1 2018-11-30 Alan Modra <amodra@gmail.com>
3 * config/rs6000/predicates.md (easy_fp_constant): Avoid long
4 dependent insn sequences.
5 * config/rs6000/rs6000.c (num_insns_constant): Support long
7 * config/rs6000/rs6000.md (mov<mode>_softfloat <FMOVE128>): Adjust
10 2018-11-30 Alan Modra <amodra@gmail.com>
12 * config/rs6000/rs6000.c (num_insns_constant_gpr): Renamed from
13 num_insns_constant_wide. Make static. Revise comment.
14 (num_insns_constant_multi): New function.
15 (num_insns_constant): Formatting. Correct CONST_WIDE_INT
16 calculation. Simplify and extract code common to both
17 CONST_INT and CONST_DOUBLE. Add gcc_unreachable for unhandled
19 * config/rs6000/rs6000-protos.h (num_insns_const_wide): Delete.
21 2018-11-30 Richard Biener <rguenther@suse.de>
23 PR tree-optimization/88274
24 * tree-vrp.c (ranges_from_anti_range): Fix handling of
27 2018-11-29 Andi Kleen <ak@linux.intel.com>
29 * config/i386/i386.c (current_fentry_section): Add.
30 (x86_function_profiler): Handle fentry section.
31 (ix86_attribute_table): Add fentry section.
32 * config/i386/i386.opt: Add -mfentry-section.
33 * doc/extend.texi: Document fentry_section attribute.
34 * doc/invoke.texi: Document -mfentry-section.
36 2018-11-29 Andi Kleen <ak@linux.intel.com>
38 * config/i386/i386.c (x86_print_call_or_nop): Handle nop name.
39 (current_fentry_name): Add.
40 (x86_function_profiler): Handle fentry_name attribute.
41 (ix86_handle_fentry_name): Add.
42 (ix86_attribute_table): Add fentry_name.
43 * config/i386/i386.opt: Add -mfentry-name
44 * doc/extend.texi: Document fentry_name.
45 * doc/invoke.texi: Document minstrument-return.
47 2018-11-29 Andi Kleen <ak@linux.intel.com>
49 * config/i386/i386-opts.h (enum instrument_return): Add.
50 * config/i386/i386.c (output_return_instrumentation): Add.
51 (ix86_output_function_return): Call output_return_instrumentation.
52 (ix86_output_call_insn): Call output_return_instrumentation.
53 * config/i386/i386.opt: Add -minstrument-return=.
54 * doc/invoke.texi (-minstrument-return): Document.
56 2018-11-29 Eric Botcazou <ebotcazou@adacore.com>
59 * config/sparc/sparc-modes.def: Minor tweak.
60 * config/sparc/sparc.c: Minor reordering.
61 (sparc_pass_by_reference): Move around.
62 (traverse_record_type): Change offset from HOST_WIDE_INT to int.
63 (classify_registers): Likewise for bitpos.
64 (function_arg_slotno): Remove dead test and tweak comments.
65 <MODE_RANDOM>: Remove useless assertion and test whether the
66 parameter is named in order to pass it in FP registers. Return
67 the regno for floating-point vector types.
68 (compute_int_layout): Change bitpos from HOST_WIDE_INT to int.
69 (compute_fp_layout): Likewise.
70 (count_registers): Likewise.
71 (assign_int_registers): Likewise.
72 (assign_fp_registers): Likewise.
73 (assign_registers): Likewise.
74 (function_arg_record_value): Change size from HOST_WIDE_INT to int
75 and use CEIL_NWORDS to compute the number of registers.
76 (function_arg_union_value): Minor tweaks.
77 (function_arg_vector_value): Add slotno and named parameters, use
78 CEIL_NWORDS to compute the number of registers.
79 (sparc_function_arg_1): Rework handling of vector types. Change
80 size from HOST_WIDE_INT to int.
81 (sparc_arg_partial_bytes): Rework handling of 32-bit ABI and deal
82 with vector types for the 64-bt ABI.
83 (sparc_function_arg_advance): Likewise.
84 (sparc_return_in_memory): Add reference to -fpcc-struct-return.
85 (sparc_struct_value_rtx): Return NULL_RTX instead of 0.
86 (sparc_function_value_1): Rework handling of vector types. Change
87 size from HOST_WIDE_INT to int.
89 2018-11-29 Jakub Jelinek <jakub@redhat.com>
92 * config/i386/sse.md (*<sse>_movmsk<ssemodesuffix><avxsizesuffix>_lt,
93 *<sse>_movmsk<ssemodesuffix><avxsizesuffix>_zext_lt,
94 *<sse>_movmsk<ssemodesuffix><avxsizesuffix>_shift,
95 *<sse>_movmsk<ssemodesuffix><avxsizesuffix>_zext_shift,
96 *<sse2_avx2>_pmovmskb_lt, *<sse2_avx2>_pmovmskb_zext_lt): New
97 define_insn_and_split patterns.
101 (*<sse4_1>_blendv<ssemodesuffix><avxsizesuffix>_lt,
102 *<sse4_1>_blendv<ssefltmodesuffix><avxsizesuffix>_ltint,
103 *<sse4_1_avx2>_pblendvb_lt): Change define_insn into
104 define_insn_and_split.
106 2018-11-29 Peter Bergner <bergner@linux.ibm.com>
109 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
110 -mabi=ieeelongdouble without both -mpopcntd and -mvsx.
112 2018-11-29 Uros Bizjak <ubizjak@gmail.com>
114 * config/i386/i386.c (inline_memory_move_cost):
115 Check "in" for 2 in MMX_CLASS_P case.
116 * config/i386/mmx.md (*mov<mode>_internal): Correct
117 TARGET_INTER_UNIT_MOVES_FROM_VEC and TARGET_INTER_UNIT_MOVES_TO_VEC
118 alternatives in preferred_for_speed attribute calculation.
120 2018-11-29 Martin Sebor <msebor@redhat.com>
124 * doc/extend.texi (attribute constructor): Clarify.
126 2018-11-29 Martin Liska <mliska@suse.cz>
129 * builtins.c (expand_movstr): Fix thinko introduced
130 when switching to the new enum.
132 2018-11-29 qing zhao <qing.zhao@oracle.com>
134 * cif-code.def (EXTERN_LIVE_ONLY_STATIC): New CIF code.
135 * common.opt: Add -flive-patching flag.
136 * doc/invoke.texi: Document -flive-patching.
137 * flag-types.h (enum live_patching_level): New enum.
138 * ipa-inline.c (can_inline_edge_p): Disable external functions from
139 inlining when flag_live_patching is LIVE_PATCHING_INLINE_ONLY_STATIC.
140 * opts.c (control_options_for_live_patching): New function.
141 (finish_options): Make flag_live_patching incompatible with flag_lto.
142 Control IPA optimizations based on different levels of
145 2018-11-29 Giuliano Belinassi <giuliano.belinassi@usp.br>
147 * match.pd (sinh (atanh (x))): New simplification rules.
148 (cosh (atanh (x))): Likewise.
150 2018-11-29 Jakub Jelinek <jakub@redhat.com>
153 * config/i386/sse.md (ssebytemode): Add V16SI, V8SI and V4SI entries.
154 (ssefltmodesuffix, ssefltvecmode): New define_mode_attrs.
155 (*<sse4_1>_blendv<ssemodesuffix><avxsizesuffix>_lt,
156 *<sse4_1>_blendv<ssefltmodesuffix><avxsizesuffix>_ltint,
157 *<sse4_1_avx2>_pblendvb_lt): New define_insns.
160 * tree.h (build_uniform_cst, uniform_integer_cst_p): Declare.
161 * tree.c (build_uniform_cst, uniform_integer_cst_p): New functions.
162 * match.pd (define_predicates): Add uniform_integer_cst_p.
163 (cmp @0 INTEGER_CST@1, cmp (convert?@2 @0) INTEGER_CST@1): Adjust
164 so that it works also for vector comparisons with uniform constants
165 with INTEGER_CST element.
168 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): For
169 vec_add and vec_sub builtins, perform PLUS_EXPR or MINUS_EXPR
170 in unsigned_type_for instead of vector integral type where overflow
173 2018-11-29 Michael Ploujnikov <michael.ploujnikov@oracle.com>
175 There can be at most one .resolver clone per function
176 * config/rs6000/rs6000.c (make_resolver_func): Generate
177 resolver symbol with clone_function_name instead of
178 clone_function_name_numbered.
180 2018-11-29 Richard Biener <rguenther@suse.de>
182 PR tree-optimization/88243
183 * tree-vect-patterns.c (vect_mark_pattern_stmts): Set the def
184 type of all pattern-sequence stmts to vect_internal_def.
186 2018-11-29 Andre Vieira <andre.simoesdiasvieira@arm.com>
189 * config/arm/arm-cpus.in (armv7-r): Add FP16conv configurations.
190 (cortex-r7, cortex-r8): Update default and add new configuration.
191 * doc/invoke.texi (armv7-r): Add two new vfp options.
192 (nofp.dp): Add cortex-r7 and cortex-r8 to the list of targets that
195 2018-11-29 Alan Modra <amodra@gmail.com>
197 * config/rs6000/rs6000.c (rs6000_emit_move): Disable long
198 double split for targets other than Darwin.
200 2018-11-29 Alan Modra <amodra@gmail.com>
202 * config/rs6000/rs6000.md (movdi_internal64): Correct lengths.
204 2018-11-29 Alan Modra <amodra@gmail.com>
206 * config/rs6000/predicates.md (easy_fp_constant): Remove code
207 dealing with integers in const_double. Assert on unexpected
208 modes. Delete superfluous ABI_V4 test.
209 * config/rs6000/rs6000.c (rs6000_emit_move): Comment fixes.
210 Don't call easy_fp_constant for Pmode.
211 (rs6000_hash_constant): Remove code dealing with integers in
213 (rs6000_legitimate_constant_p): Likewise.
214 (output_toc): Formatting, use CONST_DOUBLE_P.
216 2018-11-29 Alan Modra <amodra@gmail.com>
218 * config/rs6000/constraints.md (G, H): Comment on purpose of
219 constraint. Correct mode comments and "G" mode test.
220 * config/rs6000/rs6000.md (movdi_internal32): Remove "GH" from
221 alternative handling "F". Add length attr. Formatting.
222 (mov<mode>_softfloat <FMOVE128>): Delete "GH" from
223 alternative, and move "F" to separate alternative. Correct
226 2018-11-29 Alan Modra <amodra@gmail.com>
228 * config/rs6000/predicates.md (easy_vector_constant): Don't call
231 2018-11-29 Alan Modra <amodra@gmail.com>
233 * config.in (HAVE_AS_PLTSEQ): Add.
234 * config/rs6000/predicates.md (indirect_call_operand): New.
235 * config/rs6000/rs6000-protos.h (rs6000_pltseq_template),
236 (rs6000_sibcall_sysv): Declare.
237 * config/rs6000/rs6000.c (init_cumulative_args): Set cookie
238 CALL_LONG for -fno-plt.
239 (print_operand <T, z, 0>): Handle UNSPEC_PLTSEQ.
240 (rs6000_indirect_call_template_1): Emit .reloc directives for
242 (rs6000_pltseq_template): New function.
243 (rs6000_longcall_ref): Add arg parameter. Use PLT16 insns if
244 relocs supported by assembler. Move SYMBOL_REF test to callers.
245 (rs6000_call_aix): Adjust rs6000_longcall_ref call. Package
246 insns in UNSPEC_PLTSEQ, preserving original func_desc.
247 (rs6000_call_sysv): Likewise.
248 (rs6000_sibcall_sysv): New function.
249 * config/rs6000/rs6000.h (HAVE_AS_PLTSEQ): Provide default.
250 * config/rs6000/rs6000.md (UNSPEC_PLTSEQ, UNSPEC_PLT16_HA,
251 UNSPEC_PLT16_LO): New.
252 (pltseq_tocsave, pltseq_plt16_ha, pltseq_plt16_lo, pltseq_mtctr): New.
253 (call_indirect_nonlocal_sysv): Don't differentiate zero from non-zero
254 cookie in constraints. Test explicitly for flags in length attr.
255 Handle unspec operand 1.
256 (call_value_indirect_nonlocal_sysv): Likewise.
257 (call_indirect_aix, call_value_indirect_aix): Handle unspec operand 1.
258 (call_indirect_elfv2, call_value_indirect_elfv2): Likewise.
259 (sibcall, sibcall_value): Use rs6000_sibcall_sysv.
260 (sibcall_indirect_nonlocal_sysv): New pattern.
261 (sibcall_value_indirect_nonlocal_sysv): Likewise.
262 (sibcall_nonlocal_sysv, sibcall_value_nonlocal_sysv): Remove indirect
264 * configure.ac: Check for gas plt sequence marker support.
265 * configure: Regenerate.
267 2018-11-29 Alan Modra <amodra@gmail.com>
269 * config/rs6000/predicates.md (unspec_tls): New.
270 * config/rs6000/rs6000-protos.h (rs6000_call_template),
271 (rs6000_sibcall_template): Update prototype.
272 (rs6000_longcall_ref): Delete.
273 (rs6000_call_sysv): Declare.
274 * config/rs6000/rs6000.c (edit_tls_call_insn): New function.
275 (global_tlsarg): New variable.
276 (rs6000_legitimize_tls_address): Rewrite __tls_get_addr call
278 (print_operand): Extract UNSPEC_TLSGD address operand.
279 (rs6000_call_template, rs6000_sibcall_template): Remove arg
280 parameter, extract from second call operand instead.
281 (rs6000_longcall_ref): Make static, localize vars.
282 (rs6000_call_aix): Rename parameter to reflect new usage. Take
283 tlsarg from global_tlsarg. Don't create unused rtl or nop insns.
284 (rs6000_sibcall_aix): Rename parameter to reflect new usage. Take
285 tlsarg from global_tlsarg.
286 (rs6000_call_sysv): New function.
287 * config/rs6000/rs6000.md: Adjust rs6000_call_template and
288 rs6000_sibcall_template throughout.
289 (tls_gd_aix, tls_gd_sysv, tls_gd_call_aix, tls_gd_call_sysv): Delete.
290 (tls_ld_aix, tls_ld_sysv, tls_ld_call_aix, tls_ld_call_sysv): Delete.
291 (tls_gdld_nomark): New insn.
292 (tls_gd): Swap operand order. Simplify mode selection.
293 (tls_gd_high, tls_gd_low): Swap operand order.
294 (tls_ld): Remove const_int 0 vector element from UNSPEC_TLSLD.
295 Simplify mode selection.
296 (tls_ld_high, tls_ld_low): Similarly adjust UNSPEC_TLSLD.
297 (call, call_value): Don't assert for second call operand.
298 Use rs6000_call_sysv.
300 2018-11-29 Alan Modra <amodra@gmail.com>
302 * config/rs6000/darwin.md (call_indirect_nonlocal_darwin64),
303 (call_nonlocal_darwin64, call_value_indirect_nonlocal_darwin64),
304 (call_value_nonlocal_darwin64): Remove constraints from second call
305 arg, the rounded_stack_size_rtx arg.
306 * config/rs6000/rs6000.md (tls_gd_aix, tls_gd_sysv, tls_gd_call_aix),
307 (tls_gd_call_sysv, tls_ld_aix, tls_ld_sysv, tls_ld_call_aix),
308 (tls_ld_call_sysv, call_local32, call_local64, call_value_local32),
309 (call_value_local64, call_indirect_nonlocal_sysv),
310 (call_nonlocal_sysv, call_nonlocal_sysv_secure),
311 (call_value_indirect_nonlocal_sysv, call_value_nonlocal_sysv),
312 (call_value_nonlocal_sysv_secure, call_local_aix),
313 (call_value_local_aix, call_nonlocal_aix, call_value_nonlocal_aix),
314 (call_indirect_aix, call_value_indirect_aix, call_indirect_elfv2),
315 (call_value_indirect_elfv2, sibcall_local32, sibcall_local64),
316 (sibcall_value_local32, sibcall_value_local64, sibcall_aix),
317 (sibcall_value_aix): Likewise.
319 2018-11-29 Alan Modra <amodra@gmail.com>
321 * config/rs6000/rs6000.md (TLSmode): Delete mode iterator. Replace
322 with P throughout except for call mems which should use SI.
323 (tls_abi_suffix, tls_sysv_suffix, tls_insn_suffix): Delete mode
324 attributes. Replace with bits, mode and ptrload respectively.
326 2018-11-29 Alan Modra <amodra@gmail.com>
328 * config/rs6000/rs6000-protos.h (rs6000_indirect_call_template),
329 (rs6000_indirect_sibcall_template): Declare.
330 * config/rs6000/rs6000.c (rs6000_indirect_call_template_1),
331 (rs6000_indirect_call_template, rs6000_indirect_sibcall_template):
333 * config/rs6000/rs6000.md (call_indirect_nonlocal_sysv),
334 (call_value_indirect_nonlocal_sysv, sibcall_nonlocal_sysv),
335 (call_indirect_aix, call_value_indirect_aix): Use
336 rs6000_indirect_call_template and rs6000_indirect_sibcall_template.
337 call_indirect_elfv2, call_value_indirect_elfv2): Likewise, and
338 handle both speculation and non-speculation cases.
339 (call_indirect_aix_nospec, call_value_indirect_aix_nospec): Delete.
340 (call_indirect_elfv2_nospec, call_value_indirect_elfv2_nospec): Delete.
342 2018-11-29 Alan Modra <amodra@gmail.com>
344 * config/rs6000/rs6000-protos.h (rs6000_call_template): Declare.
345 (rs6000_sibcall_template): Declare.
346 (macho_call_template): Rename from output_call.
347 * config/rs6000/rs6000.c (rs6000_call_template_1): New function.
348 (rs6000_call_template, rs6000_sibcall_template): Likewise.
349 (macho_call_template): Rename from output_call.
350 * config/rs6000/rs6000.md (tls_gd_aix, tls_gd_sysv),
351 (tls_gd_call_aix, tls_gd_call_sysv, tls_ld_aix, tls_ld_sysv),
352 (tls_ld_call_aix, tls_ld_call_sysv, call_nonlocal_sysv),
353 (call_nonlocal_sysv_secure, call_value_nonlocal_sysv),
354 (call_value_nonlocal_sysv_secure, call_nonlocal_aix),
355 (call_value_nonlocal_aix): Use rs6000_call_template and update
356 occurrences of output_call to macho_call_template.
357 (sibcall_nonlocal_sysv, sibcall_value_nonlocal_sysv, sibcall_aix),
358 (sibcall_value_aix): Use rs6000_sibcall_template.
360 2018-11-28 Aaron Sawdey <acsawdey@linux.ibm.com>
362 * config/rs6000/rs6000-string.c (expand_block_clear): Change how
363 we determine if unaligned vsx is ok.
365 2018-11-28 Jan Hubicka <jh@suse.cz>
367 * predict.c (determine_unlikely_bbs): Forward declare; also determine
368 edges that are always known to be taken because it is only likely
369 edge out of given BB.
370 (tree_estimate_probability): Call before profile guessing to get bit
373 2018-11-28 Jan Hubicka <jh@suse.cz>
375 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Handle
376 profile_probability::always better.
378 2018-11-28 Jan Hubicka <jh@suse.cz>
380 * profile-count.h (profile_count::split): Give better result when
381 splitting profile_probability::always.
383 2018-11-28 Vladimir Makarov <vmakarov@redhat.com>
386 * ira-costs.c (scan_one_insn): Process subregs when updating costs
387 for pseudos and allocnos from insn.
389 2018-11-28 David Edelsohn <dje.gcc@gmail.com>
391 * config/rs6000/aix72.h: Update to match aix71.h changes.
393 2018-11-28 Jeff Law <law@redhat.com>
395 * config/h8300/constraints.md: Add "C" constraint for call insns.
396 * config/h8300/h8300.md (call, call_value): Turn into a define_expand
397 and define_insn pair. Move invalid call targets into a register in
398 the expander and fix constraints in the matching pattern.
399 * config/h8300/predicates.md (call_expander_operand): Renamed from
400 call_insn_operand. Reject things we shouldn't be trying to handle.
401 (call_insn_operand): New predicate for use by the call/call_value
403 (small_call_insn_operand): Update appropriately.
405 2018-11-28 Sam Tebbs <sam.tebbs@arm.com>
407 * config/aarch64/aarch64.c (aarch64_process_target_attr): Replace
408 calls to strtok with strtok_r.
410 2018-11-28 Richard Biener <rguenther@suse.de>
412 PR tree-optimization/88223
413 * tree-ssa-sccvn.c (vn_reference_lookup_3): When skipping
414 over a stored-same value may-alias store make sure to consider
415 partial overlaps which are valid when TBAA reasonings do not
416 apply and byte-granular overlaps are possible at all.
418 2018-11-28 Richard Biener <rguenther@suse.de>
420 PR tree-optimization/88217
421 * vr-values.c (vr_values::extract_range_from_phi_node): Make
422 sure to handle results > +INF and < -INF correctly when
423 trying to drop down to +INF - 1 or -INF + 1.
425 2018-11-28 Alan Modra <amodra@gmail.com>
427 * xcoffout.c (do_block): Signed/unsigned warning fix.
429 2018-11-28 Richard Biener <rguenther@suse.de>
431 PR tree-optimization/79351
432 * tree-ssa-sccvn.c (vn_reference_lookup_3): For assignments from
433 empty CONSTRUCTORs ensure the store is at a constant position.
435 2018-11-28 Richard Biener <rguenther@suse.de>
437 PR tree-optimization/88229
438 * tree-ssa.c (non_rewritable_mem_ref_base): Check DECL_SIZE_UNIT
439 is an INTEGER_CST before accessing it so.
441 2018-11-28 Sam Tebbs <sam.tebbs@arm.com>
443 * doc/invoke.texi (-mtune=): Remove obsolete CPU names.
445 2018-11-28 Jakub Jelinek <jakub@redhat.com>
448 * config/i386/i386.c (ix86_expand_sse_movcc): Handle DFmode and
449 SFmode using sse4_1_blendvs[sd] with TARGET_SSE4_1. Formatting fixes.
450 * config/i386/sse.md (sse4_1_blendv<ssemodesuffix>): New pattern.
452 2018-11-27 Martin Liska <mliska@suse.cz>
454 * gcov.c (generate_results): Append current_working_directory
457 2018-11-27 Mike Gulick <mgulick@mathworks.com>
459 PR preprocessor/83173
460 * input.c (dump_location_info): Dump reason and included_from
461 fields from line_map_ordinary struct. Fix indentation when
463 * diagnostic-show-locus.c (num_digits, num_digits): Move to
464 diagnostic.c to allow it to be utilized by input.c.
465 * diagnostic.c (num_digits, selftest::test_num_digits): Moved
467 (selftest::diagnostic_c_tests): Run selftest::test_num_digits.
468 * diagnostic.h (num_digits): Add extern definition.
470 2018-11-27 Fredrik Noring <noring@nocrew.org>
472 * config/mips/mips.c (mips_reorg_process_insns)
473 (mips_option_override): Handle `-mfix-r5900'.
474 * config/mips/mips.h (ASM_SPEC): Add `mfix-r5900' and
476 * config/mips/mips.opt (mfix-r5900): New option.
477 * doc/invoke.texi: Document the `r5900' processor name, and
478 `-mfix-r5900' and `-mno-fix-r5900' options.
480 2018-11-27 Jakub Jelinek <jakub@redhat.com>
483 * config/rs6000/rs6000.c (print_operand) <case 'D'>: Use
484 output_operand_lossage instead of gcc_assert.
485 <case 't'>: Likewise.
486 <case 'z'>: Likewise.
487 <case 'V'>: Use output_operand_lossage instead of gcc_unreachable.
489 2018-11-27 Jeff Law <law@redhat.com>
491 * config/riscv/riscv (riscv_block_mvoe_straight): Use RETURN_BEGIN
492 in call to move_by_pieces.
494 * config/sh/sh-mem.c (expand_block_move): Use RETURN_BEGIN in call
497 * config/lm32/lm32.c (lm32_block_move_inline): Use RETURN_BEGIN in
498 call to move_by_pieces.
500 * config/mips/mips.c (mips_block_move_straight): Use RETURN_BEGIN
501 in call to move_by_pieces.
503 * config/microblaze/microblaze.c (microblaze_block_move_straight): Use
504 RETURN_BEGIN in call to move_by_pieces.
505 (microblaze_expand_block_move): Likewise.
507 2018-11-27 Tamar Christina <tamar.christina@arm.com>
509 * doc/md.texi (xorsign): Document it.
511 2018-11-27 Martin Liska <mliska@suse.cz>
513 * config/gnu-user.h (TARGET_F951_OPTIONS): New.
514 * gcc.c (find_fortran_preinclude_file): New function
515 to handle Fortran pre-include.
517 2018-11-27 Martin Liska <mliska@suse.cz>
519 * asan.c (asan_emit_stack_protection): Use new enum values
520 instead of int constants.
521 * builtins.c (expand_builtin_memory_copy_args): Replace int
522 type with memop_ret enum type.
523 (expand_builtin_mempcpy_args): Likewise.
524 (expand_builtin_memcpy): Use new enum values
525 instead of int constants. Likewise.
526 (expand_builtin_mempcpy): Likewise.
527 (expand_movstr): Likewise.
528 (expand_builtin_strcpy_args): Likewise.
529 (expand_builtin_stpcpy_1): Likewise.
530 (expand_builtin_strncpy): Likewise.
531 (expand_builtin_memset_args): Likewise.
532 * expr.c (move_by_pieces_d::finish_endp): Rename to ...
533 (move_by_pieces_d::finish_retmode): ... this.
534 (move_by_pieces): Change last argument type to memop_ret.
535 (store_by_pieces): Use new enum values
536 instead of int constants.
537 (emit_block_move_hints): Likewise.
538 (emit_push_insn): Likewise.
539 (store_expr): Likewise.
540 * expr.h (store_by_pieces): Change int to newly added enum
542 * rtl.h (enum memop_ret): Define.
543 (move_by_pieces): Use the enum type.
545 2018-11-27 Alan Modra <amodra@gmail.com>
547 * config/rs6000/aix71.h (ASM_SPEC): Don't select default -maix64
549 (ASM_CPU_SPEC): Do so here. Rewrite using if .. else if .. specs
550 form. Error on missing -mcpu case.
551 * config/rs6000/driver-rs6000.c (asm_names <_AIX>): Update NULL case.
552 (asm_names <!_AIX>): Add missing cpus. Update NULL case. Apply
553 PR63177 fix for -mcpu=power8 and -mcpu=powerpc64le.
554 * config/rs6000/rs6000.h (ASM_CPU_SPEC): Rewrite using if ..
555 else if .. specs form. Error on missing -mcpu case. Don't output
556 duplicate -maltivec. Apply PR63177 fix for -mcpu=powerpc64le.
558 2018-11-26 David Malcolm <dmalcolm@redhat.com>
560 * dump-context.h (dump_context::dump_loc): Convert 1st param from
561 dump_flags_t to const dump_metadata_t &. Convert 2nd param from
562 const dump_location_t & to const dump_user_location_t &.
563 (dump_context::dump_loc_immediate): Convert 2nd param from
564 const dump_location_t & to const dump_user_location_t &.
565 (dump_context::dump_gimple_stmt): Convert 1st param from
566 dump_flags_t to const dump_metadata_t &.
567 (dump_context::void dump_gimple_stmt_loc): Likewise; convert
568 2nd param from const dump_location_t & to
569 const dump_user_location_t &.
570 (dump_context::dump_gimple_expr): Convert 1st param from
571 dump_flags_t to const dump_metadata_t &.
572 (dump_context::dump_gimple_expr_loc): Likewise; convert
573 2nd param from const dump_location_t & to
574 const dump_user_location_t &.
575 (dump_context::dump_generic_expr): Convert 1st param from
576 dump_flags_t to const dump_metadata_t &.
577 (dump_context::dump_generic_expr_loc): Likewise; convert
578 2nd param from const dump_location_t & to
579 const dump_user_location_t &.
580 (dump_context::dump_printf_va): Convert 1st param from
581 dump_flags_t to const dump_metadata_t &.
582 (dump_context::dump_printf_loc_va): Likewise; convert
583 2nd param from const dump_location_t & to
584 const dump_user_location_t &.
585 (dump_context::dump_dec): Convert 1st param from
586 dump_flags_t to const dump_metadata_t &.
587 (dump_context::dump_symtab_node): Likewise.
588 (dump_context::begin_scope): Split out 2nd param into
589 user and impl locations.
590 (dump_context::ensure_pending_optinfo): Add metadata param.
591 (dump_context::begin_next_optinfo): Replace dump_location_t param
592 with metadata and user location.
593 * dumpfile.c (dump_context::dump_loc): Convert 1st param from
594 dump_flags_t to const dump_metadata_t &. Convert 2nd param from
595 const dump_location_t & to const dump_user_location_t &.
596 (dump_context::dump_loc_immediate): Convert 2nd param from
597 const dump_location_t & to const dump_user_location_t &.
598 (dump_context::dump_gimple_stmt): Convert 1st param from
599 dump_flags_t to const dump_metadata_t &.
600 (dump_context::void dump_gimple_stmt_loc): Likewise; convert
601 2nd param from const dump_location_t & to
602 const dump_user_location_t &.
603 (dump_context::dump_gimple_expr): Convert 1st param from
604 dump_flags_t to const dump_metadata_t &.
605 (dump_context::dump_gimple_expr_loc): Likewise; convert
606 2nd param from const dump_location_t & to
607 const dump_user_location_t &.
608 (dump_context::dump_generic_expr): Convert 1st param from
609 dump_flags_t to const dump_metadata_t &.
610 (dump_context::dump_generic_expr_loc): Likewise; convert
611 2nd param from const dump_location_t & to
612 const dump_user_location_t &.
613 (dump_context::dump_printf_va): Convert 1st param from
614 dump_flags_t to const dump_metadata_t &.
615 (dump_context::dump_printf_loc_va): Likewise; convert
616 2nd param from const dump_location_t & to
617 const dump_user_location_t &.
618 (dump_context::dump_dec): Convert 1st param from
619 dump_flags_t to const dump_metadata_t &.
620 (dump_context::dump_symtab_node): Likewise.
621 (dump_context::begin_scope): Split out 2nd param into
622 user and impl locations.
623 (dump_context::ensure_pending_optinfo): Add metadata param.
624 (dump_context::begin_next_optinfo): Replace dump_location_t param
625 with metadata and user location.
626 (dump_gimple_stmt): Convert 1st param from dump_flags_t to
627 const dump_metadata_t &.
628 (dump_gimple_stmt_loc): Likewise; convert 2nd param from
629 const dump_location_t & to const dump_user_location_t &.
630 (dump_gimple_expr): Convert 1st param from dump_flags_t to
631 const dump_metadata_t &.
632 (dump_gimple_expr_loc): Likewise; convert 2nd param from
633 const dump_location_t & to const dump_user_location_t &.
634 (dump_generic_expr): Convert 1st param from dump_flags_t to
635 const dump_metadata_t &.
636 (dump_generic_expr_loc): Likewise; convert 2nd param from
637 const dump_location_t & to const dump_user_location_t &.
638 (dump_printf): Convert 1st param from dump_flags_t to
639 const dump_metadata_t &.
640 (dump_printf_loc): Likewise; convert 2nd param from
641 const dump_location_t & to const dump_user_location_t &.
642 (dump_dec): Convert 1st param from dump_flags_t to
643 const dump_metadata_t &.
644 (dump_symtab_node): Likewise.
645 (dump_begin_scope): Split out 2nd param into user and impl
647 (selftest::assert_impl_location_eq): New function.
648 (ASSERT_IMPL_LOCATION_EQ): New macro.
649 (selftest::test_impl_location): Update to use
650 ASSERT_IMPL_LOCATION_EQ.
651 (selftest::test_capture_of_dump_calls): Convert "loc" to
652 dump_user_location_t. Add ASSERT_IMPL_LOCATION_EQ throughout,
653 verifying line numbers of dump emissions.
654 * dumpfile.h (class dump_metadata_t): New class.
655 (dump_printf): Convert 1st param from dump_flags_t to
656 const dump_metadata_t &.
657 (dump_printf_loc): Likewise; convert 2nd param from
658 const dump_location_t & to const dump_user_location_t &.
659 (dump_generic_expr_loc): Likewise.
660 (dump_generic_expr): Convert 1st param from dump_flags_t to
661 const dump_metadata_t &.
662 (dump_gimple_stmt_loc): Likewise; convert 2nd param from
663 const dump_location_t & to const dump_user_location_t &.
664 (dump_gimple_stmt): Convert 1st param from dump_flags_t to
665 const dump_metadata_t &.
666 (dump_gimple_expr_loc): Likewise; convert 2nd param from
667 const dump_location_t & to const dump_user_location_t &.
668 (dump_gimple_expr): Convert 1st param from dump_flags_t to
669 const dump_metadata_t &.
670 (dump_symtab_node): Likewise.
671 (dump_dec): Likewise.
672 (dump_begin_scope): Split out 2nd param into user and impl
674 (auto_dump_scope::auto_dump_scope): Split "loc" param into a user
675 location and impl_location, and capture the impl_location.
676 (AUTO_DUMP_SCOPE): Rename param from LOC to USER_LOC.
677 * loop-unroll.c (report_unroll): Update for changes to
678 dump_printf_loc and dump_printf.
679 * opt-problem.cc (opt_problem::opt_problem): Update for change to
681 * optinfo-emit-json.cc
682 (selftest::test_building_json_from_dump_calls): Convert "loc" from
683 dump_location_t to dump_user_location_t.
684 * optinfo.cc (optinfo::emit_for_opt_problem): Update for change in
686 * profile.c (compute_branch_probabilities): Update for change to
688 * selftest.h (ASSERT_STR_CONTAINS_AT): New macro.
689 * tree-vect-slp.c (vect_print_slp_tree): Update for change to
692 2018-11-27 Jozef Lawrynowicz <jozef.l@mittosystems.com>
694 * doc/sourcebuild.texi: Document check_effective_target_int_eq_float
695 and check_effective_target_ptr_eq_long.
697 2018-11-27 Alan Modra <amodra@gmail.com>
699 * config.gcc (powerpc*-*-freebsd*, powerpc-*-netbsd*),
700 (powerpc-*-eabisimaltivec*, powerpc-*-eabisim*, powerpc-*-elf*),
701 (powerpc-*-eabialtivec*, powerpc-*-eabi*, powerpc-*-rtems*),
702 (powerpc-wrs-vxworks*, powerpc-*-lynxos*, powerpcle-*-elf*),
703 (powerpcle-*-eabisim*, powerpcle-*-eabi*): Add gnu-user.h to tm_file.
704 * config/rs6000/freebsd.h (CPLUSPLUS_CPP_SPEC),
705 (LINK_GCC_C_SEQUENCE_SPEC): Undef.
706 (ASM_APP_ON, ASM_APP_OFF): Don't define.
707 * config/rs6000/freebsd64.h (ASM_APP_ON, ASM_APP_OFF): Don't define.
708 * config/rs6000/lynx.h (ASM_APP_ON, ASM_APP_OFF): Don't define.
709 * config/rs6000/linux64.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
710 * config/rs6000/netbsd.h (CPLUSPLUS_CPP_SPEC),
711 (LINK_GCC_C_SEQUENCE_SPEC): Undef.
712 * config/rs6000/rtems.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
713 * config/rs6000/sysv4.h (GNU_USER_TARGET_CRTI): Redefine.
714 (GNU_USER_TARGET_CRTN): Redefine.
715 (CC1_SPEC): Use GNU_USER_TARGET_CC1_SPEC.
716 (LIB_LINUX_SPEC): Use GNU_USER_TARGET_LIB_SPEC.
717 (CRTOFFLOADBEGIN, CRTOFFLOADEND): Don't define.
718 (STARTFILE_LINUX_SPEC): Define as GNU_USER_TARGET_STARTFILE_SPEC.
719 (ENDFILE_LINUX_SPEC): Define as GNU_USER_TARGET_ENDFILE_SPEC.
720 (UCLIBC_DYNAMIC_LINKER, CHOOSE_DYNAMIC_LINKER): Don't define.
721 (LINK_EH_SPEC): Don't define.
723 2018-11-26 Martin Sebor <msebor@redhat.com>
726 * expr.c (string_constant): Handle top-level decls of all character
727 types and subobjects of narrow character type.
729 2018-11-27 Alan Modra <amodra@gmail.com>
731 * config.gcc (powerpc*-*-linux*): Add linux.h to tm_file.
732 * config/rs6000/linux.h (TARGET_OS_CPP_BUILTINS): Use
733 GNU_USER_TARGET_OS_CPP_BUILTINS.
734 (RS6000_ABI_NAME): Define.
735 * config/rs6000/linux64.h (TARGET_OS_CPP_BUILTINS): Use
736 GNU_USER_TARGET_OS_CPP_BUILTINS.
737 (MUSL_DYNAMIC_LINKER32): Undef before defining.
738 (UCLIBC_DYNAMIC_LINKER32, UCLIBC_DYNAMIC_LINKER64): Don't define.
739 (CHOOSE_DYNAMIC_LINKER): Don't define.
740 (GNU_USER_DYNAMIC_LINKER32, GNU_USER_DYNAMIC_LINKER64): Don't define.
741 * config/rs6000/sysv4.h (MUSL_DYNAMIC_LINKER): Undef before defining.
742 (CHOOSE_DYNAMIC_LINKER, GNU_USER_DYNAMIC_LINKER): Only define when
744 (CPP_OS_LINUX_SPEC): Remove defines and asserts handled by
745 TARGET_OS_CPP_BUILTINS.
747 2018-11-26 Jozef Lawrynowicz <jozef.l@mittosystems.com>
749 * doc/sourcebuild.texi: Document check_effective_target_newlib_nano_io.
751 2018-11-26 Jakub Jelinek <jakub@redhat.com>
754 * config/i386/i386.c (def_builtin2): If tcode == VOID_FTYPE_UINT64
755 and !TARGET_64BIT, return NULL_TREE.
758 * dwarf2out.c (secname_for_decl): For functions with
759 DECL_SECTION_NAME if in_cold_section_p, try to return
760 current_function_section's name if it is a named section.
762 2018-11-26 Maya Rashish <coypu@sdf.org>
765 * config.host (*-*-netbsd*): Use x-netbsd and host-netbsd.o.
766 * config/x-netbsd: New file.
767 * config/host-netbsd: New file.
769 2018-11-26 Jozef Lawrynowicz <jozef.l@mittosystems.com>
771 * doc/sourcebuild.texi: Document check_effective_target_size20plus.
772 Clarify documentation for check_effective_target_size32plus.
774 2018-11-26 Aaron Sawdey <acsawdey@linux.ibm.com>
776 * config/rs6000/rs6000-string.c (expand_cmp_vec_sequence): Rename
777 and modify expand_strncmp_vec_sequence.
778 (emit_final_compare_vec): Rename and modify emit_final_str_compare_vec.
779 (generate_6432_conversion): New function.
780 (expand_block_compare): Add support for vsx.
781 (expand_block_compare_gpr): New function.
782 * config/rs6000/rs6000.opt (rs6000_block_compare_inline_limit): Increase
783 default limit to 63 because of more compact vsx code.
785 2018-11-26 Uros Bizjak <ubizjak@gmail.com>
788 * config/i386/i386.c (dbx_register_map): Use IGNORED_DWARF_REGNUM
789 for registers for which no debug information can be generated.
790 Use INVALID_REGNUM for non-existent registers.
791 (dbx64_register_map): Ditto.
792 (svr4_dbx_register_map): Ditto.
794 2018-11-26 Sandra Loosemore <sandra@codesourcery.com>
797 * doc/invoke.texi (Optimize Options): Document that
798 -freorder-functions also uses "hot" and "cold" attributes.
800 2018-11-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
802 * config/aarch64/aarch64.c (aarch64_mangle_type): Fix typo in comment.
804 2018-11-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
806 * config/arm/types.md (mul64): Rename to...
807 (widen_mul64): ... This.
808 * config/arm/arm-generic.md: Rename mul64 to widen_mul64.
809 * config/arm/cortex-a15.md: Likewise.
810 * config/arm/cortex-a5.md: Likewise.
811 * config/arm/cortex-a53.md: Likewise.
812 * config/arm/cortex-a57.md: Likewise.
813 * config/arm/cortex-a7.md: Likewise.
814 * config/arm/cortex-m4.md: Likewise.
815 * config/arm/exynos-m1.md: Likewise.
816 * config/arm/marvell-pj4.md: Likewise.
817 * config/arm/xgene1.md: Likewise.
819 2018-11-26 Richard Biener <rguenther@suse.de>
821 PR tree-optimization/88182
822 * tree-vect-loop.c (vectorizable_reduction): Pick up single
823 correct reduc_def_info.
824 * tree-vect-slp.c (vect_analyze_slp_instance): Set
825 STMT_VINFO_REDUC_DEF of the first stmt.
827 2018-11-26 Andreas Krebbel <krebbel@linux.ibm.com>
829 * doc/invoke.texi: Document z14/arch12 -march option.
831 2018-11-25 Sandra Loosemore <sandra@codesourcery.com>
834 * doc/invoke.texi (Option Summary): Add -Wmissing-noreturn.
835 (Warning Options): Likewise.
837 2018-11-25 Sandra Loosemore <sandra@codesourcery.com>
840 * doc/extend.texi (Common Function Attributes): Clarify that
841 functions with "const" attribute can read const global variables.
843 2018-11-26 Alan Modra <amodra@gmail.com>
845 * config/rs6000/rs6000.h (OBJECT_PEF, TARGET_MACOS): Delete.
847 2018-11-25 Sandra Loosemore <sandra@codesourcery.com>
850 * doc/extend.texi (Common Variable Attributes): Use preferred
851 placement of type attributes in examples, plus whitespace fixes.
852 (Type Attributes): Clarify why placement of attributes
853 immediately after struct/union/enum keyword is preferred.
854 (Common Type Attributes): Use preferred placement of type
855 attributes in examples, plus more whitespace fixes.
857 2018-11-25 Paul Koning <ni1d@arrl.net>
859 * config/pdp11/pdp11.h (TARGET_HAS_NO_HW_DIVIDE): Define.
861 2018-11-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
863 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
864 attributes): Document d_runtime.
866 2018-11-25 Vladimir Makarov <vmakarov@redhat.com>
869 * ira-costs.c (record_operand_costs): Initiate register move cost
872 2018-11-23 Jeff Law <law@redhat.com>
874 PR rtl-optimization/87468
875 * tree-ssa-threadupdate.c (create_block_for_threading): Clear
876 EDGE_IGNORE on all outgoing edges of the duplicate block.
878 2018-11-23 Vladimir Makarov <vmakarov@redhat.com>
881 * ira-costs.c (record_operand_costs): Use bigger hard reg class if
882 its mode does not fit to the original class.
884 2018-11-23 Martin Sebor <msebor@redhat.com>
886 PR tree-optimization/87756
887 * expr.c (string_constant): Synthesize a string literal from
888 the address of a constant character.
889 * tree.c (build_string_literal): Add an argument.
890 * tree.h (build_string_literal): Same.
892 2018-11-23 Christoph Muellner <christoph.muellner@theobroma-systems.com>
894 * config/aarch64/aarch64-cores.def: Define emag.
895 * config/aarch64/aarch64-tune.md: Regenerated with emag.
896 * config/aarch64/aarch64.c (emag_tunings): New struct.
897 * doc/invoke.texi: Document mtune value.
899 2018-11-23 Christoph Muellner <christoph.muellner@theobroma-systems.com>
900 Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
902 * config/aarch64/aarch64.c (xgene1_tunings): Optimize Xgene1 tunings
904 * config/aarch64/aarch64.c (xgene1_tunings): Add Xgene1 specific
906 * config/aarch64/aarch64.c (xgene1_addrcost_table): Correct the post-modify
908 * config/arm/aarch-cost-tables.h (xgene1_extra_costs): Update the cost table
911 2018-11-23 Richard Biener <rguenther@suse.de>
913 PR tree-optimization/88149
914 * tree-vect-slp.c (vect_slp_analyze_node_operations): Detect
915 the case where there are two different def types for the
916 same operand at different operand position in the same stmt.
918 2018-23-11 Mihail Ionescu <mihail.ionescu@arm.com>
920 * config/arm/arm.c (arm_expand_compare_and_swap): Simplify and call
921 gen_atomic_compare_swap_1.
922 (arm_evpc_neon_vuzp): Likewise gen_neon_vuzp_internal.
923 (arm_evpc_neon_vtrn): Likewise gen_neon_vtrn_internal.
924 (arm_evpc_neon_vext): Likewise gen_neon_vext_internal.
925 (arm_evpc_neon_vzip): Likewise gen_neon_vzip_internal.
926 (arm_evpc_neon_vrev): Replace the function pointer and simplify the mode
928 * config/arm/arm.md (neon_vext<mode>),
929 (neon_vrev64<mode>, neon_vrev32<mode>),
930 (neon_vrev16<mode>, neon_vtrn<mode>_internal),
931 (neon_vzip<mode>_internal, neon_vuzp<mode>_internal): Add an '@'character
932 before the pattern name.
933 * config/arm/sync.md:
934 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1),
935 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): Likewise.
937 2018-11-23 Jakub Jelinek <jakub@redhat.com>
939 PR tree-optimization/86614
940 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return early
941 if TREE_NO_WARNING is set on ref.ref.
943 2018-11-22 Sandra Loosemore <sandra@codesourcery.com>
944 Alan Coopersmith <alan.coopersmith@oracle.com>
947 * doc/extend.texi (Designated Inits): Clarify handling of multiple
948 initializers for unions.
950 2018-11-22 Jan Hubicka <jh@suse.cz>
953 * ipa-devirt.c (type_variants_equivalent_p): Drop warn and warned
954 parameters; do not warn here.
955 (odr_subtypes_equivalent_p): Likewise.
956 (warn_odr): Fix typo.
957 (warn_types_mismatch): Do not output confused warnings on integer types.
958 (odr_types_equivalent_p): Update.
960 2018-11-22 Jakub Jelinek <jakub@redhat.com>
962 * config/i386/i386.c (ix86_option_override_internal,
963 ix86_can_inline_p, classify_argument, construct_container,
964 ix86_expand_prologue, ix86_expand_split_stack_prologue,
965 ix86_expand_carry_flag_compare, expand_set_or_movmem_via_loop,
966 expand_setmem_epilogue_via_loop, promote_duplicated_reg,
967 ix86_expand_set_or_movmem, ix86_init_builtins_va_builtins_abi):
970 * config/i386/i386.c (ix86_option_override_internal): For
971 stack_protector_guard related options, use opts_set->x_ instead
972 of global_options_set. and prefix options with opts->x_ . Move
973 defaults for offset and reg into else block.
975 2018-11-22 Uros Bizjak <ubizjak@gmail.com>
977 * config/i386/i386.c (ix86_check_avx_upper_register):
978 Return true for all SSE registers with mode bitsize > 128.
980 2018-11-22 Vladimir Makarov <vmakarov@redhat.com>
982 PR rtl-optimization/87718
983 * ira-costs.c: Remove trailing white-spaces.
984 (record_operand_costs): Add a special treatment for moves
985 involving a hard register.
987 2018-11-22 Uros Bizjak <ubizjak@gmail.com>
989 * config/i386/i386.c (ix86_avx_emit_vzeroupper): Remove.
990 (ix86_emit_mode_set) <case AVX_U128>: Emit vzeroupper here.
992 2018-11-22 Martin Liska <mliska@suse.cz>
994 * common/config/i386/i386-common.c (processor_names): Add
995 static assert and add missing "znver2".
996 (ix86_get_valid_option_values): Add checking assert for null
997 values and add "native" value if feasible.
998 * config/i386/i386.h: Do not declare size of processor_names.
999 * common/config/i386/i386-common.c:
1000 * config/i386/i386.c: Add static assert for size
1001 of processor_cost_table.
1003 2018-11-22 Thomas Preud'homme <thomas.preudhomme@linaro.org>
1005 * target-insns.def (stack_protect_combined_set): Define new standard
1007 (stack_protect_combined_test): Likewise.
1008 * cfgexpand.c (stack_protect_prologue): Try new
1009 stack_protect_combined_set pattern first.
1010 * function.c (stack_protect_epilogue): Try new
1011 stack_protect_combined_test pattern first.
1012 * config/arm/arm.c (require_pic_register): Add pic_reg and compute_now
1013 parameters to control which register to use as PIC register and force
1014 reloading PIC register respectively. Insert in the stream of insns if
1016 (legitimize_pic_address): Expose above new parameters in prototype and
1017 adapt recursive calls accordingly. Use pic_reg if non null instead of
1019 (arm_load_pic_register): Add pic_reg parameter and use it if non null.
1020 (arm_legitimize_address): Adapt to new legitimize_pic_address
1022 (thumb_legitimize_address): Likewise.
1023 (arm_emit_call_insn): Adapt to require_pic_register prototype change.
1024 (arm_expand_prologue): Adapt to arm_load_pic_register prototype change.
1025 (thumb1_expand_prologue): Likewise.
1026 * config/arm/arm-protos.h (legitimize_pic_address): Adapt to prototype
1028 (arm_load_pic_register): Likewise.
1029 * config/arm/predicated.md (guard_addr_operand): New predicate.
1030 (guard_operand): New predicate.
1031 * config/arm/arm.md (movsi expander): Adapt to legitimize_pic_address
1033 (builtin_setjmp_receiver expander): Adapt to thumb1_expand_prologue
1035 (stack_protect_combined_set): New expander..
1036 (stack_protect_combined_set_insn): New insn_and_split pattern.
1037 (stack_protect_set_insn): New insn pattern.
1038 (stack_protect_combined_test): New expander.
1039 (stack_protect_combined_test_insn): New insn_and_split pattern.
1040 (arm_stack_protect_test_insn): New insn pattern.
1041 * config/arm/thumb1.md (thumb1_stack_protect_test_insn): New insn pattern.
1042 * config/arm/unspecs.md (UNSPEC_SP_SET): New unspec.
1043 (UNSPEC_SP_TEST): Likewise.
1044 * doc/md.texi (stack_protect_combined_set): Document new standard
1046 (stack_protect_set): Clarify that the operand for guard's address is
1048 (stack_protect_combined_test): Document new standard pattern name.
1049 (stack_protect_test): Clarify that the operand for guard's address is
1052 2018-11-22 Richard Biener <rguenther@suse.de>
1054 PR tree-optimization/88148
1055 * tree-ssa-loop-niter.c (simplify_replace_tree): Get optional
1056 valueization callback parameter and handle it.
1057 * tree-ssa-loop-niter.h (simplify_replace_tree): Export.
1058 * tree-ssa-sccvn.c (process_bb): Eliminate in loop niter trees.
1060 2018-11-22 Richard Biener <rguenther@suse.de>
1064 * lto-streamer-out.c (lto_is_streamable): Allow CALL_EXPRs
1065 which can appear in size expressions.
1066 * tree-streamer-in.c (unpack_ts_base_value_fields): Stream
1067 CALL_EXPR_BY_DESCRIPTOR.
1068 (streamer_read_tree_bitfields): Stream CALL_EXPR_IFN.
1069 * tree-streamer-out.c (pack_ts_base_value_fields): Stream
1070 CALL_EXPR_BY_DESCRIPTOR.
1071 (streamer_write_tree_bitfields): Stream CALL_EXPR_IFN.
1075 * tree.c (free_lang_data_in_one_sizepos): Free non-gimple-val
1078 2018-11-22 Richard Biener <rguenther@suse.de>
1080 PR tree-optimization/88069
1081 * tree-ssa-sccvn.c (visit_phi): Tweak previous fix to not
1082 apply to default defs.
1084 2018-11-22 Jakub Jelinek <jakub@redhat.com>
1088 * config/i386/i386.c (ix86_option_override_internal): Default
1089 ix86_stack_protector_guard to SSP_TLS only if TARGET_THREAD_SSP_OFFSET
1091 * config/i386/i386.md (stack_protect_set, stack_protect_set_<mode>,
1092 stack_protect_test, stack_protect_test_<mode>): Use empty condition
1093 instead of TARGET_SSP_TLS_GUARD.
1095 2018-11-22 Martin Liska <mliska@suse.cz>
1098 * config/i386/i386.c (ix86_option_override_internal): Error
1099 about usage -mabi=ms and -fsanitize=thread.
1101 2018-11-21 Uros Bizjak <ubizjak@gmail.com>
1104 2013-10-26 Vladimir Makarov <vmakarov@redhat.com>
1107 2013-10-25 Vladimir Makarov <vmakarov@redhat.com>
1109 * lra-spills.c (lra_final_code_change): Remove useless move insns.
1111 2018-11-21 Uros Bizjak <ubizjak@gmail.com>
1114 * function.c (expand_function_end): Do not emit extra blockage insn.
1116 2018-11-21 Lokesh Janghel <lokeshjanghel91@gmail.com>
1119 * config/i386/i386.c (function_value_ms_64): Return AX_REG instead
1120 of FIRST_SSE_REG for 4 or 8 byte modes.
1122 2018-11-21 Jan Hubicka <jh@suse.cz>
1125 * tree.c (fld_decl_context): Break out from ...
1126 (free_lang_data_in_decl): ... here; free TREE_PUBLIC, TREE_PRIVATE
1127 DECL_ARTIFICIAL of TYPE_DECL; do not free TREE_TYPE of TYPE_DECL.
1128 (fld_incomplete_type_of): Build copy of TYP_DECL.
1129 * ipa-devirt.c (free_enum_values): Rename to ...
1130 (free_odr_warning_data): ... this one; free also duplicated TYPE_DECLs
1131 and TREE_TYPEs of TYPE_DECLs.
1132 (get_odr_type): Initialize odr_vtable_hash if needed.
1134 2018-11-21 Alexandre Oliva <oliva@adacore.com>
1136 * final.c (compute_discriminator): Declare. Renamed from...
1137 (maybe_set_discriminator): ... this. Set and return a local.
1138 (override_discriminator): New.
1139 (final_scan_insn_1): Set it.
1140 (notice_source_line): Adjust. Always set discriminator.
1142 2018-11-21 Jakub Jelinek <jakub@redhat.com>
1145 * config/aarch64/atomics.md (@aarch64_compare_and_swap<mode>): Use
1146 rIJ constraint for aarch64_plus_operand rather than rn.
1148 2018-11-21 Renlin Li <renlin.li@arm.com>
1151 * explow.h (get_dynamic_stack_size): Declare it as external.
1152 * explow.c (record_new_stack_level): Remove function static attribute.
1153 * function.c (assign_stack_local_1): Dynamically align the stack slot
1154 addr for parameter copy on the stack.
1156 2018-11-21 Richard Biener <rguenther@suse.de>
1159 * bitmap.c (bitmap_last_set_bit): Refactor to avoid warning.
1160 * Makefile.in (bitmap.o-warn): Remove again.
1162 2018-11-20 Jeff Law <law@redhat.com>
1164 PR tree-optimization/88069
1165 * tree-ssa-dom.c (record_equivalences_from_phis): Propagate away
1166 degenerate virtual PHIs.
1168 2018-11-21 H.J. Lu <hongjiu.lu@intel.com>
1171 * config/i386/sse.md (sse4_1_<code>v8qiv8hi2<mask_name>): Replace
1172 nonimmediate_operand with register_operand.
1173 (avx2_<code>v8qiv8si2<mask_name>): Likewise.
1174 (sse4_1_<code>v4qiv4si2<mask_name>): Likewise.
1175 (sse4_1_<code>v4hiv4si2<mask_name>): Likewise.
1176 (sse4_1_<code>v2qiv2di2<mask_name>): Likewise.
1177 (avx512f_<code>v8qiv8di2<mask_name>): Likewise.
1178 (avx2_<code>v4qiv4di2<mask_name>): Likewise.
1179 (avx2_<code>v4hiv4di2<mask_name>): Likewise.
1180 (sse4_1_<code>v2hiv2di2<mask_name>): Likewise.
1181 (sse4_1_<code>v2siv2di2<mask_name>): Likewise.
1182 (*sse4_1_<code>v8qiv8hi2<mask_name>_1): New pattern.
1183 (*sse4_1_<code>v8qiv8hi2<mask_name>_2): Likewise.
1184 (*avx2_<code>v8qiv8si2<mask_name>_1): Likewise.
1185 (*avx2_<code>v8qiv8si2<mask_name>_2): Likewise.
1186 (*sse4_1_<code>v4qiv4si2<mask_name>_1): Likewise.
1187 (*sse4_1_<code>v4qiv4si2<mask_name>_2): Likewise.
1188 (*sse4_1_<code>v4hiv4si2<mask_name>_1): Likewise.
1189 (*sse4_1_<code>v4hiv4si2<mask_name>_2): Likewise.
1190 (*avx512f_<code>v8qiv8di2<mask_name>_1): Likewise.
1191 (*avx512f_<code>v8qiv8di2<mask_name>_2): Likewise.
1192 (*avx2_<code>v4qiv4di2<mask_name>_1): Likewise.
1193 (*avx2_<code>v4qiv4di2<mask_name>_2): Likewise.
1194 (*avx2_<code>v4hiv4di2<mask_name>_1): Likewise.
1195 (*avx2_<code>v4hiv4di2<mask_name>_2): Likewise.
1196 (*sse4_1_<code>v2hiv2di2<mask_name>_1): Likewise.
1197 (*sse4_1_<code>v2hiv2di2<mask_name>_2): Likewise.
1198 (*sse4_1_<code>v2siv2di2<mask_name>_1): Likewise.
1199 (*sse4_1_<code>v2siv2di2<mask_name>_2): Likewise.
1201 2018-11-21 H.J. Lu <hongjiu.lu@intel.com>
1203 * read-rtl.c (apply_subst_iterator): Handle define_split and
1204 define_insn_and_split.
1206 2018-11-21 Jakub Jelinek <jakub@redhat.com>
1208 PR rtl-optimization/87817
1209 * config/i386/i386.md (bmi2_bzhi_<mode>3, *bmi2_bzhi_<mode>3,
1210 *bmi2_bzhi_<mode>3_1, *bmi2_bzhi_<mode>3_1_ccz): Use IF_THEN_ELSE
1211 in the pattern to avoid triggering UB when operands[2] is zero.
1212 (tbm_bextri_<mode>): New expander. Renamed the old define_insn to ...
1213 (*tbm_bextri_<mode>): ... this.
1215 2018-11-21 Tom de Vries <tdevries@suse.de>
1218 * params.def (HOT_BB_COUNT_FRACTION): Terminate help message with
1221 2018-11-21 Andreas Krebbel <krebbel@linux.ibm.com>
1223 * configure.ac: Add check for Binutils to determine whether vector
1224 load/store alignments hints are being supported.
1225 * config.in: Regenerate.
1226 * configure: Regenerate.
1227 * config/s390/s390.c (print_operand): Support new output
1229 * config/s390/s390.md ("movti"): Append alignment hint output
1230 using the new output modifier 'A'.
1231 * config/s390/vector.md ("mov<mode>", "*vec_tf_to_v1tf")
1232 ("*vec_ti_to_v1ti"): Likewise.
1234 2018-11-20 Martin Sebor <msebor@redhat.com>
1236 * doc/extend.texi (Other Builtins): Add __builtin_has_attribute.
1238 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
1241 * ipa-devirt.c (odr_types_equivalent_p): Use operand_equal_p to
1242 compare ENUM values.
1244 2018-11-20 Jakub Jelinek <jakub@redhat.com>
1246 PR tree-optimization/87895
1247 * omp-simd-clone.c (ipa_simd_modify_function_body): When removing
1248 or replacing GIMPLE_RETURN, set EDGE_FALLTHRU on the edge to EXIT.
1249 (simd_clone_adjust): Don't set EDGE_FALLTHRU here. In a loop that
1250 redirects edges to EXIT to edges to incr_bb, iterate while EXIT
1251 has any preds and always use EDGE_PRED (, 0).
1253 2018-11-20 Uros Bizjak <ubizjak@gmail.com>
1256 * mode-switching.c (create_pre_exit): After reload, always split the
1257 fallthrough edge to the exit block.
1259 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
1261 * ipa-devirt.c (add_type_duplicate): Do not ICE on incomplete enums.
1262 * tree.c (build_array_type_1): Forward declare.
1263 (fld_type_variant_equal_p): Add INNER_TYPE parameter.
1264 (fld_type_variant): Likewise.
1265 (fld_simplified_types): New hash.
1266 (fld_process_array_type): New function.
1267 (fld_incomplete_type_of): Handle array and enumeration types.
1268 (fld_simplified_type): Handle simplification of arrays.
1269 (free_lang_data): Allocate and free simplified types hash.
1271 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
1274 * ipa-devirt.c (odr_subtypes_equivalent_p): Report ODR violation
1275 when sybtype already violates ODR.
1276 (get_odr_type): Do not ICE when insert is false and type duplicate
1277 is not registered yet.
1278 (register_odr_type): Be sure to register subtypes first.
1280 2018-11-20 Andreas Krebbel <krebbel@linux.ibm.com>
1282 * config/s390/s390.md ("clztidi2"): Swap the RTX's written to the
1283 DImode parts of the target operand.
1285 2018-11-20 Nathan Sidwell <nathan@acm.org>
1288 * Makefile.in (bitmap.o-warn): Use -Wno-error=array-bounds.
1290 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
1293 * tree.c (free_lang_data_in_cgraph): Add argument fld; break out
1295 (free_lang_data) ... here; update call of free_lang_data_in_cgraph.
1297 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
1300 * ipa-fnsummary.c (pass_ipa_fnsummary): Do not remove functions
1301 * ipa.c (possible_inline_candidate_p): Break out from ..
1302 (process_references): ... here ; drop before_inlining_p;
1303 cleanup handling of alises.
1304 (walk_polymorphic_call_targets): Likewise.
1305 (symbol_table::remove_unreachable_nodes): Likewise.
1306 * passes.c (pass_data_ipa_remove_symbols): New structure.
1307 (pass_ipa_remove_symbols): New pass.
1308 (make_pass_ipa_remove_symbols): New function.
1309 * tree-pass.h (make_pass_ipa_remove_symbols): Declare.
1310 * passes.def (pass_ipa_remove_symbols): Schedule after early passes.
1312 2018-11-20 Richard Biener <rguenther@suse.de>
1314 * tree-vect-stmts.c (vectorizable_condition): Do not get
1315 at else_clause vect def for EXTRACT_LAST_REDUCTION. Remove
1316 pointless vect_is_simple_use calls.
1318 2018-11-20 Richard Biener <rguenther@suse.de>
1320 PR tree-optimization/88074
1321 * tree-ssa-dom.c (pass_dominator::execute): Do not walk
1324 2018-11-20 Richard Biener <rguenther@suse.de>
1326 PR tree-optimization/88069
1327 * tree-ssa-sccvn.c (visit_phi): Do not value-number to unvisited
1328 virtual PHI arguments.
1330 2018-11-20 Ilya Leoshkevich <iii@linux.ibm.com>
1333 * config/s390/s390.md: Skip LT(G) peephole when literal pool is
1335 * rtl.h (contains_constant_pool_address_p): New function.
1336 * rtlanal.c (contains_constant_pool_address_p): Likewise.
1338 2018-11-20 Richard Biener <rguenther@suse.de>
1341 * alias.c (component_uses_parent_alias_set_from): Remove
1342 alias-set zero and TYPE_TYPELESS_STORAGE case both already
1343 handled in other ways.
1345 2018-11-20 Richard Biener <rguenther@suse.de>
1347 PR tree-optimization/88087
1348 * tree-ssa-pre.c (create_expression_by_pieces): Re-materialize
1350 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Remember
1353 2018-11-20 Richard Biener <rguenther@suse.de>
1356 * tree-data-ref.c (lambda_matrix_right_hermite): Use abs_hwi.
1358 2018-11-19 Jan Hubicka <hubicka@ucw.cz>
1361 * ipa-devirt.c (free_enum_values): Do not ICE on ODR vilations.
1363 2018-11-19 Sandra Loosemore <sandra@codesourcery.com>
1366 * doc/invoke.texi (Link Options): Mention shared libraries
1367 in documentation for the -l option. Simplify discussion and
1368 point to the system linker documentation for details.
1370 2018-11-19 Peter Bergner <bergner@linux.ibm.com>
1372 PR rtl-optimization/88033
1373 * ira-lives.c (non_conflicting_reg_copy_p): Skip copies from a register
1374 to itself. Use HARD_REGISTER_NUM_P.
1376 2018-11-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1378 * tree-vect-loop.c (vect_transform_loop): Disable further unrolling
1379 of the loop if vf is non-constant.
1381 2018-11-19 David Malcolm <dmalcolm@redhat.com>
1383 PR tree-optimization/87025
1384 * dumpfile.c (dump_context::begin_scope): Call end_any_optinfo
1385 immediately after creating the scope optinfo.
1386 (selftest::test_pr87025): New function.
1387 (selftest::dumpfile_c_tests): Call it.
1388 * optinfo-emit-json.cc (optrecord_json_writer::pop_scope): Assert
1389 that we're not popping the top-level records array.
1390 * optinfo.cc (optinfo::handle_dump_file_kind): Assert that we're
1391 not changing the kind of a "scope" optinfo.
1393 2018-11-19 David Malcolm <dmalcolm@redhat.com>
1395 PR tree-optimization/87025
1396 * dump-context.h: Include "optinfo.h".
1397 (class optrecord_json_writer): New forward decl.
1398 (dump_context::forcibly_enable_optinfo_p): Delete.
1399 (dump_context::optinfo_enabled_p): New member function.
1400 (dump_context::optimization_records_enabled_p): New member
1402 (dump_context::set_json_writer): New member function.
1403 (dump_context::emit_optinfo): New member function.
1404 (dump_context::m_forcibly_enable_optinfo): Delete.
1405 (dump_context::m_json_writer): New member data.
1406 * dumpfile.c (dump_context::set_json_writer): New member function.
1407 (dump_context::finish_any_json_writer): New member function.
1408 (dump_context::end_scope): Replace call to
1409 optimization_records_maybe_pop_dump_scope with call to
1410 m_json_writer->pop_scope.
1411 (dump_context::optinfo_enabled_p): New member function.
1412 (dump_context::end_any_optinfo): Replace call to optinfo::emit with call
1413 to dump_context::emit_optinfo.
1414 (dump_context::emit_optinfo): New member function.
1415 (temp_dump_context::temp_dump_context): Replace
1416 m_forcibly_enable_optinfo with call to set_json_writer.
1417 (temp_dump_context::~temp_dump_context): Clean up any json writer.
1418 * optinfo-emit-json.cc (class optrecord_json_writer): Move to
1420 (the_json_writer): Delete.
1421 (optimization_records_start): Delete.
1422 (optimization_records_finish): Delete.
1423 (optimization_records_enabled_p): Delete, in favor of
1424 dump_context::optimization_records_enabled_p.
1425 (optimization_records_maybe_record_optinfo): Delete.
1426 (optimization_records_maybe_pop_dump_scope): Delete.
1427 * optinfo-emit-json.h: Include "json.h". Delete forward
1429 (optimization_records_start): Delete.
1430 (optimization_records_finish): Delete.
1431 (optimization_records_enabled_p): Delete.
1432 (optimization_records_maybe_record_optinfo): Delete.
1433 (optimization_records_maybe_pop_dump_scope): Delete.
1434 (class optrecord_json_writer): Move here from
1435 optinfo-emit-json.cc.
1436 * optinfo.cc (optinfo::emit_for_opt_problem): Replace call
1437 to optinfo::emit with call to dump_context::emit_optinfo.
1438 (optinfo::emit): Delete, in favor of dump_context::emit_optinfo.
1439 (optinfo_enabled_p): Delete, in favor of
1440 dump_context::optinfo_enabled_p.
1441 (optinfo_wants_inlining_info_p): Update for conversion o
1442 optimization_records_enabled_p to a member function of
1444 * optinfo.h (optinfo_enabled_p): Delete, in favor of
1445 dump_context::optinfo_enabled_p.
1446 (optinfo::emit): Delete, in favor of dump_context::emit_optinfo.
1447 * toplev.c: Include "dump-context.h".
1448 (compile_file): Replace call to optimization_records_finish with
1449 dump_context::finish_any_json_writer.
1450 (do_compile): Replace call to optimization_records_start with
1451 conditionally creating a optrecord_json_writer for the
1454 2018-11-19 Jakub Jelinek <jakub@redhat.com>
1456 PR tree-optimization/88071
1457 * tree-vect-loop.c (vectorize_fold_left_reduction): Pass true instead
1458 of false as last argument to gsi_remove.
1459 * tree-vect-stmts.c (vect_finish_replace_stmt): Pass true instead of
1460 false as last argument to gsi_replace.
1463 * omp-expand.c: Don't include debug.h.
1464 (adjust_context_and_scope): Add REGION argument. Find DECL_CONTEXT
1465 from innermost outer parallel, task, teams or target that has a
1466 child_fn set, or, if there is no such outer region, use
1467 current_function_decl. Do the DECL_CONTEXT adjustment regardless of
1468 whether a suitable BLOCK is found or not.
1469 (expand_parallel_call, expand_teams_call): Don't call
1470 adjust_context_and_scope here.
1471 (grid_expand_target_grid_body): Revert 2017-01-25 changes.
1472 (expand_omp_taskreg, expand_omp_target): Likewise. Call
1473 adjust_context_and_scope.
1474 * dwarf2out.c (dwarf2out_early_global_decl): For
1475 decl_function_context recurse instead of calling dwarf2out_decl.
1477 2018-11-20 Eric Botcazou <ebotcazou@adacore.com>
1479 PR rtl-optimization/85925
1480 * rtl.h (word_register_operation_p): New predicate.
1481 * combine.c (record_dead_and_set_regs_1): Only apply specific handling
1482 for WORD_REGISTER_OPERATIONS targets to word_register_operation_p RTX.
1483 * rtlanal.c (nonzero_bits1): Likewise. Adjust couple of comments.
1484 (num_sign_bit_copies1): Likewise.
1486 2018-11-19 Richard Biener <rguenther@suse.de>
1489 * tree.c (free_lang_data_in_one_sizepos): Free non-gimple-val
1492 2018-11-19 Eric Botcazou <ebotcazou@adacore.com>
1494 * compare-elim.c (struct comparison): Add not_in_a field.
1495 (is_not): New static function.
1496 (strip_not): Likewise.
1497 (conforming_compare): Handle a NOT in the first operand.
1498 (can_eliminate_compare): Likewise.
1499 (find_comparison_dom_walker::before_dom_children): Likewise.
1500 (try_eliminate_compare): Likewise.
1501 * config/visium/visium.md (negsi2_insn_set_carry): Turn into...
1502 (neg<mode>2_insn_set_carry): ...this and add missing NEG operation.
1504 2018-11-19 Jonathan Wakely <jwakely@redhat.com>
1506 * doc/extend.texi (Common Type Attributes): Fix typos.
1508 2018-11-19 Martin Liska <mliska@suse.cz>
1510 PR gcov-profile/88045
1511 * coverage.c (coverage_begin_function): Add assert.
1513 2018-11-18 Sandra Loosemore <sandra@codesourcery.com>
1516 * doc/extend.texi (Common Function Attributes): Document that
1517 no_instrument_function applies to -p and -pg, too.
1518 * doc/invoke.texi (Instrumentation Options): Add cross-references
1519 to docs for -p, -pg, and -finstrument-functions.
1521 2018-11-18 Alan Modra <amodra@gmail.com>
1523 * config/gnu-user.h (GNU_USER_TARGET_CRTI): Define.
1524 (GNU_USER_TARGET_STARTFILE_SPEC): Use it here.
1525 (GNU_USER_TARGET_CRTN): Define.
1526 (GNU_USER_TARGET_ENDFILE_SPEC): Use it here.
1527 (GNU_USER_TARGET_LINK_GCC_C_SEQUENCE_SPEC): Define.
1529 2018-11-18 Alan Modra <amodra@gmail.com>
1531 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Delete
1532 !HAVE_LD_PIE variant.
1533 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
1535 2018-11-17 Nick Clifton <nickc@redhat.com>
1536 Sandra Loosemore <sandra@codesourcery.com>
1539 * doc/invoke.texi (Overall Options): Document that --help and
1540 --help=<value> options cannot be combined.
1542 2018-11-17 Jim Wilson <jimw@sifive.com>
1544 * config/riscv/riscv.c (epilogue_cfa_sp_offset): New.
1545 (riscv_restore_reg): If restoring HARD_FRAME_POINTER_REGNUM, and
1546 epilogue_cfa_sp_offset set, then add REG_CFA_DEF_CFA regnote.
1547 (riscv_expand_epilogue): Initialize epilogue_cfa_sp_offset. Set it
1548 to step2 if frame_pointer_needed and step1 is 0.
1550 2018-11-17 Sandra Loosemore <sandra@codesourcery.com>
1553 * doc/cpp.texi (System Headers): Add note about implicit
1554 extern "C" block on targets that define SYSTEM_IMPLICIT_EXTERN_C.
1556 2018-11-17 Sandra Loosemore <sandra@codesourcery.com>
1559 * doc/invoke.texi (C++ Dialect Options): Clarify usage of
1560 -fno-implicit-templates.
1562 2018-11-17 Jakub Jelinek <jakub@redhat.com>
1564 PR tree-optimization/87546
1565 * tree-vect-patterns.c (vect_look_through_possible_promotion): Add
1566 min_precision variable, initially set it to orig_precision, only does
1567 something if op_type's precision is <= min_precision and update
1568 min_precision whenever calling set_op.
1570 2018-11-16 Jan Hubicka <hubicka@ucw.cz>
1573 * ipa-devirt.c (warn_odr): Look for main variant to get TYPE_DECL.
1575 2018-11-16 Sandra Loosemore <sandra@codesourcery.com>
1577 * doc/invoke.texi (Option Summary): Fix whitespace and line
1578 breaks in @gccoptlist environments.
1579 (Warning Options): Likewise.
1580 (Optimize Options): Likewise.
1581 (PowerPC SPE Options): Likewise.
1582 (RS/6000 and PowerPC Options): Likewise.
1584 2018-11-16 Jeff Law <law@redhat.com>
1586 * config/mn10300/mn10300.md (adddi3_degenerate): Remove bogus
1589 2018-11-16 Sandra Loosemore <sandra@codesourcery.com>
1592 * doc/invoke.texi (Optimize Options): Update options enabled by
1593 fprofile-generate, -fprofile-use, and -fauto-profile.
1595 2018-11-16 Jozef Lawrynowicz <jozef.l@mittosystems.com>
1598 * target-def.h: Initialize TARGET_ASM_{,UN}ALIGNED_P{S,D,T}I_OP.
1599 Add them to the TARGET_ASM_{,UN}ALIGNED_INT_OP structs.
1600 * target.def: Enumerate TARGET_ASM_{,UN}ALIGNED_P{S,D,T}I_OP in
1602 * target.h: Add psi, pdi, pti to struct asm_int_op definition.
1603 * targhooks.c (default_print_patchable_function_entry): Assert
1604 asm_int_op does not return a NULL string.
1605 * varasm.c (integer_asm_op): Return the op for a partial int type
1606 when the requested size does not correspond to an integer type.
1607 * config/msp430/msp430.c: Initialize TARGET_ASM_{,UN}ALIGNED_PSI_OP.
1608 * doc/tm.texi: Regenerate.
1610 2018-11-12 Jason Merrill <jason@redhat.com>
1612 * gimplify.c (gimplify_case_label_expr): Handle hot/cold attributes.
1614 2018-11-16 Michael Meissner <meissner@linux.ibm.com>
1616 * config/rs6000/constraints.md (wF constraint): Remove power9
1617 fusion documentation. Just document wF's use for power8 fusion.
1618 * config/rs6000/predicates.md (p9_fusion_reg_operand): Delete.
1619 (fusion_gpr_addis): Delete power9 fusion support. Change power8
1620 fusion support to require the upper 12 bits to be all 0's or all
1622 (fusion_gpr_mem_load): Add comment.
1623 (fusion_addis_mem_combo_load): Remove power9 fusion support.
1624 (fusion_addis_mem_combo_store): Delete.
1625 (fusion_offsettable_mem_operand): Delete.
1626 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Do not set
1628 (ISA_3_0_MASKS_SERVER): Delete power9 fusion.
1629 (POWERPC_MASKS): Delete power9 fusion.
1630 * config/rs6000/rs6000-protos.h (emit_fusion_load_store): Delete.
1631 (fusion_p9_p): Delete.
1632 (expand_fusion_p9_load): Delete.
1633 (expand_fusion_p9_store): Delete.
1634 (emit_fusion_p9_load): Delete.
1635 (emit_fusion_p9_store): Delete.
1636 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Delete power9
1638 (rs6000_option_override_internal): Set power8 fusion based on
1639 whether we are tuning for power8. Delete power9 fusion support.
1640 (rs6000_opt_masks): Delete -mpower9-fusion switch.
1641 (emit_fusion_load): Rename emit_fusion_load_store to
1642 emit_fusion_load, and drop fusion store support. Update callers.
1643 (emit_fusion_load_store): Likewise.
1644 (emit_fusion_gpr_load): Likewise.
1645 (fusion_p9_p): Delete.
1646 (expand_fusion_p9_load): Delete.
1647 (expand_fusion_p9_store): Delete.
1648 (emit_fusion_p9_load): Delete.
1649 (emit_fusion_p9_store): Delete.
1650 * config/rs6000/rs6000.md (UNSPEC_FUSION_P9): Delete.
1651 (GPR_FUSION): Delete.
1652 (FPR_FUSION): Delete.
1653 (power9 fusion peephole2s): Delete.
1654 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Delete.
1655 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Delete.
1656 (fusion_vsx_<P:mode>_<FPR_FUSION:mode>_load): Delete.
1657 (fusion_vsx_<P:mode>_<FPR_FUSION:mode>_store): Delete.
1658 (fusion_p9_<mode>_constant): Delete.
1659 * config/rs6000/rs6000.opt (-mpower9-fusion): Delete undocumented
1660 power9 fusion switch.
1661 * doc/md.texi (PowerPC constraints): Update wF constraint
1662 documentation for power8 fusion only.
1664 2018-11-16 Jakub Jelinek <jakub@redhat.com>
1666 PR rtl-optimization/87475
1667 * cfgrtl.c (patch_jump_insn): Allow redirection failure for
1668 CROSSING_JUMP_P insns.
1669 (cfg_layout_redirect_edge_and_branch): Don't ICE if ret is NULL.
1671 2018-11-16 Uros Bizjak <ubizjak@gmail.com>
1674 * config/i386/i386.md (floatunsdidf2): Allow only 64bit AVX512F targets.
1675 * config/i386/sse.md (UNSPEC_MOVDI_TO_SSE): New UNSPEC.
1676 (movdi_to_sse): Rewrite using UNSPEC_MOVDI_TO_SSE unspec.
1678 2018-11-16 Jakub Jelinek <jakub@redhat.com>
1681 * optabs.c (expand_binop): For op0_mode use GET_MODE (op0), unless it
1682 is VOIDmode, in which case use int_mode. Similarly for op1_mode.
1684 2018-11-16 Eric Botcazou <ebotcazou@adacore.com>
1686 * tree-switch-conversion.h (switch_decision_tree::emit_case_nodes): Add
1687 location_t parameter.
1688 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
1689 (switch_decision_tree::do_jump_if_equal): Likewise.
1690 * tree-switch-conversion.c (switch_decision_tree::emit): Pass location
1691 of switch statement to emit_case_nodes.
1692 (switch_decision_tree::emit_cmp_and_jump_insns): Add LOC parameter and
1693 set it on the newly built GIMPLE comparison statement.
1694 (switch_decision_tree::do_jump_if_equal): Likewise.
1695 (switch_decision_tree::emit_case_nodes): Add LOC parameter and pass it
1696 in calls to do_jump_if_equal as well as recursive calls.
1698 2018-11-16 Andi Kleen <ak@linux.intel.com>
1700 * config/i386/i386.md: Allow memory operands to ptwrite.
1702 2018-11-16 Michael Matz <matz@suse.de>
1704 * system.h (PRsa): New macro.
1705 (SIZE_AMOUNT): Cast number to uint64_t.
1706 * alloc-pool.h (pool_usage::dump): Don't use %zu but PRsa.
1707 (pool_usage::dump_footer): Likewise and also use PRIu64.
1708 * bitmap.h (bitmap_usage::dump): Likewise.
1709 * ggc-common.c (ggc_usage::dump): Likewise.
1710 * ggc-page.c (ggc_print_statistics): Likewise.
1711 * input.c (dump_line_table_statistics): Likewise.
1712 * mem-stats.h (mem_usage::dump): Likewise.
1713 (mem_usage::dump_footer): Likewise.
1714 * rtl.c (dump_rtx_statistics): Likewise.
1715 * tree-cfg.c (dump_cfg_stats): Likewise.
1716 * tree-dfa.c (dump_dfa_stats): Likewise.
1717 * tree-phinodes.c (phinodes_print_statistics): Likewise.
1718 * tree-ssanames (ssanames_print_statistics): Likewise.
1719 * vec.c (vec_usage::dump): Likewise.
1720 (vec_usage::dump_footer): Likewise.
1722 2018-11-16 Richard Biener <rguenther@suse.de>
1724 PR tree-optimization/88011
1725 * tree-vrp.c (extract_range_from_binary_expr): Fix error in
1726 replacing set_value_range_to_undefined and
1727 set_value_range_to_varying with method calls.
1729 2018-11-16 Ilya Leoshkevich <iii@linux.ibm.com>
1731 * config/s390/s390.md
1732 (*r<noxa>sbg_<mode>_srl_bitmask): Do not delegate arithmetic to
1734 (*r<noxa>sbg_<mode>_sll): Likewise.
1735 (*r<noxa>sbg_<mode>_srl): Likewise.
1736 (*r<noxa>sbg_sidi_srl): New pattern.
1738 2018-11-16 Jerome Lambourg <lambourg@adacore.com>
1740 * config/vxworks.h (TARGET_VXWORKS_HAVE_CTORS_DTORS): New macro.
1741 Default to TARGET_VXWORKS_RTP.
1742 (SUPPORTS_INIT_PRIORITY): Use TARGET_VXWORKS_HAVE_CTORS_DTORS instead
1743 of TARGET_VXWORKS_RTP.
1744 * config/vxworksae.h: Also define TARGET_VXWORKS_HAVE_CTORS_DTORS.
1745 * config/vxworks.c: Use TARGET_VXWORKS_HAVE_CTORS_DTORS instead
1746 of TARGET_VXWORKS_RTP to set targetm.have_ctors_dtors.
1748 2018-11-15 Sandra Loosemore <sandra@codesourcery.com>
1751 * doc/extend.texi (Common Type Attributes): Make it explicit
1752 that attribute "packed" can apply to C++ classes.
1754 2018-11-15 Martin Sebor <msebor@redhat.com>
1758 * tree.c (type_argument_type): New function.
1759 * tree.h (type_argument_type): Declare it.
1760 * gcc/doc/extend.texi (alloc_align): Update and clarify.
1761 (alloc_size, nonnull, sentinel): Same.
1763 2018-11-15 Andrew Stubbs <ams@codesourcery.com>
1764 Kwok Cheung Yeung <kcy@codesourcery.com>
1766 * tree-vect-stmts.c (vectorizable_store): Don't ICE when
1767 int_mode_for_size fails.
1768 (vectorizable_load): Likewise.
1770 2018-11-15 David Malcolm <dmalcolm@redhat.com>
1772 * doc/ux.texi (Group logically-related diagnostics): Move
1773 discussion of auto_diagnostic_group into this new subsection.
1774 Give an example of where this grouping is used.
1776 2018-11-15 David Malcolm <dmalcolm@redhat.com>
1779 * Makefile.in (OBJS): Move json.o to...
1780 (OBJS-libcommon): ...here and add diagnostic-format-json.o.
1781 * common.opt (fdiagnostics-format=): New option.
1782 (diagnostics_output_format): New enum.
1783 * diagnostic-format-json.cc: New file.
1784 * diagnostic.c (default_diagnostic_final_cb): New function, taken
1785 from start of diagnostic_finish.
1786 (diagnostic_initialize): Initialize final_cb to
1787 default_diagnostic_final_cb.
1788 (diagnostic_finish): Move "being treated as errors" messages to
1789 default_diagnostic_final_cb. Call any final_cb.
1790 (default_diagnostic_finalizer): Add diagnostic_t param.
1791 (diagnostic_report_diagnostic): Pass "orig_diag_kind" to
1792 diagnostic_finalizer callback.
1793 * diagnostic.h (enum diagnostics_output_format): New enum.
1794 (diagnostic_finalizer_fn): Reimplement, adding diagnostic_t param.
1795 (struct diagnostic_context): Add "final_cb".
1796 (default_diagnostic_finalizer): Add diagnostic_t param.
1797 (diagnostic_output_format_init): New decl.
1798 * doc/invoke.texi (-fdiagnostics-format): New option.
1799 * dwarf2out.c (gen_producer_string): Ignore
1800 OPT_fdiagnostics_format_.
1801 * gcc.c (driver_handle_option): Handle OPT_fdiagnostics_format_.
1802 * lto-wrapper.c (append_diag_options): Ignore it.
1803 * opts.c (common_handle_option): Handle it.
1805 2018-11-15 David Malcolm <dmalcolm@redhat.com>
1807 PR tree-optimization/88015
1808 * graphite-isl-ast-to-gimple.c
1809 (translate_isl_ast_to_gimple::scop_to_isl_ast): Add missing check
1811 * graphite-sese-to-poly.c (build_poly_scop): Likewise.
1813 2018-11-15 Richard Biener <rguenther@suse.de>
1816 * gimple.c (gimple_call_flags): Union flags from decl, type
1818 * trans-mem.c (is_tm_pure_call): Simplify.
1820 2018-11-15 Richard Biener <rguenther@suse.de>
1822 PR tree-optimization/88031
1823 * tree-vect-loop.c (vectorizable_reduction): Move check
1824 for multiple types earlier so we get the expected dump.
1825 Simplify calls to vectorizable_condition.
1826 * tree-vect-stmts.h (vectorizable_condition): Update prototype.
1827 * tree-vect-stmts.c (vectorizable_condition): Instead of
1828 reduc_def and reduc_index take just a flag. Simplify
1829 code-generation now that we can rely on the defs being set up.
1830 (vectorizable_comparison): Remove unused argument.
1832 2018-11-15 Richard Biener <rguenther@suse.de>
1834 PR tree-optimization/88030
1835 * tree-complex.c (need_eh_cleanup): New global.
1836 (update_complex_assignment): Mark blocks that need EH update.
1837 (expand_complex_comparison): Likewise.
1838 (tree_lower_complex): Allocate and deallocate need_eh_cleanup,
1839 perform EH cleanup and schedule CFG cleanup if that did anything.
1841 2018-11-15 Jakub Jelinek <jakub@redhat.com>
1843 PR rtl-optimization/88018
1844 * cfgrtl.c (fixup_abnormal_edges): Guard moving insns to fallthru edge
1845 on the presence of fallthru edge, rather than if it is a USE or not.
1847 2018-11-15 Richard Biener <rguenther@suse.de>
1850 * tree-data-ref.c (analyze_miv_subscript): Guard calls to
1851 analyze_subscript_affine_affine properly.
1853 2018-11-15 Xianmiao Qu <xianmiao_qu@c-sky.com>
1855 * config/csky/csky-linux-elf.h (LINUX_DYNAMIC_LINKER): Remove.
1856 (GLIBC_DYNAMIC_LINKER): Define.
1857 (LINUX_TARGET_LINK_SPEC): Update the dynamic linker's name.
1859 2018-11-15 Bin Cheng <bin.cheng@linux.alibaba.com>
1861 PR tree-optimization/84648
1862 * tree-ssa-loop-niter.c (adjust_cond_for_loop_until_wrap): New.
1863 (number_of_iterations_cond): Adjust exit cond for loop-until-wrap case
1864 by calling adjust_cond_for_loop_until_wrap.
1866 2018-11-15 Sandra Loosemore <sandra@codesourcery.com>
1869 * doc/extend.texi (Common Function Attributes): Clarify linker
1870 restrictions on "aligned" attribute.
1871 (Common Variable Attributes): Likewise. Mention that linker
1872 restrictions don't apply to stack-allocated variables.
1874 2018-11-15 Joern Rennecke <joern.rennecke@riscy-ip.com>
1876 * gensupport.c (add_predicate_code): Properly handle ZERO_EXTRACT
1879 2018-11-14 Sandra Loosemore <sandra@codesourcery.com>
1883 * doc/invoke.texi (Optimize Options): Remove bad example about
1884 interaction between -flto and -O. Replace it with a note that
1885 you need to compile with -O and not just link. Copy-edit -flto
1886 discussion to reduce verbiage and improve flow.
1888 2018-11-14 Sandra Loosemore <sandra@codesourcery.com>
1891 * doc/invoke.texi (Optimize Options): Clarify that -O0 and -Og
1892 also suppress many optimizations. Alphabetize option lists for
1893 -O1, -O2, and -Os. Add list of options disabled with -Og, and
1894 correct documentation for those options to say that.
1895 * opts.c (default_options_table): Sort table by level and option
1896 name, to make it easier to correlate to the manual.
1898 2018-11-14 Uros Bizjak <ubizjak@gmail.com>
1900 * config/i386/i386.c (ix86_print_operand_address_as): Simplify
1901 printing of the names of segment registers.
1903 2018-11-14 Nathan Sidwell <nathan@acm.org>
1907 * dwarf2out.c (dwarf2out_finish): Apply resolve_addr to comdat
1910 2018-11-14 David Malcolm <dmalcolm@redhat.com>
1912 * Makefile.in (CFLAGS-optinfo-emit-json.o): Add $(ZLIBINC).
1914 2018-11-14 Jakub Jelinek <jakub@redhat.com>
1916 P1236R1 - Signed integers are two's complement
1917 * doc/invoke.texi (Wshift-overflow): Adjust documentation for c++2a.
1920 * hash-map.h (hash_map::iterator::reference_pair): New class.
1921 (hash_map::iterator::operator*): Return it rather than std::pair.
1923 2018-11-14 Jeff Law <law@redhat.com>
1925 * optabs.c (expand_binop): Pass INT_MODE to operand_subword_force
1926 iff the operand is a constant.
1928 2018-11-14 Aldy Hernandez <aldyh@redhat.com>
1930 * gimple-ssa-evrp-analyze.c
1931 (evrp_range_analyzer::record_ranges_from_incoming_edge): Rename
1932 ignore_equivs_equal_p to equal_p.
1933 * ipa-cp.c (meet_with_1): Use equal_p instead of
1934 ignore_equivs_equal_p.
1935 * ipa-prop.c (ipa_vr_ggc_hash_traits::equal): Same.
1936 * tree-vrp.c (value_range::ignore_equivs_equal_p): Remove.
1937 (value_range::operator==): Remove.
1938 (value_range::operator!=): Remove.
1939 (vrp_prop::visit_stmt): Use equal_p.
1940 * tree-vrp.h (value_range): Remove operator==, operator!=,
1941 ignore_equivs_equal_p.
1942 * vr-values.c (update_value_range): Use equal_p.
1944 2018-11-14 Michael Matz <matz@suse.de>
1947 * gimplify.c (collect_fallthrough_labels): Add new argument,
1948 return location via that, don't modify statements.
1949 (warn_implicit_fallthrough_r): Adjust call, don't use
1950 statement location directly.
1952 2018-11-14 Richard Biener <rguenther@suse.de>
1955 * tree-data-ref.c (split_constant_offset): Add wrapper
1956 allocating a cache hash-map.
1957 (split_constant_offset_1): Cache results of expanding
1958 expressions from SSA def stmts.
1960 2018-11-14 Richard Biener <rguenther@suse.de>
1963 * tree-data-ref.c (lambda_matrix_row_add): Change const1 argument
1965 (lambda_vector_mult_const): Likewise.
1966 (lambda_matrix_right_hermite): Use lambda_int temporaries.
1968 2018-11-14 Wilco Dijkstra <wdijkstr@arm.com>
1969 Jackson Woodruff <jackson.woodruff@arm.com>
1971 PR 71026/tree-optimization
1972 * match.pd: Simplify floating point comparisons.
1974 2018-11-14 Jakub Jelinek <jakub@redhat.com>
1976 PR rtl-optimization/87817
1977 * config/i386/i386.c (ix86_fold_builtin): For _bzhi_u{32,64} if
1978 last argument has low 8 bits clear, fold to 0.
1980 2018-11-14 Iain Sandoe <iain@sandoe.co.uk>
1982 * tree-vect-data-refs.c (vect_can_force_dr_alignment_p): Cast
1983 MAX_OFILE_ALIGNMENT as needed.
1985 2018-11-14 Richard Biener <rguenther@suse.de>
1987 PR tree-optimization/88019
1988 * graphite-isl-ast-to-gimple.c (ternary_op_to_tree): Rewrite
1989 COND_EXPR arguments to non-trapping overflow.
1991 2018-11-14 Claudiu Zissulescu <claziss@synopsys.com>
1993 * config/arc/arc.h (reg_class): Reorder registers classes, remove
1994 unused register classes.
1995 (REG_CLASS_NAMES): Likewise.
1996 (REG_CLASS_CONTENTS): Likewise.
1997 (FIXED_REGISTERS): Make lp_count fixed.
1998 (BASE_REG_CLASS): Remove ACC16_BASE_REGS reference.
1999 (PROGRAM_COUNTER_REGNO): Remove.
2000 * config/arc/arc.c (arc_conditional_register_usage): Remove unused
2001 register classes, use constants for register numbers, remove
2002 reg_class_contents references.
2003 (arc_process_double_reg_moves): Add asserts.
2004 (arc_secondary_reload): Remove LPCOUNT_REG reference, use
2005 lra_in_progress predicate.
2006 (arc_init_reg_tables): Remove unused register classes.
2007 (arc_register_move_cost): Likewise.
2008 (arc_preferred_reload_class): Likewise.
2009 (hwloop_optimize): Update rtx patterns involving lp_count register.
2010 (arc_return_address_register): Rename ILINK1, INLINK2 regnums macros.
2011 * config/arc/constraints.md ("c"): Choose between GENERAL_REGS and
2012 CHEAP_CORE_REGS. Former one will be used for LRA.
2013 ("Rac"): Choose between GENERAL_REGS and ALL_CORE_REGS. Former
2014 one will be used for LRA.
2015 ("w"): Choose between GENERAL_REGS and WRITABLE_CORE_REGS. Former
2016 one will be used for LRA.
2017 ("W"): Choose between GENERAL_REGS and MPY_WRITABLE_CORE_REGS.
2018 Former one will be used for LRA.
2019 ("f"): Delete constraint.
2022 ("l"): Change it from register constraint to constraint.
2023 * config/arc/arc.md (movqi_insn): Remove unsed lp_count constraints.
2024 (movhi_insn): Likewise.
2025 (movsi_insn): Update pattern.
2028 (stack_tie): Remove 'b' constraint letter.
2030 (R9_REG, R15_REG, R16_REG, R25_REG): Likewise.
2031 (R32_REG, R40_REG, R41_REG, R42_REG, R43_REG, R44_REG): Likewise.
2032 (R57_REG, R59_REG, PCL_REG): Likewise.
2033 (ILINK1_REGNUM): Renamed to ILINK1_REG.
2034 (ILINK2_REGNUM): Renamed to ILINK2_REG.
2036 (SP_REGS): Likewise.
2037 (Rcw): Remove unused reg classes.
2038 * config/arc/predicates.md (dest_reg_operand): Just default on
2039 register_operand predicate.
2040 (mpy_dest_reg_operand): Likewise.
2041 (move_dest_operand): Use macros instead of constants.
2042 * config/arc/arc.opt (mlra): Switch to lra as default.
2044 2018-11-14 Richard Biener <rguenther@suse.de>
2046 PR tree-optimization/87974
2047 * tree-vect-loop.c (vectorizable_reduction): When computing
2048 the vectorized reduction PHI vector type ignore constant
2051 2018-11-14 Jakub Jelinek <jakub@redhat.com>
2053 PR tree-optimization/87977
2054 * tree-ssa-math-opts.c (optimize_recip_sqrt): Don't reuse division
2055 stmt, build a new one and replace the old one with it. Formatting fix.
2056 Call release_ssa_name (x) if !has_other_use and !delete_div.
2057 (pass_cse_reciprocals::execute): Before calling optimize_recip_sqrt
2058 verify lhs of stmt is still def.
2060 2018-11-13 Peter Bergner <bergner@linux.ibm.com>
2062 PR rtl-optimization/87507
2063 * lower-subreg.c (operand_for_swap_move_operator): New function.
2064 (simple_move): Strip simple operators.
2065 (find_pseudo_copy): Likewise.
2066 (resolve_operand_for_swap_move_operator): New function.
2067 (resolve_simple_move): Strip simple operators and swap operands.
2069 2018-11-13 Jakub Jelinek <jakub@redhat.com>
2071 PR tree-optimization/87898
2072 * omp-simd-clone.c (ipa_simd_modify_stmt_ops): Formatting fix.
2073 (ipa_simd_modify_function_body): Remove debug stmts where the first
2074 argument was changed into a non-decl.
2076 2018-11-13 Peter Bergner <bergner@linux.ibm.com>
2078 PR rtl-optimization/87899
2079 * lra-lives.c (start_living): Update white space in comment.
2080 (enum point_type): New.
2081 (sparseset_contains_pseudos_p): New function.
2082 (update_pseudo_point): Likewise.
2083 (make_hard_regno_live): Use HARD_REGISTER_NUM_P macro.
2084 (make_hard_regno_dead): Likewise. Remove ignore_reg_for_conflicts
2085 handling. Move early exit after adding conflicts.
2086 (mark_pseudo_live): Use HARD_REGISTER_NUM_P macro. Add early exit
2087 if regno is already live. Remove all handling of program points.
2088 (mark_pseudo_dead): Use HARD_REGISTER_NUM_P macro. Add early exit
2089 after adding conflicts. Remove all handling of program points and
2090 ignore_reg_for_conflicts.
2091 (mark_regno_live): Use HARD_REGISTER_NUM_P macro. Remove return value
2092 and do not guard call to mark_pseudo_live.
2093 (mark_regno_dead): Use HARD_REGISTER_NUM_P macro. Remove return value
2094 and do not guard call to mark_pseudo_dead.
2095 (check_pseudos_live_through_calls): Use HARD_REGISTER_NUM_P macro.
2096 (process_bb_lives): Use HARD_REGISTER_NUM_P and HARD_REGISTER_P macros.
2097 Use new function update_pseudo_point. Handle register copies by
2098 removing the source register from the live set. Handle INOUT operands.
2099 Update to the next program point using the unused_set, dead_set and
2101 (lra_create_live_ranges_1): Use HARD_REGISTER_NUM_P macro.
2103 2018-11-13 David Malcolm <dmalcolm@redhat.com>
2105 * builtins.c: Replace "source_location" with "location_t".
2106 * diagnostic-show-locus.c: Likewise.
2107 * diagnostic.c: Likewise.
2108 * dumpfile.c: Likewise.
2109 * gcc-rich-location.h: Likewise.
2110 * genmatch.c: Likewise.
2111 * gimple.h: Likewise.
2112 * gimplify.c: Likewise.
2113 * input.c: Likewise.
2114 * input.h: Likewise. Eliminate the typedef.
2115 * omp-expand.c: Likewise.
2116 * selftest.h: Likewise.
2117 * substring-locations.h (get_source_location_for_substring):
2119 (get_location_within_string): ...this.
2120 * tree-cfg.c: Replace "source_location" with "location_t".
2121 * tree-cfgcleanup.c: Likewise.
2122 * tree-diagnostic.c: Likewise.
2123 * tree-into-ssa.c: Likewise.
2124 * tree-outof-ssa.c: Likewise.
2125 * tree-parloops.c: Likewise.
2126 * tree-phinodes.c: Likewise.
2127 * tree-phinodes.h: Likewise.
2128 * tree-ssa-loop-ivopts.c: Likewise.
2129 * tree-ssa-loop-manip.c: Likewise.
2130 * tree-ssa-phiopt.c: Likewise.
2131 * tree-ssa-phiprop.c: Likewise.
2132 * tree-ssa-threadupdate.c: Likewise.
2133 * tree-ssa.c: Likewise.
2134 * tree-ssa.h: Likewise.
2135 * tree-vect-loop-manip.c: Likewise.
2137 2018-11-13 Michael Ploujnikov <michael.ploujnikov@oracle.com>
2139 * doc/extend.texi: Fix typo in the weakref description.
2141 2018-11-13 Richard Biener <rguenther@suse.de>
2143 PR tree-optimization/86991
2144 * tree-vect-loop.c (vect_is_slp_reduction): Delay reduction
2145 group building until we have successfully detected the SLP
2147 (vect_is_simple_reduction): Remove fixup code here.
2149 2018-11-13 David Malcolm <dmalcolm@redhat.com>
2151 * dumpfile.c (VERIFY_DUMP_ENABLED_P): New macro.
2152 (dump_gimple_stmt): Use it.
2153 (dump_gimple_stmt_loc): Likewise.
2154 (dump_gimple_expr): Likewise.
2155 (dump_gimple_expr_loc): Likewise.
2156 (dump_generic_expr): Likewise.
2157 (dump_generic_expr_loc): Likewise.
2158 (dump_printf): Likewise.
2159 (dump_printf_loc): Likewise.
2160 (dump_dec): Likewise.
2161 (dump_dec): Likewise.
2162 (dump_hex): Likewise.
2163 (dump_symtab_node): Likewise.
2164 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
2165 Guard dump call with dump_enabled_p.
2166 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl): Likewise.
2167 * graphite-optimize-isl.c (optimize_isl): Likewise.
2168 * graphite.c (graphite_transform_loops): Likewise.
2169 * tree-loop-distribution.c (pass_loop_distribution::execute): Likewise.
2170 * tree-parloops.c (parallelize_loops): Likewise.
2171 * tree-ssa-loop-niter.c (number_of_iterations_exit): Likewise.
2172 * tree-vect-data-refs.c (vect_analyze_group_access_1): Likewise.
2173 (vect_prune_runtime_alias_test_list): Likewise.
2174 * tree-vect-loop.c (vect_update_vf_for_slp): Likewise.
2175 (vect_estimate_min_profitable_iters): Likewise.
2176 * tree-vect-slp.c (vect_record_max_nunits): Likewise.
2177 (vect_build_slp_tree_2): Likewise.
2178 (vect_supported_load_permutation_p): Likewise.
2179 (vect_slp_analyze_operations): Likewise.
2180 (vect_slp_analyze_bb_1): Likewise.
2181 (vect_slp_bb): Likewise.
2182 * tree-vect-stmts.c (vect_analyze_stmt): Likewise.
2183 * tree-vectorizer.c (try_vectorize_loop_1): Likewise.
2184 (pass_slp_vectorize::execute): Likewise.
2185 (increase_alignment): Likewise.
2187 2018-11-13 David Malcolm <dmalcolm@redhat.com>
2190 * ipa-inline.c (report_inline_failed_reason): Guard calls to
2191 cl_target_option_print_diff and cl_optimization_print_diff with
2194 2018-11-13 David Malcolm <dmalcolm@redhat.com>
2196 * doc/invoke.texi (-fsave-optimization-record): Note that the
2197 output is compressed.
2198 * optinfo-emit-json.cc: Include <zlib.h>.
2199 (optrecord_json_writer::write): Compress the output.
2201 2018-11-13 Aldy Hernandez <aldyh@redhat.com>
2203 * tree-vrp.c (value_range_base::dump): Dump type.
2204 Do not use INF nomenclature for 1-bit types.
2205 (dump_value_range): Group all variants to common dumping code.
2206 (debug): New overloaded functions for value_ranges.
2207 (value_range_base::dump): Remove no argument version.
2208 (value_range::dump): Same.
2210 2018-11-13 Richard Biener <rguenther@suse.de>
2212 PR tree-optimization/87931
2213 * tree-vect-loop.c (vect_is_simple_reduction): Restrict
2214 nested cycles we support to latch computations vectorizable_reduction
2217 2018-11-13 Martin Liska <mliska@suse.cz>
2219 PR tree-optimization/87885
2220 * cfghooks.c (account_profile_record): Rename
2222 (profile_record_check_consistency): ... this.
2223 Calculate missing num_mismatched_freq_in.
2224 (profile_record_account_profile): New function
2225 that calculates time and size of a function.
2226 * cfghooks.h (struct profile_record): Remove
2228 (struct cfg_hooks): Remove after_pass flag.
2229 (account_profile_record): Rename to ...
2230 (profile_record_check_consistency): ... this.
2231 (profile_record_account_profile): New.
2232 * cfgrtl.c (rtl_account_profile_record): Remove
2234 * passes.c (check_profile_consistency): Do only
2236 (account_profile): Calculate size and time of
2238 (pass_manager::dump_profile_report): Reformat
2240 (execute_one_ipa_transform_pass): Call
2241 consistency check before clean upand call account_profile
2242 after a clean up is done.
2243 (execute_one_pass): Call check_profile_consistency and
2244 account_profile instead of using after_pass flag..
2245 * tree-cfg.c (gimple_account_profile_record): Likewise.
2247 2018-11-13 Martin Liska <mliska@suse.cz>
2250 * config/i386/i386.c (ix86_option_override_internal): Error
2251 about usage -mabi=ms and -fsanitize={,kernel-}address.
2253 2018-11-13 Andre Vieira <andre.simoesdiasvieira@arm.com>
2255 * config/aarch64/aarch64.c
2256 (aarch64_vectorize_preferred_vector_alignment): Change return type to
2258 (aarch64_simd_vector_alignment_reachable): Adapt to preferred vector
2259 alignment being a poly int.
2260 * doc/tm.texi (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Change
2261 return type to poly_uint64.
2262 * target.def (default_preferred_vector_alignment): Likewise.
2263 * targhooks.c (default_preferred_vector_alignment): Likewise.
2264 * targhooks.h (default_preferred_vector_alignment): Likewise.
2265 * tree-vect-data-refs.c (vect_calculate_target_alignment): Likewise.
2266 (vect_compute_data_ref_alignment): Adapt to vector alignment being a
2268 (vect_update_misalignment_for_peel): Likewise.
2269 (vect_enhance_data_refs_alignment): Likewise.
2270 (vect_find_same_alignment_drs): Likewise.
2271 (vect_duplicate_ssa_name_ptr_info): Likewise.
2272 (vect_setup_realignment): Likewise.
2273 (vect_can_force_dr_alignment_p): Change alignment parameter type to
2275 * tree-vect-loop-manip.c (get_misalign_in_elems): Learn to construct a
2276 mask with a compile time variable vector alignment.
2277 (vect_gen_prolog_loop_niters): Adapt to vector alignment being a poly
2279 (vect_do_peeling): Exit early if vector alignment is not constant.
2280 * tree-vect-stmts.c (ensure_base_align): Adapt to vector alignment being
2282 (vectorizable_store): Likewise.
2283 (vectorizable_load): Likweise.
2284 * tree-vectorizer.h (struct dr_vec_info): Make target_alignment field a
2286 (vect_known_alignment_in_bytes): Adapt to vector alignment being a
2288 (vect_can_force_dr_alignment_p): Change alignment parameter type to
2291 2018-11-13 Richard Biener <rguenther@suse.de>
2293 PR tree-optimization/87962
2294 * tree-vect-loop.c (vect_is_simple_reduction): More reliably
2295 detect outer reduction for disqualifying in-loop uses.
2297 2018-11-13 Richard Biener <rguenther@suse.de>
2299 PR tree-optimization/87967
2300 * tree-vect-loop.c (vect_transform_loop): Also copy PHIs
2301 for constants for the scalar loop.
2303 2018-11-13 Alan Modra <amodra@gmail.com>
2305 * config/rs6000/linux64.h (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Exclude
2306 integer constants when -mcmodel=medium.
2308 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
2310 * config/arc/arc.h (FUNCTION_PROFILER): Redefine to empty.
2311 * config/arc/elf.h (PROFILE_HOOK): Define.
2312 * config/arc/linux.h (PROFILE_HOOK): Likewise.
2314 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
2316 * config/arc/arc.c (hwloop_optimize): Bailout when detecting a
2317 jump table data in the text section.
2319 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
2321 * config/arc/arc.c (arc_eh_return_address_location): Repurpose it
2322 to fit the eh_return pattern.
2323 * config/arc/arc.md (eh_return): Define.
2324 (VUNSPEC_ARC_EH_RETURN): Likewise.
2325 * config/arc/arc-protos.h (arc_eh_return_address_location): Match
2327 * config/arc/arc.h (EH_RETURN_HANDLER_RTX): Remove it.
2330 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
2332 * common/config/arc/arc-common.c (arc_option_optimization_table):
2333 Millicode optimization is default on for size optimizations.
2334 * config/arc/arc-protos.h (arc_check_multi): New function.
2335 * config/arc/arc.c (RTX_OK_FOR_OFFSET_P): Rearange.
2336 (ENTER_LEAVE_START_REG): Define.
2337 (ENTER_LEAVE_END_REG): Likewise.
2338 (arc_override_options): Disable millicode when long calls option
2340 (arc_frame_info): Change it from int to bool.
2341 (arc_compute_frame_size): Clean up.
2342 (arc_save_restore): Remove.
2343 (frame_save_reg): New function.
2344 (frame_restore_reg): Likewise.
2345 (arc_enter_leave_p): Likewise.
2346 (arc_save_callee_saves): Likewise.
2347 (arc_restore_callee_saves): Likewise.
2348 (arc_save_callee_enter): Likewise.
2349 (arc_restore_callee_leave): Likewise.
2350 (arc_save_callee_milli): Likewise.
2351 (arc_restore_callee_milli): Likewise.
2352 (arc_expand_prologue): Reimplement to emit enter/leave
2354 (arc_expand_epilogue): Likewise.
2355 (arc_check_multi): New function.
2356 * config/arc/arc.md (push_multi_fp): New pattern.
2357 (push_multi_fp_blink): Likewise.
2358 (pop_multi_fp): Likewise.
2359 (pop_multi_fp_blink): Likewise.
2360 (pop_multi_fp_ret): Likewise.
2361 (pop_multi_fp_blink_ret): Likewise.
2362 * config/arc/arc.opt (mmillicode): Update option.
2363 (mcode-density-frame): New option.
2364 * config/arc/predicates.md (push_multi_operand): New predicate.
2365 (pop_multi_operand): Likewise.
2366 * doc/invoke.texi (ARC): Update ARC options information.
2369 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
2371 * config/arc/arc-protos.h (gen_operands_ldd_std): Add.
2372 * config/arc/arc.c (operands_ok_ldd_std): New function.
2373 (mem_ok_for_ldd_std): Likewise.
2374 (gen_operands_ldd_std): Likewise.
2375 * config/arc/arc.md: Add peephole2 rules for std/ldd.
2377 2018-11-13 Eric Botcazou <ebotcazou@adacore.com>
2379 * toplev.c (output_stack_usage): Turn test on flag_stack_usage into
2380 test on stack_usage_file.
2381 (lang_dependent_init): Do not open the .su file if generating LTO.
2383 2018-11-13 Jakub Jelinek <jakub@redhat.com>
2385 PR rtl-optimization/87918
2386 * simplify-rtx.c (simplify_merge_mask): For COMPARISON_P, use
2387 simplify_gen_relational rather than simplify_gen_binary.
2389 2018-11-13 Richard Biener <rguenther@suse.de>
2391 * tree-ssanames.h (set_range_info): Use value_range_base.
2392 (get_range_info): Likewise.
2393 * tree-ssanames.c (set_range_info): Likewise.
2394 (get_range_info): Likewise.
2395 * tree-vrp.c (value_range_base::union_helper): Split
2396 out common parts of value_range[_base]::union_.
2397 (value_range_base::union_): Update.
2398 (value_range::union_): Likewise.
2399 (determine_value_range_1): Use value_range_base.
2400 (determine_value_range): Likewise.
2401 * tree-vrp.h (value_range_base::union_helper): Move ...
2402 (value_range::union_helper): ... from here.
2404 2018-11-13 Alan Modra <amodra@gmail.com>
2406 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Negate
2408 (rs6000_secondary_reload_gpr): Don't call find_replacement.
2410 2018-11-13 Sandra Loosemore <sandra@codesourcery.com>
2413 * doc/invoke.texi (Optimize Options): Clarify that the
2414 l1-cache-line-size, l1-cache-size, and l2-cache-size parameters
2415 apply to data cache size.
2417 2018-11-13 Alan Modra <amodra@gmail.com>
2419 * config/rs6000/predicates.md (logical_const_operand),
2420 (logical_operand): Correct comment.
2421 * config/rs6000/rs6000.c (rs6000_emit_prologue): Comment fix.
2423 2018-11-13 Alan Modra <amodra@gmail.com>
2425 * config/rs6000/rs6000.md (addsi3_high): Prefix with '*'.
2427 2018-11-13 Alan Modra <amodra@gmail.com>
2429 * gcc/doc/md.texi (Machine Constraints): Correct rs6000 h constraint
2431 * config/rs6000/rs6000.md (movsi_internal1): Delete MT%0 case
2432 covered by alternative.
2433 (movcc_internal1): Ignore h for register preference.
2434 (mov<mode>_hardfloat64): Likewise.
2435 (mov<mode>_softfloat): Ignore c, l, h for register preference.
2437 2018-11-12 Sandra Loosemore <sandra@codesourcery.com>
2439 PR preprocessor/47823
2440 * doc/cpp.texi (Alternatives to Wrapper #ifndef): Move #pragma once
2443 * doc/extend.texi (Pragmas): Note additional pragmas documented
2446 2018-11-13 Kugan Vivekanandarajah <kuganv@linaro.org>
2450 * tree-scalar-evolution.c (expression_expensive_p): Make
2451 BUILTIN POPCOUNT as expensive when backend does not define it.
2453 2018-11-12 Fredrik Noring <noring@nocrew.org>
2455 * config.gcc: Update with-llsc defaults for MIPS r5900.
2457 2018-11-12 Martin Liska <mliska@suse.cz>
2459 PR gcov-profile/87442
2460 * common.opt: Add -fprofile-filter-files and -fprofile-exclude-files
2462 * doc/invoke.texi: Document them.
2463 * tree-profile.c (parse_profile_filter): New.
2464 (parse_profile_file_filtering): Likewise.
2465 (release_profile_file_filtering): Likewise.
2466 (include_source_file_for_profile): Likewise.
2467 (tree_profiling): Filter source files based on the
2468 newly added options.
2470 2018-11-12 Martin Liska <mliska@suse.cz>
2473 * doc/extend.texi: Add missing values for __builtin_cpu_is and
2474 __builtin_cpu_supports for x86 target.
2476 2018-11-12 Sandra Loosemore <sandra@codesourcery.com>
2479 * doc/rtl.texi (Constants): Clarify that mode of "high" doesn't
2481 (Arithmetic): Likewise for "lo_sum".
2483 2018-11-12 Renlin Li <renlin.li@arm.com>
2486 * dse.c (get_stored_val): Add check for compile-time constantness
2489 2018-11-12 Sudakshina Das <sudi.das@arm.com>
2491 * config/arm/arm-cpus.in (armv8_5, sb, predres): New features.
2492 (ARMv8_5a): New fgroup.
2493 (armv8.5-a): New arch.
2494 (armv8-a, armv8.1-a, armv8.2-a, armv8.3-a, armv8.4-a): New options
2496 * config/arm/arm-tables.opt: Regenerate.
2497 * config/arm/t-aprofile: Add matching rules for -march=armv8.5-a
2498 * config/arm/t-arm-elf (all_v8_archs): Add armv8.5-a.
2499 * config/arm/t-multilib (v8_5_a_simd_variants): New variable.
2500 Add matching rules for -march=armv8.5-a and extensions.
2501 * doc/invoke.texi (ARM options): Document -march=armv8.5-a.
2502 Add sb and predres to all armv8-a except armv8.5-a.
2504 2018-11-12 Richard Biener <rguenther@suse.de>
2506 * tree-vrp.h (value_range[_base]::set): Make public. Provide
2507 overload for single value.
2508 (value_range[_base]::set_nonnull): New.
2509 (value_range[_base]::set_null): Likewise.
2510 (value_range): Document bitmap copying behavior, mark
2511 copy constructor and assignment operator deleted.
2512 (value_range::move): New.
2513 (value_range::set_and_canonicalize): Default bitmap to zero.
2514 (set_value_range_to_nonnull): Remove.
2515 (set_value_range_to_null): Likewise.
2516 (set_value_range): Likewise.
2517 (set_value_range_to_value): Likewise.
2518 (extract_range_from_unary_expr): Work on value_range_base.
2519 (extract_range_from_binary_expr_1): Likewise. Rename to...
2520 (extract_range_from_binary_expr): ... this.
2521 * tree-vrp.c (value_range::update): Clear equiv bitmap
2523 (value_range::move): New, move equiv bitmap.
2524 (value_range_base::set_undefined): Avoid assignment.
2525 (value_range::set_undefined): Likewise.
2526 (value_range_base::set_varying): Likewise.
2527 (value_range::set_varying): Likewise.
2528 (set_value_range): Remove.
2529 (value_range_base::set): New overload for value.
2530 (value_range::set): Likewise.
2531 (set_value_range_to_nonnull): Remove.
2532 (value_range_base::set_nonnull): New.
2533 (value_range::set_nonnull): Likewise.
2534 (set_value_range_to_null): Remove.
2535 (value_range_base::set_null): New.
2536 (value_range::set_null): Likewise.
2537 (range_is_null): Work on value_range_base.
2538 (range_is_nonnull): Likewise.
2539 (ranges_from_anti_range): Likewise.
2540 (extract_range_into_wide_ints): Likewise.
2541 (extract_range_from_multiplicative_op): Likewise.
2542 (extract_range_from_binary_expr): Likewise. Update for API changes.
2543 (extract_range_from_unary_expr): Likewise. Remove OBJ_TYPE_REF
2545 (value_range::intersect_helper): Avoid copy and assignment.
2546 (value_range::union_helper): Likewise.
2547 (determine_value_range_1): Adjust.
2548 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
2549 Avoid assignment by using move.
2550 (evrp_range_analyzer::record_ranges_from_stmt): Avoid assignment.
2551 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
2553 * tree-ssanames.c (get_range_info): Likewise.
2554 * vr-values.h (vr_values::get_vr_for_comparison): Adjust API.
2555 * vr-values.c (vr_values::get_value_range): Adjust.
2556 (vr_values::update_value_range): Likewise.
2557 (symbolic_range_based_on_p): Work on value_range_base.
2558 (vr_values::extract_range_from_binary_expr): Use value_range_base.
2559 (vr_values::extract_range_from_unary_expr): Likewise.
2560 (vr_values::extract_range_from_cond_expr): Avoid assignment.
2561 (vr_values::extract_range_from_comparison): Adjust.
2562 (vr_values::check_for_binary_op_overflow): Use value_range_base.
2563 (vr_values::extract_range_basic): Adjust.
2564 (vr_values::adjust_range_with_scev): Likewise.
2565 (vr_values::vrp_visit_assignment_or_call): Likewise.
2566 (vr_values::get_vr_for_comparison): Change API to avoid
2567 assignment and copy construction.
2568 (vr_values::compare_name_with_value): Adjust accordingly.
2569 (vr_values::compare_names): Likewise.
2570 (vr_values::extract_range_from_phi_node): Avoid assignment and
2571 bogus in-place modify of equiv bitmap.
2572 (vr_values::simplify_bit_ops_using_ranges): Use value_range_base.
2573 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Adjust
2574 for extract_range_from_unary_expr API change.
2575 * ipa-cp.c (ipa_vr_operation_and_type_effects): Likewise.
2577 2018-11-12 Eric Botcazou <ebotcazou@adacore.com>
2579 * config/mcore/mcore.h (WORD_REGISTER_OPERATIONS): Remove duplicate.
2581 2018-11-12 Richard Biener <rguenther@suse.de>
2583 * tree-vrp.h (value_range_base::symbolic_p,
2584 value_range_base::constant_p, value_range_base::zero_p,
2585 value_range_base::singleton_p): Move from value_range.
2586 (value_range::dump): Add.
2587 * gimple-ssa-evrp-analyze.c
2588 (evrp_range_analyzer::record_ranges_from_phis): Use set_varying.
2589 * ipa-cp.c (ipcp_vr_lattice::print): Use dump_value_range.
2590 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
2592 * tree-vrp.c (value_range::symbolic_p): Move to value_range_base.
2593 (value_range::constant_p): Likewise.
2594 (value_range::singleton_p): Likewise.
2595 (value_range_base::dump): Add.
2596 (set_value_range_to_undefined): Remove.
2597 (set_value_range_to_varying): Likewise.
2598 (range_int_cst_p): Take value_range_base argument.
2599 (range_int_cst_singleton_p): Likewise.
2600 (value_range_constant_singleton): Likewise.
2601 (vrp_set_zero_nonzero_bits): Likewise.
2602 (extract_range_from_multiplicative_op): Use set_varying.
2603 (extract_range_from_binary_expr_1): Likewise. Use set_undefined.
2604 (extract_range_from_unary_expr): Likewise.
2605 (dump_value_range_base): Change to overload of dump_value_range.
2606 (vrp_prop::vrp_initialize): Use set_varying and set_undefined.
2607 (vrp_prop::visit_stmt): Likewise.
2608 (value_range::intersect_helper): Likewise.
2609 (value_range::union_helper): Likewise.
2610 (determine_value_range_1): Likewise.
2612 2018-11-12 Richard Biener <rguenther@suse.de>
2614 * tree-vrp.c (set_value_range_to_nonnull): Clear equiv.
2615 (set_value_range_to_null): Likewise.
2616 * vr-values.c (vr_values::extract_range_from_comparison):
2617 Clear equiv for constant singleton ranges.
2619 2018-11-12 Wei Xiao <wei3.xiao@intel.com>
2621 * config/i386/sse.md: Combine VFIXUPIMM* patterns
2622 (<avx512>_fixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
2623 (<avx512>_fixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
2624 (<avx512>_fixupimm<mode>_mask<round_saeonly_name>): Remove.
2625 (avx512f_sfixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
2626 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
2627 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Remove.
2629 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
2632 * doc/extend.texi (Common Type Attributes): For the align type
2633 attribute, copy language about decreasing alignment from the
2634 corresponding variable attribute.
2636 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
2638 * config/csky/csky.md (*fpuv2_nmulsf3_1, *fpuv2_nmuldf3_1): Handle
2641 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
2644 * doc/invoke.texi (C++ Dialect Options): Add warning about mixing
2645 -frtti and -fno-rtti code.
2647 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
2650 * doc/extend.texi (Other Builtins): Document probability associated
2651 with __builtin_expect.
2653 2018-11-11 Uros Bizjak <ubizjak@gmail.com>
2656 * config/i386/i386.h (STACK_BOUNDARY): Use TARGET_64BIT_MS_ABI
2657 instead of (TARGET_64BIT && ix86_abi == MS_ABI).
2658 * config/i386/darwin.h (STACK_BOUNDARY): Ditto.
2659 * config/i386/cygming.h (STACK_BOUNDARY): Remove.
2661 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
2663 * config/csky/csky-linux-elf.h (CC1_SPEC): Support -profile.
2665 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
2667 * config/csky/csky.h (ASM_PREFERRED_EH_DATA_FORMAT): Define.
2669 2018-11-11 Richard Biener <rguenther@suse.de>
2671 * tree-vrp.h (class value_range_base): New base class for
2672 value_range containing all but the m_equiv member.
2673 (dump_value_range_base): Add.
2674 (range_includes_zero_p): Work on value_range_base.
2675 * tree-vrp.c (value_range_base::set): Split out base handling
2677 (value_range::set): this.
2678 (value_range::set_equiv): New.
2679 (value_range_base::value_range_base): New constructors.
2680 (value_range_base::check): Split out base handling from...
2681 (value_range::check): this.
2682 (value_range::equal_p): Refactor in terms of
2683 ignore_equivs_equal_p which is now member of the base.
2684 (value_range_base::set_undefined): New.
2685 (value_range_base::set_varying): Likewise.
2686 (value_range_base::dump):Split out base handling from...
2687 (value_range::dump): this.
2688 (value_range_base::set_and_canonicalize): Split out base handling
2690 (value_range::set_and_canonicalize): this.
2691 (value_range_base::union_): New.
2692 * ipa-prop.h (struct ipa_jump_func): Use value_range_base *
2694 * ipa-cp.c (class ipcp_vr_lattice): Use value_range_base
2695 instead of value_range everywhere.
2696 (ipcp_vr_lattice::print): Use dump_value_range_base.
2697 (ipcp_vr_lattice::meet_with): Adjust.
2698 (ipcp_vr_lattice::meet_with_1): Likewise.
2699 (ipa_vr_operation_and_type_effects): Likewise.
2700 (propagate_vr_across_jump_function): Likewise.
2701 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Likewise.
2702 (ipa_get_value_range): Likewise.
2703 (ipa_set_jfunc_vr): Likewise.
2704 (ipa_compute_jump_functions_for_edge): Likewise.
2706 2018-11-10 Sandra Loosemore <sandra@codesourcery.com>
2709 * doc/invoke.texi (Optimize Options): Add @opindex entries
2710 for the positive forms of -fno-xxx and -mno-xxx options
2711 that were lacking them.
2713 2018-11-10 Segher Boessenkool <segher@kernel.crashing.org>
2715 * combine.c (make_more_copies): Only make an intermediate copy if the
2716 dest of a move is a pseudo.
2718 2018-11-09 Maya Rashish <coypu@sdf.org>
2721 * config/netbsd-elf.h (NETBSD_STARTFILE_SPEC): Use crtbeginS.o for PIE.
2722 (NETBSD_ENDFILE_SPEC): Use crtendS.o for PIE.
2724 2018-11-09 Sandra Loosemore <sandra@codesourcery.com>
2728 * doc/invoke.texi (Optimize Options): Clarify default behavior
2729 for -fno-toplevel-reorder, -fno-defer-pop, and -fno-branch-count-reg.
2731 2018-11-09 Bernd Edlinger <bernd.edlinger@hotmail.de>
2733 PR tree-optimization/87940
2734 * expr.c (string_constant): Don't strip NOPS in subexpressions.
2735 Fold PLUS_EXPR correctly.
2737 2018-11-09 Ilya Leoshkevich <iii@linux.ibm.com>
2740 * config/s390/s390.c (s390_safe_relative_long_p): New function.
2741 (annotate_constant_pool_refs): Skip insns which support
2742 relative addressing.
2743 (annotate_constant_pool_refs_1): New helper function.
2744 (find_constant_pool_ref): Skip insns which support relative
2746 (find_constant_pool_ref_1): New helper function.
2747 (replace_constant_pool_ref): Skip insns which support
2748 relative addressing.
2749 (replace_constant_pool_ref_1): New helper function.
2750 (s390_mainpool_start): Adapt to the new signature.
2751 (s390_mainpool_finish): Likewise.
2752 (s390_chunkify_start): Likewise.
2753 (s390_chunkify_finish): Likewise.
2754 (pass_s390_early_mach::execute): Likewise.
2755 (s390_prologue_plus_offset): Likewise.
2756 (s390_emit_prologue): Likewise.
2757 (s390_emit_epilogue): Likewise.
2759 2018-11-09 Jakub Jelinek <jakub@redhat.com>
2761 * gimplify.c (gimplify_scan_omp_clauses): Call sorry_at for valid
2762 but unsupported lastprivate with conditional modifier.
2764 2018-11-09 Jeff Law <law@redhat.com>
2766 * config/mips/mips.c (mips_loongson_ext2_prefetch_cookie): Handle
2767 unused argument better. Add gcc_unreachable to silence warning.
2769 2018-11-09 Martin Sebor <msebor@redhat.com>
2772 * attribs.c (has_attribute): New helper function.
2773 (decls_mismatched_attributes, maybe_diag_alias_attributes): Same.
2774 * attribs.h (decls_mismatched_attributes): Declare.
2775 * cgraphunit.c (handle_alias_pairs): Call maybe_diag_alias_attributes.
2776 (maybe_diag_incompatible_alias): Use OPT_Wattribute_alias_.
2777 * common.opt (-Wattribute-alias): Take an argument.
2778 (-Wno-attribute-alias): New option.
2779 * doc/extend.texi (Common Function Attributes): Document copy.
2780 (Common Variable Attributes): Same.
2781 * doc/invoke.texi (-Wmissing-attributes): Document enhancement.
2782 (-Wattribute-alias): Document new option argument.
2784 2018-11-09 Richard Earnshaw <rearnsha@arm.com>
2786 * config/arm/parsecpu.awk (/alias/): Tighten invisible alias
2787 matching criteria. Remove unused array initializer.
2789 2018-11-09 Bill Schmidt <wschmidt@linux.ibm.com>
2790 Jinsong Ji <jji@us.ibm.com>
2792 * config/rs6000/xmmintrin.h (_mm_cvtss_si32): Fix incorrect
2793 constraints by introducing a new temporary.
2794 (_mm_cvtss_si64): Likewise.
2796 2018-11-09 Martin Liska <mliska@suse.cz>
2798 * common.opt: Add -fipa-stack-alignment flag.
2799 * doc/invoke.texi: Document it.
2800 * final.c (rest_of_clean_state): Guard stack
2801 shrinking with flag.
2803 2018-11-09 Martin Liska <mliska@suse.cz>
2805 * cgraph.h (ipa_discover_readonly_nonaddressable_vars): Rename
2807 (ipa_discover_variable_flags): ... this.
2808 * common.opt: Come up with new flag -fipa-reference-addressable.
2809 * doc/invoke.texi: Document it.
2810 * ipa-reference.c (propagate): Call the renamed fn.
2811 * ipa-visibility.c (whole_program_function_and_variable_visibility):
2813 * ipa.c (ipa_discover_readonly_nonaddressable_vars): Renamed to
2815 (ipa_discover_variable_flags): ... this. Discover
2816 non-addressable variables only with the newly added flag.
2817 * opts.c: Enable the newly added flag with -O1 and higher
2820 2018-11-09 David Malcolm <dmalcolm@redhat.com>
2822 * json.cc (selftest::test_writing_literals): Fix comment.
2824 2018-11-09 Martin Liska <mliska@suse.cz>
2826 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not copy
2827 string to a stack buffer.
2828 (aarch64_parse_cpu): Likewise.
2829 (aarch64_parse_tune): Likewise.
2831 2018-11-09 Richard Biener <rguenther@suse.de>
2833 PR tree-optimization/87953
2834 * tree-vect-loop.c (vectorizable_reduction): For analysis
2835 always pass ops[0] to vectorizable_condition.
2837 2018-11-09 Stafford Horne <shorne@gmail.com>
2838 Richard Henderson <rth@twiddle.net>
2839 Joel Sherrill <joel@rtems.org>
2841 * common/config/or1k/or1k-common.c: New file.
2842 * config/or1k/*: New.
2843 * config.gcc (or1k*-*-*): New.
2844 * configure.ac (or1k*-*-*): New test for openrisc tls.
2845 * configure: Regenerated.
2846 * doc/install.texi: Document OpenRISC triplets.
2847 * doc/invoke.texi: Document OpenRISC arguments.
2848 * doc/md.texi: Document OpenRISC.
2850 2018-11-09 Richard Earnshaw <rearnsha@arm.com>
2852 * config/arm/arm-cpus.in (arm7tdmi): Add an alias for arm7tdmi-s.
2853 (arm7tdmi-s): Delete CPU.
2854 (arm710t): Add aliases for arm720t and arm740t.
2855 (arm720t, arm740t): Delete CPUs.
2856 (arm920t): Add aliases for arm920, arm922t and arm940t.
2857 (arm920, arm922t, arm940t): Delete CPUs.
2858 (arm10tdmi): Add alias for arm1020t.
2859 (arm1020t): Delete CPU.
2860 (arm9e): Add aliases for arm946e-s, arm966e-s and arm968e-s.
2861 (arm946e-s, arm966e-s, arm968e-s): Delete CPUs.
2862 (arm10e): Add aliases for arm1020e and arm1022e.
2863 (arm1020e, arm1022e): Delete CPU.
2864 * config/arm/arm.md (generic_sched): Remove entries that are now
2866 (generic_vfp): Likewise.
2867 * config/arm/arm1020e.md: Simplify tuning selection based on alias
2869 * config/arm/arm-tune.md: Regenerated.
2870 * config/arm/arm-tables.opt: Regenerated.
2872 2018-11-09 Richard Biener <rguenther@suse.de>
2874 PR tree-optimization/87621
2875 * tree-vect-loop.c (vectorizable_reduction): Handle reduction
2876 op with only phi inputs.
2877 * tree-ssa-loop-ch.c: Include tree-ssa-sccvn.h.
2878 (ch_base::copy_headers): Run CSE on copied loop headers.
2879 (pass_ch_vect::process_loop_p): Simplify.
2881 2018-11-09 Alexandre Oliva <oliva@adacore.com>
2883 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Adjust
2884 for non-w64 x86_64 biarch.
2886 2018-11-09 Alexandre Oliva <aoliva@redhat.com>
2888 PR rtl-optimization/86438
2889 * compare-elim.c (try_eliminate_compare): Use SET_SRC instead
2890 of in_b for the compare if in_b is SET_DEST.
2893 * config/i386/i386.c (ix86_const_not_ok_for_debug_p): Reject
2894 non-toplevel UNSPEC.
2896 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
2898 * tree-vrp.c (value_range::check): Do not access internals
2900 (value_range::singleton_p): Same.
2901 (value_range::type): Same.
2902 (vrp_finalize): Use value_range API.
2904 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
2906 * tree-vrp.c (may_contain_p): Do not access m_min/m_max directly.
2908 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
2910 * vr-values.c (vr_values::get_value_range): Use value_range API
2911 instead of piecing together ranges.
2912 (vr_values::update_value_range): Same.
2914 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
2916 * gimple-fold.c (size_must_be_zero_p): Use value_range API instead
2917 of performing ad-hoc calculations.
2918 * tree-ssanames.c (set_range_info): New overloaded function
2919 accepting value_range &.
2920 (get_range_info): Same.
2921 * tree-ssanames.h (set_range_info_raw): Remove.
2922 (set_range_info): New prototype.
2923 (get_range_info): Same.
2924 * tree-vrp.h (value_range::null_p): Rename to zero_p.
2925 * tree-vrp.c (value_range::null_p): Same.
2927 2018-11-09 Jan Hubicka <jh@suse.cz>
2929 * tree.c (fld_type_variant_equal_p): Test user align flag.
2930 (flt_type_variant): Copy user align flag.
2931 (fld_incomplete_type_of): Clear it.
2933 2018-11-09 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2935 * config/arm/neon.md (div<mode>3): New pattern.
2937 2018-11-08 Andi Kleen <ak@linux.intel.com>
2939 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PTWRITE_SET): New.
2940 (OPTION_MASK_ISA_PTWRITE_UNSET): New.
2941 (ix86_handle_option): Handle OPT_mptwrite.
2942 * config/i386/cpuid.h (bit_PTWRITE): Add.
2943 * config/i386/driver-i386.c (host_detect_local_cpu): Detect ptwrite.
2944 * config/i386/i386-builtin.def (BDESC): Add ptwrite32/64.
2945 * config/i386/i386-c.c (ix86_target_macros_internal): Define __PTWRITE__.
2946 * config/i386/i386.c (ix86_target_string): Handle ptwrite.
2947 (ix86_option_override_internal): Handle PTA_PTWRITE.
2948 (ix86_valid_target_attribute_inner_p): Define ptwrite.
2949 (def_builtin2): Force UINT64 to be 64bit only.
2950 * config/i386/i386.h (TARGET_PTWRITE): Add.
2951 (TARGET_PTWRITE_P): Add.
2953 * config/i386/i386.md: Define ptwrite.
2954 * config/i386/i386.opt: Add -mptwrite.
2955 * config/i386/immintrin.h (_ptwrite64): Add.
2957 * doc/extend.texi: Document __builtin_ia32_ptwrite*.
2958 * doc/invoke.texi: Document -mptwrite.
2960 2018-11-08 Peter Bergner <bergner@linux.ibm.com>
2962 PR rtl-optimization/87600
2963 * cfgexpand.c (expand_asm_stmt): Catch illegal asm constraint usage.
2964 * lra-constraints.c (process_alt_operands): Skip illegal hard
2965 register usage. Prefer reloading non hard register operands.
2967 2018-11-08 Sandra Loosemore <sandra@codesourcery.com>
2970 * doc/invoke.texi (Optimize Options): Clarify default behavior
2971 for -fno-sched-interblock and -fno-sched-spec.
2973 2018-11-08 Roman Geissler <roman.geissler@amadeus.com>
2975 * collect2.c (linker_select): Add USE_LLD_LD.
2976 (ld_suffixes): Add ld.lld.
2977 (main): Handle -fuse-ld=lld.
2978 * common.opt (-fuse-ld=lld): New option.
2979 * doc/invoke.texi (-fuse-ld=lld): Document.
2980 * opts.c (common_handle_option): Handle OPT_fuse_ld_lld.
2982 2018-11-08 Paul Koning <ni1d@arrl.net>
2984 * config/pdp11/constraints.md: Add "Z" series constraints for use
2985 with pre-dec and post-inc addressing.
2986 * config/pdp11/pdp11-protos.m (expand_block_move): Delete.
2987 (pdp11_expand_operands): Add int argument (word count).
2988 (pdp11_sp_frame_offset): Delete.
2989 (pdp11_cmp_length): New function.
2990 (pushpop_regeq): New function.
2991 * config/pdp11/pdp11.c (TARGET_STACK_PROTECT_RUNTIME_ENABLED_P):
2993 (pdp11_expand_prologue, pdp11_expand_epilogue): Rewrite for new
2995 (pdp11_initial_elimination_offset): Ditto.
2996 (pdp11_expand_operands): Add word count argument. Bugfixes.
2997 (output_move_multiple): Change how pointer adjustment is done.
2998 (pdp11_gen_int_label): Correct format.
2999 (output_ascii): Ditto.
3000 (pdp11_asm_output_var): Add code for DEC assembler case.
3001 (pdp11_asm_print_operand): Bugfix for CONST_DOUBLE holding integer
3003 (legitimate_const_double_p): Ditto.
3004 (pdp11_register_move_cost): Adjust for new register classes.
3005 (pdp11_regno_reg_class): Ditto.
3006 (expand_block_move): Delete.
3007 (pushpop_regeq): New function.
3008 (pdp11_legitimate_address_p): Bugfix in check for constant
3010 (pdp11_sp_frame_offset): Delete.
3011 (pdp11_reg_save_size): New helper function for new frame layout.
3012 (output_addr_const_pdp11): Remove CONST_DOUBLE case.
3013 (pdp11_expand_shift): Bugfix in check for constant shift count.
3014 (pdp11_shift_length): Ditto.
3015 (pdp11_assemble_shift): Copy input to pdp11_expand_operands.
3016 (pdp11_cmp_length): New function.
3017 * config/pdp11/pdp11.h (TARGET_CPU_CPP_BUILTINS): Add macros for
3018 some compile options.
3019 (FIXED_REGISTERS): Remove HARD_FRAME_POINTER_REGNUM.
3020 (CALL_USED_REGISTERS): Ditto.
3021 (ELIMINABLE_REGS): Ditto.
3022 (REGISTER_NAMES): Ditto.
3023 (reg_class): Add classes NOTR0_REG through NOTSP_REG for use by Z
3025 (REG_CLASS_NAMES): Ditto.
3026 (REG_CLASS_CONTENTS): Ditto. Also remove
3027 HARD_FRAME_POINTER_REGNUM.
3028 (CPU_REG_CLASS): New macro.
3029 (CLASS_MAX_NREGS): Adjust for new register classes.
3030 (FUNCTION_PROFILER): Make no-op.
3031 (may_call_alloca): Remove unused declaration.
3032 (ASM_OUTPUT_ALIGN): Add workaround for PR87795.
3033 (ASM_OUTPUT_SKIP): Fix format.
3034 * config/pdp11/pdp11.md (unspecv): Add UNSPECV_MOVMEM.
3035 (HARD_FRAME_POINTER_REGNUM): Remove.
3037 (*rts): Rename. Remove epilogue related checks.
3038 (cmpsi, cmpdi): New insn.
3039 (cbranch<mode>4): Change to apply to SI and DI modes as well.
3040 (mov<mode>): Change constraints to enforce that push/pop
3041 destination cannot use the same register as source.
3042 (*mov<mode><cc_cc>): Ditto.
3043 (movmemhi, movmemhi1, movmemhi_nocc): Change to expand block move
3044 at assembly output rather than as RTL expander.
3045 (zero_extendqihi2): Bugfix in check for same registers.
3046 (adddi3_nocc): Bugfix in check for constant operand.
3047 (addsi3_nocc): Ditto.
3048 (subdi3_nocc): Ditto.
3049 (subsi3_nocc): Ditto.
3050 (negdi2_nocc): Copy input to pdp11_expand_operands.
3051 (negsi2_nocc): Ditto.
3052 (bswap2_nocc): Ditto.
3053 * config/pdp11/pdp11.opt (mlra): Fix documentation.
3054 * config/pdp11/t-pdp11: Use -Os.
3056 2018-11-08 Richard Earnshaw <rearnsha@arm.com>
3058 * config/arm/parsecpu.awk (/alias/): New parsing rule.
3059 (/begin cpu/): Check that the cpu name hasn't been previously defined.
3060 (gen_comm_data): Print out CPU alias tables.
3061 (check_cpu): Match aliases when checking the CPU name.
3062 * config/arm/arm-protos.h (cpu_alias): New structure.
3063 (cpu_option): Add entry for aliases.
3064 * config/arm/arm-cpus.in (strongarm): Add aliases for strongarm110
3065 strongarm1100 and strongarm1110.
3066 (strongarm110, strongarm1100, strongarm1110): Delete CPU entries.
3067 (config/arm/arm-generic.md): Remove redundant references to
3068 strongarm110, strongarm1100 and strongarm1110.
3069 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
3070 Scan aliases for additional hints.
3071 (arm_parse_cpu_option_name): Also match a cpu name against the list
3073 * config/arm/arm-tables.opt: Regenerated.
3074 * config/arm/arm-tune.md: Regenerated.
3076 2018-11-08 Jakub Jelinek <jakub@redhat.com>
3078 * builtin-types.def (BT_FN_VOID_BOOL, BT_FN_VOID_SIZE_SIZE_PTR,
3079 BT_FN_UINT_UINT_PTR_PTR, BT_FN_UINT_OMPFN_PTR_UINT_UINT,
3080 BT_FN_BOOL_UINT_LONGPTR_LONG_LONG_LONGPTR_LONGPTR_PTR_PTR,
3081 BT_FN_BOOL_UINT_ULLPTR_LONG_ULL_ULLPTR_ULLPTR_PTR_PTR,
3082 BT_FN_BOOL_LONG_LONG_LONG_LONG_LONG_LONGPTR_LONGPTR_PTR_PTR,
3083 BT_FN_BOOL_BOOL_ULL_ULL_ULL_LONG_ULL_ULLPTR_ULLPTR_PTR_PTR): New.
3084 * gengtype.c (open_base_files): Add omp-general.h.
3085 * gimple.c (gimple_build_omp_critical):
3086 (gimple_build_omp_taskgroup): Add CLAUSES argument. Call
3087 gimple_omp_taskgroup_set_clauses.
3088 (gimple_build_omp_atomic_load): Add mo argument, call
3089 gimple_omp_atomic_set_memory_order.
3090 (gimple_build_omp_atomic_store): Likewise.
3091 (gimple_copy): Adjust handling of GIMPLE_OMP_TASKGROUP.
3092 * gimple.def (GIMPLE_OMP_TASKGROUP): Use GSS_OMP_SINGLE_LAYOUT
3094 (GIMPLE_OMP_TEAMS): Use GSS_OMP_PARALLEL_LAYOUT instead
3095 of GSS_OMP_SINGLE_LAYOUT, adjust comments.
3096 * gimple.h (enum gf_mask): Add GF_OMP_TEAMS_HOST, GF_OMP_TASK_TASKWAIT
3097 and GF_OMP_ATOMIC_MEMORY_ORDER. Remove GF_OMP_ATOMIC_SEQ_CST, use
3098 different value for GF_OMP_ATOMIC_NEED_VALUE.
3099 (struct gimple_statement_omp_taskreg): Add GIMPLE_OMP_TEAMS to
3101 (struct gimple_statement_omp_single_layout): And remove here.
3102 (struct gomp_teams): Inherit from gimple_statement_omp_taskreg rather
3103 than gimple_statement_omp_single_layout.
3104 (is_a_helper <gimple_statement_omp_taskreg *>::test): Allow
3106 (is_a_helper <const gimple_statement_omp_taskreg *>::test): Likewise.
3107 (gimple_omp_subcode): Formatting fix.
3108 (gimple_omp_teams_child_fn, gimple_omp_teams_child_fn_ptr,
3109 gimple_omp_teams_set_child_fn, gimple_omp_teams_data_arg,
3110 gimple_omp_teams_data_arg_ptr, gimple_omp_teams_set_data_arg,
3111 gimple_omp_teams_host, gimple_omp_teams_set_host,
3112 gimple_omp_task_taskwait_p, gimple_omp_task_set_taskwait_p,
3113 gimple_omp_taskgroup_clauses, gimple_omp_taskgroup_clauses_ptr,
3114 gimple_omp_taskgroup_set_clauses): New inline functions.
3115 (gimple_build_omp_atomic_load): Add enum omp_memory_order argument.
3116 (gimple_build_omp_atomic_store): Likewise.
3117 (gimple_omp_atomic_seq_cst_p): Remove.
3118 (gimple_omp_atomic_memory_order): New function.
3119 (gimple_omp_atomic_set_seq_cst): Remove.
3120 (gimple_omp_atomic_set_memory_order): New function.
3121 (gimple_build_omp_taskgroup): Add clauses argument.
3122 * gimple-pretty-print.c (dump_gimple_omp_taskgroup): New function.
3123 (dump_gimple_omp_task): Print taskwait with depend clauses.
3124 (dump_gimple_omp_atomic_load, dump_gimple_omp_atomic_store): Use
3125 dump_omp_atomic_memory_order.
3126 (pp_gimple_stmt_1): Handle GIMPLE_OMP_TASKGROUP.
3127 * gimplify.c (enum gimplify_omp_var_data): Add GOVD_MAP_ALLOC_ONLY,
3128 GOVD_MAP_FROM_ONLY and GOVD_NONTEMPORAL.
3129 (enum omp_region_type): Reserve bits 1 and 2 for auxiliary flags,
3130 renumber values of most of ORT_* enumerators, add ORT_HOST_TEAMS,
3131 ORT_COMBINED_HOST_TEAMS, ORT_TASKGROUP, ORT_TASKLOOP and
3132 ORT_UNTIED_TASKLOOP enumerators.
3133 (enum gimplify_defaultmap_kind): New.
3134 (struct gimplify_omp_ctx): Remove target_map_scalars_firstprivate and
3135 target_map_pointers_as_0len_arrays members, add defaultmap.
3136 (new_omp_context): Initialize defaultmap member.
3137 (gimple_add_tmp_var): Handle ORT_TASKGROUP like ORT_WORKSHARE.
3138 (maybe_fold_stmt): Don't fold even in host teams regions.
3139 (omp_firstprivatize_variable): Handle ORT_TASKGROUP like
3140 ORT_WORKSHARE. Test ctx->defaultmap[GDMK_SCALAR] instead of
3141 ctx->omp_firstprivatize_variable.
3142 (omp_add_variable): Don't add private/firstprivate for VLAs in
3144 (omp_default_clause): Print "taskloop" rather than "task" if
3146 (omp_notice_variable): Handle ORT_TASKGROUP like ORT_WORKSHARE.
3147 Handle new defaultmap clause kinds.
3148 (omp_is_private): Handle ORT_TASKGROUP like ORT_WORKSHARE. Allow simd
3149 iterator to be lastprivate or private. Fix up diagnostics if linear
3150 is used on collapse>1 simd iterator.
3151 (omp_check_private): Handle ORT_TASKGROUP like ORT_WORKSHARE.
3152 (gimplify_omp_depend): New function.
3153 (gimplify_scan_omp_clauses): Add shared clause on parallel for
3154 combined parallel master taskloop{, simd} if taskloop has
3155 firstprivate, lastprivate or reduction clause. Handle
3156 OMP_CLAUSE_REDUCTION_TASK diagnostics. Adjust tests for
3157 ORT_COMBINED_TEAMS. Gimplify depend clauses with iterators. Handle
3158 cancel and simd OMP_CLAUSE_IF_MODIFIERs. Handle
3159 OMP_CLAUSE_NONTEMPORAL. Handle new defaultmap clause kinds. Handle
3160 OMP_CLAUSE_{TASK,IN}_REDUCTION. Diagnose invalid conditional
3162 (gimplify_adjust_omp_clauses_1): Ignore GOVD_NONTEMPORAL. Handle
3163 GOVD_MAP_ALLOC_ONLY and GOVD_MAP_FROM_ONLY.
3164 (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_NONTEMPORAL. Handle
3165 OMP_CLAUSE_{TASK,IN}_REDUCTION.
3166 (gimplify_omp_task): Handle taskwait with depend clauses.
3167 (gimplify_omp_for): Add shared clause on parallel for combined
3168 parallel master taskloop{, simd} if taskloop has firstprivate,
3169 lastprivate or reduction clause. Use ORT_TASKLOOP or
3170 ORT_UNTIED_TASKLOOP instead of ORT_TASK or ORT_UNTIED_TASK. Adjust
3171 tests for ORT_COMBINED_TEAMS. Handle C++ range for loops with
3172 NULL TREE_PURPOSE in OMP_FOR_ORIG_DECLS. Firstprivatize
3173 __for_end and __for_range temporaries on OMP_PARALLEL for
3174 distribute parallel for{, simd}. Move OMP_CLAUSE_REDUCTION
3175 and OMP_CLAUSE_IN_REDUCTION from taskloop to the task construct
3176 sandwiched in between two taskloops.
3177 (computable_teams_clause): Test ctx->defaultmap[GDMK_SCALAR]
3178 instead of ctx->omp_firstprivatize_variable.
3179 (gimplify_omp_workshare): Set ort to ORT_HOST_TEAMS or
3180 ORT_COMBINED_HOST_TEAMS if not inside of target construct. If
3181 host teams, use gimplify_and_return_first etc. for body like
3182 for target or target data constructs, and at the end call
3183 gimple_omp_teams_set_host on the GIMPLE_OMP_TEAMS object.
3184 (gimplify_omp_atomic): Use OMP_ATOMIC_MEMORY_ORDER instead
3185 of OMP_ATOMIC_SEQ_CST, pass it as new argument to
3186 gimple_build_omp_atomic_load and gimple_build_omp_atomic_store, remove
3187 gimple_omp_atomic_set_seq_cst calls.
3188 (gimplify_expr) <case OMP_TASKGROUP>: Move handling into a separate
3189 case, handle taskgroup clauses.
3190 * lto-streamer-out.c (hash_tree): Handle
3191 OMP_CLAUSE_{TASK,IN}_REDUCTION.
3192 * Makefile.in (GTFILES): Add omp-general.h.
3193 * omp-builtins.def (BUILT_IN_GOMP_TASKWAIT_DEPEND,
3194 BUILT_IN_GOMP_LOOP_NONMONOTONIC_RUNTIME_START,
3195 BUILT_IN_GOMP_LOOP_MAYBE_NONMONOTONIC_RUNTIME_START,
3196 BUILT_IN_GOMP_LOOP_START, BUILT_IN_GOMP_LOOP_ORDERED_START,
3197 BUILT_IN_GOMP_LOOP_DOACROSS_START,
3198 BUILT_IN_GOMP_LOOP_NONMONOTONIC_RUNTIME_NEXT,
3199 BUILT_IN_GOMP_LOOP_MAYBE_NONMONOTONIC_RUNTIME_NEXT,
3200 BUILT_IN_GOMP_LOOP_ULL_NONMONOTONIC_RUNTIME_START,
3201 BUILT_IN_GOMP_LOOP_ULL_MAYBE_NONMONOTONIC_RUNTIME_START,
3202 BUILT_IN_GOMP_LOOP_ULL_START, BUILT_IN_GOMP_LOOP_ULL_ORDERED_START,
3203 BUILT_IN_GOMP_LOOP_ULL_DOACROSS_START,
3204 BUILT_IN_GOMP_LOOP_ULL_NONMONOTONIC_RUNTIME_NEXT,
3205 BUILT_IN_GOMP_LOOP_ULL_MAYBE_NONMONOTONIC_RUNTIME_NEXT,
3206 BUILT_IN_GOMP_PARALLEL_LOOP_NONMONOTONIC_RUNTIME,
3207 BUILT_IN_GOMP_PARALLEL_LOOP_MAYBE_NONMONOTONIC_RUNTIME,
3208 BUILT_IN_GOMP_PARALLEL_REDUCTIONS, BUILT_IN_GOMP_SECTIONS2_START,
3209 BUILT_IN_GOMP_TEAMS_REG, BUILT_IN_GOMP_TASKGROUP_REDUCTION_REGISTER,
3210 BUILT_IN_GOMP_TASKGROUP_REDUCTION_UNREGISTER,
3211 BUILT_IN_GOMP_TASK_REDUCTION_REMAP,
3212 BUILT_IN_GOMP_WORKSHARE_TASK_REDUCTION_UNREGISTER): New builtins.
3213 * omp-expand.c (workshare_safe_to_combine_p): Return false for
3214 non-worksharing loops.
3215 (omp_adjust_chunk_size): Don't adjust anything if chunk_size is zero.
3216 (determine_parallel_type): Don't combine parallel with worksharing
3217 which has _reductemp_ clause.
3218 (expand_parallel_call): Emit the GOMP_*nonmonotonic_runtime* or
3219 GOMP_*maybe_nonmonotonic_runtime* builtins instead of GOMP_*runtime*
3220 if there is nonmonotonic modifier or if there is no modifier and no
3221 ordered clause. For dynamic and guided schedule without monotonic
3222 and nonmonotonic modifier, default to nonmonotonic.
3223 (expand_omp_for): Likewise. Adjust expand_omp_for_generic caller, use
3224 GOMP_loop{,_ull}{,_ordered,_doacross}_start builtins if there are
3226 (expand_task_call): Add GOMP_TASK_FLAG_REDUCTION flag to flags if
3227 there are any reduction clauses.
3228 (expand_taskwait_call): New function.
3229 (expand_teams_call): New function.
3230 (expand_omp_taskreg): Allow GIMPLE_OMP_TEAMS and call
3231 expand_teams_call for it. Formatting fix. Handle taskwait with
3233 (expand_omp_for_generic): Add SCHED_ARG argument. Handle expansion
3234 of worksharing loops with task reductions.
3235 (expand_omp_for_static_nochunk, expand_omp_for_static_chunk): Handle
3236 expansion of worksharing loops with task reductions.
3237 (expand_omp_sections): Handle expansion of sections with task
3239 (expand_omp_synch): For host teams call expand_omp_taskreg.
3240 (omp_memory_order_to_memmodel): New function.
3241 (expand_omp_atomic_load, expand_omp_atomic_store,
3242 expand_omp_atomic_fetch_op): Use it and gimple_omp_atomic_memory_order
3243 instead of gimple_omp_atomic_seq_cst_p.
3244 (build_omp_regions_1, omp_make_gimple_edges): Treat taskwait with
3245 depend clauses as a standalone directive.
3246 * omp-general.c (enum omp_requires): New variable.
3247 (omp_extract_for_data): Initialize have_reductemp member. Allow
3248 NE_EXPR even in OpenMP loops, transform them into LT_EXPR or
3249 GT_EXPR loops depending on incr sign. Formatting fixes.
3250 * omp-general.h (struct omp_for_data): Add have_reductemp member.
3251 (enum omp_requires): New enum.
3252 (omp_requires_mask): Declare.
3253 * omp-grid.c (grid_eliminate_combined_simd_part): Formatting fix.
3255 * omp-low.c (struct omp_context): Add task_reductions and
3256 task_reduction_map fields.
3257 (is_host_teams_ctx): New function.
3258 (is_taskreg_ctx): Return true also if is_host_teams_ctx.
3259 (use_pointer_for_field): Use is_global_var instead of
3260 TREE_STATIC || DECL_EXTERNAL, and apply only if not privatized
3262 (build_outer_var_ref): Ignore taskgroup outer contexts.
3263 (delete_omp_context): Release task_reductions and task_reduction_map.
3264 (scan_sharing_clauses): Don't add any fields for reduction clause on
3265 taskloop. Handle OMP_CLAUSE__REDUCTEMP_. Handle
3266 OMP_CLAUSE_{IN,TASK}_REDUCTION and OMP_CLAUSE_REDUCTION with task
3267 modifier. Don't ignore shared clauses in is_host_teams_ctx contexts.
3268 Handle OMP_CLAUSE_NONTEMPORAL.
3269 (add_taskreg_looptemp_clauses): Add OMP_CLAUSE__REDUCTEMP_ clause if
3271 (scan_omp_parallel): Add _reductemp_ clause if there are any reduction
3272 clauses with task modifier.
3273 (scan_omp_task): Handle taskwait with depend clauses.
3274 (finish_taskreg_scan): Move field corresponding to _reductemp_ clause
3275 first. Move also OMP_CLAUSE__REDUCTEMP_ clause in front if present.
3276 Handle GIMPLE_OMP_TEAMS like GIMPLE_OMP_PARALLEL.
3277 (scan_omp_for): Fix comment formatting.
3278 (scan_omp_teams): Handle host teams constructs.
3279 (check_omp_nesting_restrictions): Allow teams with no outer
3280 OpenMP context. Adjust diagnostics for teams strictly nested into
3281 some explicit OpenMP construct other than target. Allow OpenMP atomics
3282 inside of simd regions.
3283 (scan_omp_1_stmt): Call scan_sharing_clauses for taskgroups.
3284 (scan_omp_1_stmt) <case GIMPLE_OMP_TEAMS>: Temporarily bump
3285 taskreg_nesting_level while scanning host teams construct.
3286 (task_reduction_read): New function.
3287 (lower_rec_input_clauses): Handle OMP_CLAUSE_REDUCTION on taskloop
3288 construct. Handle OMP_CLAUSE_IN_REDUCTION and OMP_CLAUSE__REDUCTEMP_
3289 clauses. Handle OMP_CLAUSE_REDUCTION with task modifier. Remove
3290 second argument create_tmp_var if it is NULL. Don't ignore shared
3291 clauses in is_host_teams_ctx contexts. Handle
3292 OMP_CLAUSE_FIRSTPRIVATE_NO_REFERENCE on OMP_CLAUSE_FIRSTPRIVATE
3294 (lower_reduction_clauses): Ignore reduction clauses with task
3295 modifier. Remove second argument create_tmp_var if it is NULL.
3296 Initialize OMP_ATOMIC_MEMORY_ORDER to relaxed.
3297 (lower_send_clauses): Ignore reduction clauses with task modifier.
3298 Handle OMP_CLAUSE__REDUCTEMP_. Don't send anything for
3299 OMP_CLAUSE_REDUCTION on taskloop. Handle OMP_CLAUSE_IN_REDUCTION.
3300 (maybe_add_implicit_barrier_cancel): Add OMP_RETURN argument, don't
3301 rely that it is the last stmt in body so far. Ignore outer taskgroup
3303 (omp_task_reductions_find_first, omp_task_reduction_iterate,
3304 lower_omp_task_reductions): New functions.
3305 (lower_omp_sections): Handle reduction clauses with taskgroup
3306 modifiers. Adjust maybe_add_implicit_barrier_cancel caller.
3307 (lower_omp_single): Adjust maybe_add_implicit_barrier_cancel caller.
3308 (lower_omp_for): Likewise. Handle reduction clauses with taskgroup
3310 (lower_omp_taskgroup): Handle taskgroup reductions.
3311 (create_task_copyfn): Copy over OMP_CLAUSE__REDUCTEMP_ pointer.
3312 Handle OMP_CLAUSE_IN_REDUCTION and OMP_CLAUSE_REDUCTION clauses.
3313 (lower_depend_clauses): If there are any
3314 OMP_CLAUSE_DEPEND_DEPOBJ or OMP_CLAUSE_DEPEND_MUTEXINOUTSET
3315 depend clauses, use a new array format. If OMP_CLAUSE_DEPEND_LAST is
3316 seen, assume lowering is done already and return early. Set kind
3317 on artificial depend clause to OMP_CLAUSE_DEPEND_LAST.
3318 (lower_omp_taskreg): Handle reduction clauses with task modifier on
3319 parallel construct. Handle reduction clause on taskloop construct.
3320 Handle taskwait with depend clauses.
3321 (lower_omp_1): Use lower_omp_taskreg instead of lower_omp_teams
3322 for host teams constructs.
3323 * tree.c (omp_clause_num_ops): Add in_reduction, task_reduction,
3324 nontemporal and _reductemp_ clause entries.
3325 (omp_clause_code_name): Likewise.
3326 (walk_tree_1): Handle OMP_CLAUSE_{IN,TASK}_REDUCTION,
3327 OMP_CLAUSE_NONTEMPORAL and OMP_CLAUSE__REDUCTEMP_.
3328 * tree-core.h (enum omp_clause_code): Add
3329 OMP_CLAUSE_{{IN,TASK}_REDUCTION,NONTEMPORAL,_REDUCTEMP_}.
3330 (enum omp_clause_defaultmap_kind, enum omp_memory_order): New.
3331 (struct tree_base): Add omp_atomic_memory_order field into union.
3332 Remove OMP_ATOMIC_SEQ_CST comment.
3333 (enum omp_clause_depend_kind): Add OMP_CLAUSE_DEPEND_MUTEXINOUTSET
3334 and OMP_CLAUSE_DEPEND_DEPOBJ.
3335 (struct tree_omp_clause): Add subcode.defaultmap_kind.
3336 * tree.def (OMP_TASKGROUP): Add another operand, move next to other
3337 OpenMP constructs with body and clauses operands.
3338 * tree.h (OMP_BODY): Use OMP_MASTER instead of OMP_TASKGROUP.
3339 (OMP_CLAUSES): Use OMP_TASKGROUP instead of OMP_SINGLE.
3340 (OMP_TASKGROUP_CLAUSES): Define.
3341 (OMP_CLAUSE_DECL): Use OMP_CLAUSE__REDUCTEMP_ instead of
3342 OMP_CLAUSE__LOOPTEMP_.
3343 (OMP_ATOMIC_SEQ_CST): Remove.
3344 (OMP_ATOMIC_MEMORY_ORDER, OMP_CLAUSE_FIRSTPRIVATE_NO_REFERENCE,
3345 OMP_CLAUSE_LASTPRIVATE_CONDITIONAL): Define.
3346 (OMP_CLAUSE_REDUCTION_CODE, OMP_CLAUSE_REDUCTION_INIT,
3347 OMP_CLAUSE_REDUCTION_MERGE, OMP_CLAUSE_REDUCTION_PLACEHOLDER,
3348 OMP_CLAUSE_REDUCTION_DECL_PLACEHOLDER,
3349 OMP_CLAUSE_REDUCTION_OMP_ORIG_REF): Handle
3350 OMP_CLAUSE_{,IN_,TASK_}REDUCTION.
3351 (OMP_CLAUSE_REDUCTION_TASK, OMP_CLAUSE_REDUCTION_INSCAN,
3352 OMP_CLAUSE_DEFAULTMAP_KIND, OMP_CLAUSE_DEFAULTMAP_CATEGORY,
3353 OMP_CLAUSE_DEFAULTMAP_BEHAVIOR, OMP_CLAUSE_DEFAULTMAP_SET_KIND):
3355 * tree-inline.c (remap_gimple_stmt): Remap taskgroup clauses.
3356 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
3357 OMP_CLAUSE__REDUCTEMP_, OMP_CLAUSE_NONTEMPORAL.
3358 (convert_local_omp_clauses): Likewise. Remove useless test.
3359 * tree-parloops.c (create_call_for_reduction_1): Pass
3360 OMP_MEMORY_ORDER_RELAXED as new argument to
3361 dump_gimple_omp_atomic_load and dump_gimple_omp_atomic_store.
3362 * tree-pretty-print.c (dump_omp_iterators): New function.
3363 (dump_omp_clause): Handle OMP_CLAUSE__REDUCTEMP_,
3364 OMP_CLAUSE_NONTEMPORAL, OMP_CLAUSE_{TASK,IN}_REDUCTION. Print
3365 reduction modifiers. Handle OMP_CLAUSE_DEPEND_DEPOBJ and
3366 OMP_CLAUSE_DEPEND_MUTEXINOUTSET. Print iterators in depend clauses.
3367 Print __internal__ for OMP_CLAUSE_DEPEND_LAST. Handle cancel and
3368 simd OMP_CLAUSE_IF_MODIFIERs. Handle new kinds of
3369 OMP_CLAUSE_DEFAULTMAP. Print conditional: for
3370 OMP_CLAUSE_LASTPRIVATE_CONDITIONAL.
3371 (dump_omp_atomic_memory_order): New function.
3372 (dump_generic_node): Use it. Print taskgroup clauses. Print
3373 taskwait with depend clauses.
3374 * tree-pretty-print.h (dump_omp_atomic_memory_order): Declare.
3375 * tree-streamer-in.c (unpack_ts_omp_clause_value_fields):
3376 Handle OMP_CLAUSE_{TASK,IN}_REDUCTION.
3377 * tree-streamer-out.c (pack_ts_omp_clause_value_fields,
3378 write_ts_omp_clause_tree_pointers): Likewise.
3380 2018-11-08 David Malcolm <dmalcolm@redhat.com>
3383 * doc/invoke.texi (-fdump-ipa-): Document the "-optimized",
3384 "-missed", "-note", and "-all" sub-options.
3385 * ipa-inline.c (caller_growth_limits): Port from fprintf to dump
3387 (can_early_inline_edge_p): Likewise.
3388 (want_early_inline_function_p): Likewise.
3389 (want_inline_self_recursive_call_p): Likewise.
3390 (recursive_inlining): Likewise.
3391 (inline_small_functions): Likewise.
3392 (flatten_function): Likewise.
3393 (ipa_inline): Likewise.
3394 (inline_always_inline_functions): Likewise.
3395 (early_inline_small_functions): Likewise.
3396 (early_inliner): Likewise.
3397 * tree-inline.c (expand_call_inline): Likewise.
3399 2018-11-08 David Malcolm <dmalcolm@redhat.com>
3401 * pretty-print.c (pp_format): Handle %f.
3402 (selftest::test_pp_format): Add test of %f.
3403 * pretty-print.h (pp_double): New macro.
3405 2018-11-08 David Malcolm <dmalcolm@redhat.com>
3407 * dump-context.h (ASSERT_IS_CGRAPH_NODE): New macro.
3408 * dumpfile.c (make_item_for_dump_cgraph_node): Move to before...
3409 (dump_pretty_printer::decode_format): Implement "%C" for
3411 (selftest::test_capture_of_dump_calls): Rename "where" to
3412 "stmt_loc". Convert test_decl to a function decl and set its
3413 location. Add a symbol_table_test RAII instance and a
3414 cgraph_node, using it to test "%C" and dump_symtab_node.
3416 2018-11-08 Eric Botcazou <ebotcazou@adacore.com>
3419 * cgraphclones.c (duplicate_thunk_for_node): Also set DECL_IGNORED_P.
3421 2018-11-08 David Malcolm <dmalcolm@redhat.com>
3423 * cgraph.c: Include "selftest.h".
3424 (saved_symtab): New variable.
3425 (selftest::symbol_table_test::symbol_table_test): New ctor.
3426 (selftest::symbol_table_test::~symbol_table_test): New dtor.
3427 (selftest::test_symbol_table_test): New test.
3428 (selftest::cgraph_c_tests): New.
3429 * cgraph.h (saved_symtab): New decl.
3430 (selftest::symbol_table_test): New class.
3431 * selftest-run-tests.c (selftest::run_tests): Call
3432 selftest::cgraph_c_tests.
3433 * selftest.h (selftest::cgraph_c_tests): New decl.
3435 2018-11-08 Richard Biener <rguenther@suse.de>
3437 * tree-data-ref.h (lambda_int): New typedef.
3438 (lambda_vector_gcd): Adjust.
3439 (lambda_vector_new): Likewise.
3440 (lambda_matrix_new): Likewise.
3441 * tree-data-ref.c (print_lambda_vector): Adjust.
3443 2018-11-08 Richard Biener <rguenther@suse.de>
3445 PR tree-optimization/87929
3446 * tree-complex.c (expand_complex_comparison): Clean EH.
3448 2018-11-08 Martin Liska <mliska@suse.cz>
3450 * doc/extend.texi: Reword.
3451 * predict.c (expr_expected_value_1): Likewise.
3453 2018-11-08 Richard Biener <rguenther@suse.de>
3455 PR tree-optimization/87913
3456 * tree-ssa-phiopt.c (minmax_replacement): Turn EQ/NE compares
3457 of extreme values to ordered comparisons.
3459 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
3462 * doc/invoke.texi (Code Gen Options): Clarify -fno-common behavior.
3464 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
3466 * doc/invoke.texi: Remove leading dash from @opindex entries
3467 throughout the file.
3469 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
3472 * doc/invoke.texi (Option Summary): Add -e and --entry.
3473 (Link Options): Likewise.
3475 2018-11-07 Nathan Sidwell <nathan@acm.org>
3478 * Makefile.in (bitmap.o-warn): Add -Wno-error to unbreak
3479 --disable-checking bootstrap.
3481 2018-11-07 Hafiz Abid Qadeer <abidh@codesourcery.com>
3483 * configure: Regenerated.
3485 2018-11-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
3488 * stor-layout.c (compute_record_mode): Set TYPE_MODE of UNION_TYPE
3489 to the mode of the widest field iff the widest field has mode class
3490 MODE_INT, or MODE_PARTIAL_INT and the union would be passed by
3493 2018-11-07 Nikolai Merinov <n.merinov@inango-systems.com>
3495 * common.opt: Add -Wattribute-warning.
3496 * doc/invoke.texi: Add documentation for -Wno-attribute-warning.
3497 * expr.c (expand_expr_real_1): Add new attribute to warning_at
3498 call to allow user configure behavior of "warning" attribute.
3500 2018-11-07 Segher Boessenkool <segher@kernel.crashing.org>
3502 * target.def: Put @: after every vs., e.g., and i.e. where it is
3503 followed by whitespace.
3504 * doc/extend.texi: Ditto.
3505 * doc/fragments.texi: Ditto.
3506 * doc/gimple.texi: Ditto.
3507 * doc/implement-c.texi: Ditto.
3508 * doc/install.texi: Ditto.
3509 * doc/invoke.texi: Ditto.
3510 * doc/md.texi: Ditto.
3511 * doc/plugins.texi: Ditto.
3512 * doc/rtl.texi: Ditto.
3513 * doc/sourcebuild.texi: Ditto.
3514 * doc/tm.texi.in: Ditto.
3515 * doc/ux.texi: Ditto.
3516 * doc/tm.texi: Regenerate.
3518 2018-11-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3520 * config/arm/arm-cpus.in (ares): New entry.
3521 * config/arm/arm-tables.opt: Regenerate.
3522 * config/arm/arm-tune.md: Likewise.
3523 * doc/invoke.texi (ARM Options): Document ares.
3525 2018-11-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3527 * config/aarch64/aarch64-cores.def (ares): Define.
3528 * config/aarch64/aarch64-tune.md: Regenerate.
3529 * doc/invoke.texi (AArch64 Options): Document ares value for mtune.
3531 2018-11-07 Jan Hubicka <jh@suse.cz>
3533 * tree.c (fld_incomplete_type_of): Clear TREE_ADDRESSABLE.
3534 (free_lang_data_in_decl): Set TREE_ADDRESSABLE for public vars and
3535 functions; clear TYPE_DECL_SUPPRESS_DEBUG and DECL_MODE for
3538 2018-11-07 Richard Biener <rguenther@suse.de>
3540 PR tree-optimization/87914
3541 * tree-vect-loop.c (vect_is_simple_reduction): Improve detection
3543 (vectorizable_reduction): Handle shifts and rotates by dispatching
3544 to vectorizable_shift.
3545 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Handle
3546 in-loop uses of vect_nested_cycle defs. Merge cycle and internal
3548 (vectorizable_shift): Export and handle being called as
3550 (vect_analyze_stmt): Call vectorizable_shift after
3551 vectorizable_reduction.
3552 * tree-vectorizer.h (vectorizable_shift): Declare.
3554 2018-11-07 Jan Hubicka <jh@suse.cz>
3556 * ipa-devirt.c (odr_types_equivalent_p): Expect constants
3557 than const decls in TREE_VALUE of enum.
3558 (dump_type_inheritance_graph): Improve duplicate dumping.
3559 (free_enum_values): New.
3560 (build_type_inheritance_graph): Use it.
3561 * tree.c (free_lang_data_in_type): Free TYPE_VALUES of enums
3562 which are not main variants or not ODR types.
3563 (verify_type_variant): Expect variants to have no TYPE_VALUES.
3565 2018-11-07 Richard Biener <rguenther@suse.de>
3567 * ipa-inline.c (want_inline_small_function_p): Compute
3568 big_speedup_p lazily and last.
3570 2018-11-07 Jan Hubicka <jh@suse.cz>
3572 * tree.c (fld_type_variant_equal_p): Skip TYPE_ALIGN check when
3573 building incomplete variant of complete type.
3574 (fld_type_variant): Do not copy TYPE_ALIGN when building incomplete
3575 variant of complete type.
3577 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
3579 * config/mips/mips.c: Fix typo in documentation of
3580 mips_loongson_ext2_prefetch_cookie.
3581 (mips_option_override): fix brain twister logical.
3582 * config/mips/mips.h: Fix typo in documentation of
3583 ISA_HAS_CTZ_CTO and define pattern.
3584 * config/mips/mips.md (prefetch): Hoist EXT2 above
3586 (prefetch_indexed): Hoist EXT2 above the EXT block.
3588 2018-11-07 Jan Hubicka <jh@suse.cz>
3590 * tree.c (free_lang_data_in_type): Add fld parameter; simplify
3591 return and parameter types of function and method types.
3592 (free_lang_data_in_cgraph): Update.
3594 2018-11-07 Martin Liska <mliska@suse.cz>
3596 PR rtl-optimization/87868
3597 * postreload-gcse.c (eliminate_partially_redundant_load): Set
3598 threshold to max_count if we would overflow.
3599 * profile-count.h: Make max_count a public constant.
3601 2018-11-07 Martin Liska <mliska@suse.cz>
3603 * mem-stats.h: Fix GNU coding style.
3605 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
3607 * config/mips/gs264e.md: New.
3608 * config/mips/mips-cpus.def: Define gs264e.
3609 * config/mips/mips-tables.opt: Regenerate.
3610 * config/mips/mips.c (mips_rtx_cost_data): Add DEFAULT_COSTS for
3612 (mips_issue_rate): Add support for gs264e.
3613 (mips_multipass_dfa_lookahead): Likewise.
3614 * config/mips/mips.h: Define TARGET_GS264E and TUNE_GS264E.
3615 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs264e.
3616 (MIPS_ASE_MSA_SPEC): New.
3617 (BASE_DRIVER_SELF_SPECS): march=gs264e implies -mmsa.
3618 (ISA_HAS_FUSED_MADD4): Enable for TARGET_GS264E.
3619 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_GS264E.
3620 * config/mips/mips.md: Include gs264e.md.
3621 (processor): Add gs264e.
3622 * config/mips/mips.opt (MSA): Use Mask instead of Var.
3623 * doc/invoke.texi: Add gs264e to supported architectures.
3625 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
3627 * config/mips/gs464e.md: New.
3628 * config/mips/mips-cpus.def: Define gs464e.
3629 * config/mips/mips-tables.opt: Regenerate.
3630 * config/mips/mips.c (mips_rtx_cost_data): Add DEFAULT_COSTS for
3632 (mips_issue_rate): Add support for gs464e.
3633 (mips_multipass_dfa_lookahead): Likewise.
3634 (mips_option_override): Enable MMI, EXT and EXT2 for gs464e.
3635 * config/mips/mips.h: Define TARGET_GS464E and TUNE_GS464E.
3636 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs464e.
3637 (ISA_HAS_FUSED_MADD4): Enable for TARGET_GS464E.
3638 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_GS464E.
3639 * config/mips/mips.md: Include gs464e.md.
3640 (processor): Add gs464e.
3641 * doc/invoke.texi: Add gs464e to supported architectures.
3643 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
3645 * config/mips/loongson3a.md: Rename to ...
3646 * config/mips/gs464.md: ... here.
3647 * config/mips/mips-cpus.def: Define gs464; Add loongson3a
3648 as an alias of gs464 processor.
3649 * config/mips/mips-tables.opt: Regenerate.
3650 * config/mips/mips.c (mips_issue_rate): Use PROCESSOR_GS464
3651 instead of PROCESSOR_LOONGSON_3A.
3652 (mips_multipass_dfa_lookahead): Use TUNE_GS464 instead of
3654 (mips_option_override): Enable MMI and EXT for gs464.
3655 * config/mips/mips.h: Rename TARGET_LOONGSON_3A to TARGET_GS464;
3656 Rename TUNE_LOONGSON_3A to TUNE_GS464.
3657 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs464.
3658 (ISA_HAS_ODD_SPREG, ISA_AVOID_DIV_HILO, ISA_HAS_FUSED_MADD4,
3659 ISA_HAS_UNFUSED_MADD4): Use TARGET_GS464 instead of
3661 * config/mips/mips.md: Include gs464.md instead of loongson3a.md.
3662 (processor): Add gs464;
3663 * doc/invoke.texi: Add gs464 to supported architectures.
3665 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
3667 * config/mips/mips-protos.h
3668 (mips_loongson_ext2_prefetch_cookie): New prototype.
3669 * config/mips/mips.c (mips_loongson_ext2_prefetch_cookie): New.
3670 (mips_option_override): Enable TARGET_LOONGSON_EXT when
3671 TARGET_LOONGSON_EXT2 is true.
3672 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define
3673 __mips_loongson_ext2, __mips_loongson_ext_rev=2.
3674 (ISA_HAS_CTZ_CTO): New, true if TARGET_LOONGSON_EXT2.
3675 (ISA_HAS_PREFETCH): Include TARGET_LOONGSON_EXT and
3676 TARGET_LOONGSON_EXT2.
3677 (ASM_SPEC): Add mloongson-ext2 and mno-loongson-ext2.
3678 (define_insn "ctz<mode>2"): New insn pattern.
3679 (define_insn "prefetch"): Include TARGET_LOONGSON_EXT2.
3680 (define_insn "prefetch_indexed_<mode>"): Include
3681 TARGET_LOONGSON_EXT and TARGET_LOONGSON_EXT2.
3682 * config/mips/mips.opt (-mloongson-ext2): Add option.
3683 * gcc/doc/invoke.texi (-mloongson-ext2): Document.
3685 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
3687 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Add
3688 __mips_loongson_ext.
3689 (MIPS_ASE_LOONGSON_EXT_SPEC): New.
3690 (BASE_DRIVER_SELF_SPECS): march=loongson3a implies
3692 (ASM_SPEC): Add mloongson-ext and mno-loongson-ext.
3693 * config/mips/mips.md (mul<mode>3, mul<mode>3_mul3_nohilo,
3694 <u>div<mode>3, <u>mod<mode>3, prefetch): Use TARGET_LOONGSON_EXT
3695 instead of TARGET_LOONGSON_3A.
3696 * config/mips/mips.opt (-mloongson-ext): Add option.
3697 * gcc/doc/invoke.texi (-mloongson-ext): Document.
3699 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
3701 * config.gcc (extra_headers): Add loongson-mmiintrin.h.
3702 * config/mips/loongson.md: Move to ...
3703 * config/mips/loongson-mmi.md: here; Adjustment.
3704 * config/mips/loongson.h: Move to ...
3705 State as deprecated. Include loongson-mmiintrin.h for back
3706 compatibility and warning.
3707 * config/mips/loongson-mmiintrin.h: ... here.
3708 * config/mips/mips.c (mips_hard_regno_mode_ok_uncached,
3709 mips_vector_mode_supported_p, AVAIL_NON_MIPS16): Use
3710 TARGET_LOONGSON_MMI instead of TARGET_LOONGSON_VECTORS.
3711 (mips_option_override): Make sure MMI use hard float;
3712 (mips_shift_truncation_mask, mips_expand_vpc_loongson_even_odd,
3713 mips_expand_vpc_loongson_pshufh, mips_expand_vpc_loongson_bcast,
3714 mips_expand_vector_init): Use TARGET_LOONGSON_MMI instead of
3715 TARGET_LOONGSON_VECTORS.
3716 * gcc/config/mips/mips.h (TARGET_LOONGSON_VECTORS): Delete.
3717 (TARGET_CPU_CPP_BUILTINS): Add __mips_loongson_mmi.
3718 (MIPS_ASE_DSP_SPEC, MIPS_ASE_LOONGSON_MMI_SPEC): New.
3719 (BASE_DRIVER_SELF_SPECS): march=loongson2e/2f/3a implies
3721 (SHIFT_COUNT_TRUNCATED): Use TARGET_LOONGSON_MMI instead of
3722 TARGET_LOONGSON_VECTORS.
3723 * gcc/config/mips/mips.md (MOVE64, MOVE128): Use
3724 TARGET_LOONGSON_MMI instead of TARGET_LOONGSON_VECTORS.
3725 (Loongson MMI patterns): Include loongson-mmi.md instead of
3727 * gcc/config/mips/mips.opt (-mloongson-mmi): New option.
3728 * gcc/doc/invoke.texi (-mloongson-mmi): Document.
3730 2018-11-07 Richard Biener <rguenther@suse.de>
3733 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Fixup
3734 BLOCK_ABSTRACT_ORIGIN to be the ultimate origin.
3736 2018-11-07 Alexandre Oliva <aoliva@redhat.com>
3738 PR rtl-optimization/87874
3739 * lra.c (lra_substitute_pseudo): Do not create a subreg for
3742 2018-11-06 Aaron Sawdey <acsawdey@linux.ibm.com>
3744 * config/rs6000/rs6000.md (bswap<mode>2): Force address into register
3745 if not in indexed or indirect form.
3746 (bswap<mode>2_load): Change predicate to indexed_or_indirect_operand.
3747 (bswap<mode>2_store): Ditto.
3749 2018-11-06 Richard Earnshaw <rearnsha@arm.com>
3751 * config/aarch64/aarch64.md (speculation_tracker): Set the mode for
3754 2018-11-06 Richard Biener <rguenther@suse.de>
3756 PR tree-optimization/86850
3757 * vec.h (vec<T, va_heap, vl_ptr>::splice): Check src.length ()
3758 instead of src.m_vec.
3760 2018-11-06 Jan Hubicka <jh@suse.cz>
3762 * tree.c (fld_simplified_type_name): Break out form ...
3763 (free_lang_data_in_type): ... here.
3764 (fld_type_variant_equal_p): Use it.
3766 2018-11-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3768 * config/default-d.c: Include memmodel.h.
3770 * config/sol2-d.c: New file.
3771 * config/t-sol2 (sol2-d.o): New rule.
3772 * config.gcc <*-*-solaris2*>: Set d_target_objs,
3773 target_has_targetdm.
3775 2018-11-06 Jan Hubicka <jh@suse.cz>
3777 * tree.c (fld_type_variant): Also copy alignment; be sure that
3778 new variant is equal.
3780 2018-11-06 Ilya Leoshkevich <iii@linux.ibm.com>
3783 * config/s390/s390.md: Add relative_long attribute.
3785 2018-11-06 Jan Hubicka <jh@suse.cz>
3787 * ipa-pure-const.c (check_decl): Do not test TYPE_NEEDS_CONSTRUCTING.
3788 * lto-streamer-out.c (hash_tree): Do not hash TYPE_NEEDS_CONSTRUCTING.
3789 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Do not
3790 stream TYPE_NEEDS_CONSTRUCTING.
3791 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
3792 * tree.c (free_lang_data_in_type): Clear TYPE_NEEDS_CONSTRUCTING.
3794 2018-11-06 Richard Biener <rguenther@suse.de>
3796 * tree-vect-slp.c (vect_slp_bb): Move opening of vect_slp_analyze_bb
3798 (vect_slp_analyze_bb_1): ... here to avoid hiding optimized locations.
3800 2018-11-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
3802 * gcc/config/msp430/msp430.h (REG_CLASS_CONTENTS): Add R0 to
3803 REG_CLASS_CONTENTS[GEN_REGS].
3804 (REGNO_REG_CLASS): Return NO_REGS for R2 and R3.
3806 2018-11-06 Jan Hubicka <jh@suse.cz>
3808 * tree.c (fld_simplified_type_of): Clear TYPELESS_STORAGE flag.
3810 2018-11-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3813 * config/sol2.h (ASAN_CC1_SPEC): Define.
3814 (LD_WHOLE_ARCHIVE_OPTION): Define.
3815 (LD_NO_WHOLE_ARCHIVE_OPTION): Define.
3816 (ASAN_REJECT_SPEC): Provide default.
3817 (LIBASAN_EARLY_SPEC): Define.
3818 (LIBTSAN_EARLY_SPEC): Define.
3819 (LIBLSAN_EARLY_SPEC): Define.
3820 * config/i386/sol2.h (CC1_SPEC): Redefine.
3821 (ASAN_REJECT_SPEC): Define.
3823 * config/sparc/sparc.c (sparc_asan_shadow_offset): Declare.
3824 (TARGET_ASAN_SHADOW_OFFSET): Define.
3825 (sparc_asan_shadow_offset): New function.
3826 * config/sparc/sol2.h (CC1_SPEC): Append ASAN_CC1_SPEC.
3827 (ASAN_REJECT_SPEC): Define.
3829 2018-11-06 Jan Hubicka <jh@suse.cz>
3831 * tree.c (fld_type_variant): Copy canonical type.
3832 (fld_incomplete_type_of): Check that canonical types looks sane;
3833 copy canonical type.
3834 (verify_type): Accept when incomplete type has complete canonical type.
3836 2018-11-06 Jan Hubicka <jh@suse.cz>
3838 * tree.c (free_lang_data): Reset overwite_assembler_name,
3839 print_xnode, print_decl, print_type and print_identifier of
3842 2018-11-06 Richard Biener <rguenther@suse.de>
3844 PR tree-optimization/87889
3845 * tree-vect-loop-manip.c (slpeel_duplicate_current_defs_from_edges):
3846 Do nothing if old and new arg are the same
3848 2018-11-06 Andreas Krebbel <krebbel@linux.ibm.com>
3851 * config/s390/s390.md ("*r<noxa>sbg_di_rotl"): Remove mode
3852 attributes for operands 3 and 4.
3854 2018-11-06 Richard Biener <rguenther@suse.de>
3857 * simplify-rtx.c (simplify_binary_operation_1): Add pattern
3858 matching bitfield insertion.
3860 2018-11-06 Alexandre Oliva <aoliva@redhat.com>
3862 * auto-inc-dec.c: Include valtrack.h. Improve comments.
3863 (reg_next_debug_use): New.
3864 (attempt_change): Propagate adjusted expression into affected
3866 (merge_in_block): Track uses in debug insns.
3867 (pass_inc_dec::execute): Allocate and release
3870 2018-11-06 Wei Xiao <wei3.xiao@intel.com>
3872 * config/i386/avx512fintrin.h: Update VFIXUPIMM* intrinsics.
3873 (_mm512_fixupimm_round_pd): Update parameters and builtin.
3874 (_mm512_maskz_fixupimm_round_pd): Ditto.
3875 (_mm512_fixupimm_round_ps): Ditto.
3876 (_mm512_maskz_fixupimm_round_ps): Ditto.
3877 (_mm_fixupimm_round_sd): Ditto.
3878 (_mm_maskz_fixupimm_round_sd): Ditto.
3879 (_mm_fixupimm_round_ss): Ditto.
3880 (_mm_maskz_fixupimm_round_ss): Ditto.
3881 (_mm512_fixupimm_pd): Ditto.
3882 (_mm512_maskz_fixupimm_pd): Ditto.
3883 (_mm512_fixupimm_ps): Ditto.
3884 (_mm512_maskz_fixupimm_ps): Ditto.
3885 (_mm_fixupimm_sd): Ditto.
3886 (_mm_maskz_fixupimm_sd): Ditto.
3887 (_mm_fixupimm_ss): Ditto.
3888 (_mm_maskz_fixupimm_ss): Ditto.
3889 (_mm512_mask_fixupimm_round_pd): Update builtin.
3890 (_mm512_mask_fixupimm_round_ps): Ditto.
3891 (_mm_mask_fixupimm_round_sd): Ditto.
3892 (_mm_mask_fixupimm_round_ss): Ditto.
3893 (_mm512_mask_fixupimm_pd): Ditto.
3894 (_mm512_mask_fixupimm_ps): Ditto.
3895 (_mm_mask_fixupimm_sd): Ditto.
3896 (_mm_mask_fixupimm_ss): Ditto.
3897 * config/i386/avx512vlintrin.h:
3898 (_mm256_fixupimm_pd): Update parameters and builtin.
3899 (_mm256_maskz_fixupimm_pd): Ditto.
3900 (_mm256_fixupimm_ps): Ditto.
3901 (_mm256_maskz_fixupimm_ps): Ditto.
3902 (_mm_fixupimm_pd): Ditto.
3903 (_mm_maskz_fixupimm_pd): Ditto.
3904 (_mm_fixupimm_ps): Ditto.
3905 (_mm_maskz_fixupimm_ps): Ditto.
3906 (_mm256_mask_fixupimm_pd): Update builtin.
3907 (_mm256_mask_fixupimm_ps): Ditto.
3908 (_mm_mask_fixupimm_pd): Ditto.
3909 (_mm_mask_fixupimm_ps): Ditto.
3910 * config/i386/i386-builtin-types.def: Add new types and remove useless ones.
3911 * config/i386/i386-builtin.def: Update builtin definitions.
3912 * config/i386/i386.c: Handle new builtin types and remove useless ones.
3913 * config/i386/sse.md: Update VFIXUPIMM* patterns.
3914 (<avx512>_fixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
3915 (<avx512>_fixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
3916 (<avx512>_fixupimm<mode>_mask<round_saeonly_name>): Update.
3917 (avx512f_sfixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
3918 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
3919 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Update.
3920 * config/i386/subst.md:
3921 (round_saeonly_sd_mask_operand4): Add new subst_attr.
3922 (round_saeonly_sd_mask_op4): Ditto.
3923 (round_saeonly_expand_operand5): Ditto.
3924 (round_saeonly_expand): Update.
3926 2018-11-05 Max Filippov <jcmvbkbc@gmail.com>
3928 * config/xtensa/uclinux.h (XTENSA_ALWAYS_PIC): Change to 0.
3930 2018-11-05 Segher Boessenkool <segher@kernel.crashing.org>
3932 PR rtl-optimization/87871
3933 * combine.c (make_more_copies): Skip if dest is frame_pointer_rtx.
3935 2018-11-05 Paul Koning <ni1d@arrl.net>
3937 * doc/sourcebuild.texi (target attributes): Document new "inf"
3938 effective target keyword.
3940 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
3942 * config/s390/s390.c (s390_register_move_cost): Increase costs for
3943 moves involving the CC reg.
3945 2018-11-05 Richard Biener <rguenther@suse.de>
3947 PR tree-optimization/87873
3948 * tree-ssa-loop-manip.h (split_loop_exit_edge): Add copy_constants_p
3950 * tree-ssa-loop-manip.c (split_loop_exit_edge): Likewise.
3951 * tree-vect-loop.c (vect_transform_loop): When splitting the
3952 loop exit also create forwarder PHIs for constants.
3953 * tree-vect-loop-manip.c (slpeel_duplicate_current_defs_from_edges):
3954 Handle constant to_arg, add extra checking we match up the correct
3957 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
3959 * config/s390/s390.md: QImode and HImode for load on condition.
3961 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
3963 * config/s390/predicates.md: Fix typo.
3964 * config/s390/s390.md: Allow immediates for load on condition.
3966 2018-11-05 Martin Liska <mliska@suse.cz>
3968 * alloc-pool.h (struct pool_usage): Use SIZE_AMOUNT.
3969 * bitmap.h (struct bitmap_usage): Likewise.
3970 * ggc-common.c (SCALE): Remove.
3972 (struct ggc_usage): Use SIZE_AMOUNT. And update
3974 * ggc-page.c (SCALE): Remove.
3975 (STAT_LABEL): Remove.
3976 (ggc_print_statistics): Use SIZE_AMOUNT.
3977 * gimple.h (SCALE): Remove.
3979 * input.c (ONE_K): Remove.
3982 (STAT_LABEL): Likewise.
3983 (FORMAT_AMOUNT): Likewise.
3984 (dump_line_table_statistics): Use SIZE_AMOUNT.
3985 * mem-stats.h (struct mem_usage): Likewise.
3986 * rtl.c (dump_rtx_statistics): Likewise.
3987 (rtx_alloc_counts): Change type to size_t.
3988 (rtx_alloc_sizes): Likewise.
3989 (rtx_count_cmp): New.
3990 (dump_rtx_statistics): Sort first based on counts.
3991 * tree.c (tree_nodes_cmp): New.
3992 (tree_codes_cmp): New.
3993 (dump_tree_statistics): Sort first based on counts.
3994 * system.h (ONE_K): New.
3996 (SIZE_SCALE): Likewise.
3997 (SIZE_LABEL): Likewise.
3998 (SIZE_AMOUNT): Likewise.
3999 * tree-cfg.c (dump_cfg_stats): Use SIZE_AMOUNT.
4000 * tree-dfa.c (dump_dfa_stats): Likewise.
4001 * tree-phinodes.c (phinodes_print_statistics): Likewise.
4002 * tree-ssanames.c (ssanames_print_statistics): Likewise.
4003 * tree.c (dump_tree_statistics): Likewise.
4004 * vec.c (struct vec_usage): Likewise.
4005 * trans-mem.c (tm_mangle): Enlarge buffer in order to not
4006 trigger a -Werror=format-overflow with
4007 --enable-gather-detailed-stats.
4009 2018-11-05 Martin Liska <mliska@suse.cz>
4011 * mem-stats.h (mem_alloc_description::release_instance_overhead):
4013 * vec.c (struct vec_usage): Register m_element_size.
4014 (vec_prefix::register_overhead): New arguments: elements and
4016 (vec_prefix::release_overhead): Subtract elements.
4017 * vec.h (struct vec_prefix): Change signature.
4018 (va_heap::reserve): Pass proper arguments.
4019 (va_heap::release): Likewise.
4021 2018-11-05 Martin Liska <mliska@suse.cz>
4023 * mem-stats.h (mem_alloc_description::get_list): Fix GNU coding
4027 2018-11-05 Richard Biener <rguenther@suse.de>
4029 * tree-scalar-evolution.h (final_value_replacement_loop): Update
4031 * tree-scalar-evolution.c (final_value_replacement_loop): Return
4032 whether anything was done.
4033 (scev_const_prop): Remove constant propagation part, fold
4035 * tree-ssa-loop.c (pass_scev_cprop::execute): ... here.
4036 (pass_data_scev_cprop): TODO_cleanup_cfg is now done
4039 2018-11-05 Jakub Jelinek <jakub@redhat.com>
4041 PR tree-optimization/87859
4042 * gimple-ssa-store-merging.c (struct merged_store_group): Add
4043 only_constants and first_nonmergeable_order members.
4044 (merged_store_group::merged_store_group): Initialize them.
4045 (merged_store_group::do_merge): Clear only_constants member if
4046 adding something other than INTEGER_CST store.
4047 (imm_store_chain_info::coalesce_immediate_stores): Don't merge
4048 stores with order >= first_nonmergeable_order. Use
4049 merged_store->only_constants instead of always recomputing it.
4050 Set merged_store->first_nonmergeable_order if we've skipped any
4051 stores. Attempt to merge overlapping INTEGER_CST stores that
4052 we would otherwise skip.
4055 * match.pd (X + Y < X): Don't optimize if TYPE_OVERFLOW_SANITIZED.
4057 2018-11-05 Xuepeng Guo <xuepeng.guo@intel.com>
4060 * config/i386/emmintrin.h (__v16qs): New to cope with option
4062 (_mm_cmpeq_epi8): Replace __v16qi with __v16qs.
4063 (_mm_cmplt_epi8): Likewise.
4064 (_mm_cmpgt_epi8): Likewise.
4066 2018-11-05 Richard Biener <rguenther@suse.de>
4068 PR rtl-optimization/87852
4069 * fwprop.c (use_killed_between): Only consider single-defs of the
4070 use whose definition statement dominates the use.
4072 2018-11-05 Martin Liska <mliska@suse.cz>
4075 * doc/invoke.texi: Remove options that are
4076 not disabled with -Os.
4078 2018-11-05 Martin Liska <mliska@suse.cz>
4081 * doc/extend.texi: Update constrain about the last argument
4082 of __builtin_expect_with_probability.
4084 2018-11-05 Martin Liska <mliska@suse.cz>
4087 * predict.c (expr_expected_value_1): Verify
4088 that last argument is a real constants and emit
4091 2018-11-05 Martin Liska <mliska@suse.cz>
4093 PR gcov-profile/77698
4094 * ipa-profile.c (ipa_profile): Adjust hotness threshold
4097 2018-11-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
4099 PR tree-optimization/86572
4100 * builtins.c (c_strlen): Handle negative offsets in a safe way.
4102 2018-11-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
4104 PR tree-optimization/87672
4105 * gimple-fold.c (gimple_fold_builtin_stxcpy_chk): Gimplify.
4106 * tree-ssa-strlen.c (handle_builtin_strcat): Adjust object size.
4108 2018-11-04 Uros Bizjak <ubizjak@gmail.com>
4111 * cfgexpand.c (pass_expand::execute): Move the call to
4112 finish_eh_generation in front of the call to expand_stack_alignment.
4114 2018-11-04 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
4116 * common/config/i386/i386-common.c (processor_alias_table): Add
4118 * config.gcc (i[34567]86-*-linux* | ...): Add znver2.
4119 (case ${target}): Add znver2.
4120 * config/i386/driver-i386.c: (host_detect_local_cpu): Let
4121 -march=native recognize znver2 processors.
4122 * config/i386/i386-c.c (ix86_target_macros_internal): Add znver2.
4123 * config/i386/i386.c (m_znver2): New definition.
4124 (m_ZNVER): New definition.
4125 (m_AMD_MULTIPLE): Includes m_znver2.
4126 (processor_cost_table): Add znver2 entry.
4127 (processor_target_table): Add znver2 entry.
4128 (get_builtin_code_for_version): Set priority for
4130 (processor_model): Add M_AMDFAM17H_ZNVER2.
4131 (arch_names_table): Ditto.
4132 (ix86_reassociation_width): Include znver2.
4133 * config/i386/i386.h (TARGET_znver2): New definition.
4134 (struct ix86_size_cost): Add TARGET_ZNVER2.
4135 (enum processor_type): Add PROCESSOR_ZNVER2.
4136 * config/i386/i386.md (define_attr "cpu"): Add znver2.
4137 * config/i386/x86-tune-costs.h: (processor_costs) Add znver2 costs.
4138 * config/i386/x86-tune-sched.c: (ix86_issue_rate): Add znver2.
4139 (ix86_adjust_cost): Add znver2.
4140 * config/i386/x86-tune.def: Replace m_ZNVER1 by m_ZNVER.
4141 * gcc/doc/extend.texi: Add details about znver2.
4142 * gcc/doc/invoke.texi: Add details about znver2.
4144 2018-11-03 Sandra Loosemore <sandra@codesourcery.com>
4148 * config/nios2/nios2.c (nios2_rtx_costs): Recognize <mul>sidi3
4151 2018-11-02 Aaron Sawdey <acsawdey@linux.ibm.com>
4153 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Pay
4154 attention to TARGET_AVOID_XFORM and BYTES_BIG_ENDIAN.
4156 2018-11-02 Richard Earnshaw <rearnsha@arm.com>
4158 * config/aarch64/aarch64.c ((aarch64_override_options): Disable
4159 shrink-wrapping when -mtrack-speculation.
4161 2018-11-02 Richard Biener <rguenther@suse.de>
4163 * tree-ssa-coalesce.c (struct coalesce_list): Add obstack member.
4164 (pop_cost_one_pair): Do not free pair.
4165 (pop_best_coalesce): Likewise.
4166 (create_coalesce_list): Initialize obstack.
4167 (delete_coalesce_list): Free obstack.
4168 (find_coalesce_pair): Obstack-allocate coalesce pairs.
4169 (add_cost_one_coalesce): Likewise.
4170 (struct live_track): Remove bitmap pointer indirections.
4171 (new_live_track): Adjust.
4172 (delete_live_track): Likewise.
4173 (live_track_remove_partition): Likewise.
4174 (live_track_add_partition): Likewise.
4175 (live_track_live_p): Likewise.
4176 (live_track_process_def): Likewise.
4177 (live_track_clear_base_vars): Likewise.
4179 2018-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4181 * configure.ac (gcc_cv_as_sparc_register_op): Remove.
4182 * configure: Regenerate.
4183 * config.in: Regenerate.
4184 * config/sparc/sparc.c (sparc_output_scratch_registers): Remove
4185 HAVE_AS_REGISTER_PSEUDO_OP guard.
4186 * config/sparc/sparc.h (ASM_ARCH64_SPEC): Remove
4187 !HAVE_AS_REGISTER_PSEUDO_OP support.
4188 (ASM_DECLARE_REGISTER_GLOBAL): Remove HAVE_AS_REGISTER_PSEUDO_OP
4191 2018-11-02 Richard Biener <rguenther@suse.de>
4193 * dwarf2out.c (build_abbrev_table): Guard lookup_external_ref call
4196 2018-11-02 Richard Biener <rguenther@suse.de>
4198 PR tree-optimization/87776
4199 * tree-ssa-sccvn.c (do_rpo_vn): Do not mark backedges
4200 executable when iterating but running into rpo-vn-max-loop-depth
4201 and not eliding the iteration.
4203 2018-11-30 Jan Hubicka <jh@suse.cz>
4205 * tree.c (free_lang_data_in_decl): Free TREE_TYPE and alignment of
4208 2018-11-01 Aaron Sawdey <acsawdey@linux.ibm.com>
4210 * config/rs6000/rs6000-protos.h (rs6000_address_for_fpconvert): Remove
4212 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem):
4213 Combine with rs6000_address_for_fpconvert.
4214 (rs6000_address_for_fpconvert): Combine with
4215 rs6000_force_indexed_or_indirect_mem.
4216 (rs6000_expand_vector_init): Change function call from
4217 rs6000_address_for_fpconvert to rs6000_force_indexed_or_indirect_mem.
4218 * config/rs6000/rs6000.md (floatsi<mode>2_lfiwax): Change call from
4219 rs6000_address_for_fpconvert to rs6000_force_indexed_or_indirect_mem.
4220 (floatsi<mode>2_lfiwax_mem): Ditto.
4221 (floatunssi<mode>2_lfiwzx): Ditto.
4222 (floatunssi<mode>2_lfiwzx_mem): Ditto.
4223 (float<QHI:mode><FP_ISA3:mode>2): Ditto.
4224 (floatuns<QHI:mode><FP_ISA3:mode>2): Ditto.
4225 (fix_trunc<mode>si2_stfiwx): Ditto.
4226 (fixuns_trunc<mode>si2_stfiwx): Ditto.
4227 (float_<mode>si2_hw): Ditto.
4228 (floatuns_<mode>si2_hw): Ditto.
4229 * config/rs6000/vsx.md (*vsx_extract_si): Ditto.
4230 (vsx_splat_<mode>): Ditto.
4232 2018-11-01 Joseph Myers <joseph@codesourcery.com>
4234 * configure.ac (gcc_cv_as_ia64_ltoffx_ldxmov_relocs): Add newline
4235 at end of assembler input text.
4236 * configure: Regenerate.
4238 2018-11-01 Jakub Jelinek <jakub@redhat.com>
4240 PR tree-optimization/87826
4241 * gimple-ssa-store-merging.c (do_shift_rotate): Punt if count is
4242 negative or larger or equal to type's precision.
4244 2018-10-31 Alexandre Oliva <aoliva@redhat.com>
4246 * opts.c (default_options_table): Do not enable
4247 OPT_fdelayed_branch at -Og.
4248 * doc/invoke.texi (-fdelayed-branch): Document it.
4250 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
4252 * optabs-libfuncs.c (build_libfunc_function_visibility):
4253 New, split out from...
4254 (build_libfunc_function): ... here.
4255 (init_one_libfunc_visibility): New, split out from ...
4256 (init_one_libfunc): ... here.
4258 * config/aarch64/atomics.md (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse):
4259 scratch register need not be early-clobber. Document the reason
4260 why we cannot use ST<OP>.
4262 2018-10-31 Joseph Myers <joseph@codesourcery.com>
4265 * configure.ac: Remove AC_PREREQ. Use AC_LANG_SOURCE. Use single
4266 line for second argument of AC_DEFINE_UNQUOTED.
4267 * doc/install.texi (Tools/packages necessary for modifying GCC):
4268 Update to autoconf 2.69 and automake 1.15.1.
4269 * aclocal.m4, config.in, configure: Regenerate.
4271 2018-10-31 Pat Haugen <pthaugen@us.ibm.com>
4273 * tree-ssa-loop-ivcanon.c (try_peel_loop): Fix typo and remove dead
4276 2018-10-31 Martin Liska <mliska@suse.cz>
4279 * common/config/aarch64/aarch64-common.c (aarch64_parse_extension):
4280 Add new argument invalid_extension.
4281 (aarch64_get_all_extension_candidates): New function.
4282 (aarch64_rewrite_selected_cpu): Add NULL to function call.
4283 * config/aarch64/aarch64-protos.h (aarch64_parse_extension): Add
4285 (aarch64_get_all_extension_candidates): New function.
4286 * config/aarch64/aarch64.c (aarch64_parse_arch): Add new
4287 argument invalid_extension.
4288 (aarch64_parse_cpu): Likewise.
4289 (aarch64_print_hint_for_extensions): New function.
4290 (aarch64_validate_mcpu): Provide hint about invalid extension.
4291 (aarch64_validate_march): Likewise.
4292 (aarch64_handle_attr_arch): Pass new argument.
4293 (aarch64_handle_attr_cpu): Provide hint about invalid extension.
4294 (aarch64_handle_attr_isa_flags): Likewise.
4296 2018-10-31 Richard Biener <rguenther@suse.de>
4300 * tree-outof-ssa.c (insert_backedge_copies): Restrict
4301 copy generation to useful cases. Place the copy before
4302 the definition of the backedge value when possible.
4304 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
4306 * config/arc/arc-arch.h (ARC_TUNE_ARC7XX): New tune value.
4307 * config/arc/arc.c (arc_active_insn): New function.
4308 (check_store_cacheline_hazard): Likewise.
4309 (workaround_arc_anomaly): Use check_store_cacheline_hazard.
4310 (arc_override_options): Disable delay slot scheduler for older
4312 (arc_store_addr_hazard_p): New implementation, old one renamed to
4314 (arc_store_addr_hazard_internal_p): Renamed.
4315 (arc_reorg): Don't combine into brcc instructions which are part
4316 of hardware hazard solution.
4317 * config/arc/arc.md (attr tune): Consider new arc7xx tune value.
4318 (tune_arc700): Likewise.
4319 * config/arc/arc.opt (arc7xx): New tune value.
4320 * config/arc/arc700.md: Improve A7 scheduler.
4322 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
4324 * config/arc/arc.c (arc_override_options): Remove
4325 TARGET_COMPACT_CASESI.
4326 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Update.
4327 (CASE_VECTOR_MODE): Likewise.
4328 (CASE_VECTOR_PC_RELATIVE): Likewise.
4329 (CASE_VECTOR_SHORTEN_MODE): Likewise.
4330 (CASE_VECTOR_SHORTEN_MODE1): Delete.
4331 (ADDR_VEC_ALIGN): Update.
4332 (ASM_OUTPUT_CASE_LABEL): Undefine.
4333 (ASM_OUTPUT_BEFORE_CASE_LABEL): Undefine.
4334 (TARGET_BI_BIH): Define.
4335 (DEFAULT_BRANCH_INDEX): Likewise.
4336 * config/arc/arc.md (casesi): Rework to accept BI/BIH
4337 instructions, remove compact_casesi use case.
4338 (casesi_compact_jump): Remove.
4339 (casesi_dispatch): New pattern.
4340 * config/arc/arc.opt: Add mbranch-index option. Deprecate
4341 compact_casesi option.
4342 * doc/invoke.texi: Document mbranch-index option.
4344 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
4346 * config/arc/arc.c (arc_get_tp): Remove function.
4347 (arc_emit_call_tls_get_addr): Likewise.
4348 (arc_call_tls_get_addr): New function.
4349 (arc_legitimize_tls_address): Make use of arc_call_tls_get_addr.
4350 * config/arc/arc.md (tls_load_tp_soft): Remove.
4351 (tls_gd_get_addr): Likewise.
4353 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
4355 * config/arc/arc.md (mulsi3): Remove call to mulsi_600_lib.
4356 (mulsi3_600_lib): Remove pattern.
4357 (umulsi3_highpart_600_lib_le): Likewise.
4358 (umulsi3_highpart): Remove call to umulsi3_highpart_600_lib_le.
4359 (umulsidi3): Remove call to umulsidi3_600_lib.
4360 (umulsidi3_600_lib): Remove pattern.
4361 (peephole2): Remove peephole using the above deprecated patterns.
4363 2018-10-31 Thomas Preud'homme <thomas.preudhomme@linaro.org>
4366 * config/arm/arm.c (arm_option_check_internal): Disable the combined
4367 use of -mslow-flash-data and -mword-relocations.
4368 (arm_option_override): Enable -mword-relocations if -fpic or -fPIC.
4369 * config/arm/arm.md (SYMBOL_REF MOVT splitter): Stop checking for
4371 * doc/invoke.texi (-mword-relocations): Mention conflict with
4373 (-mslow-flash-data): Reciprocally.
4375 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
4377 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Force
4378 16-byte modes held in GP registers to use an even regno.
4380 * config/aarch64/aarch64.c (aarch64_emit_bic): Remove.
4381 (aarch64_atomic_ldop_supported_p): Remove.
4382 (aarch64_gen_atomic_ldop): Remove.
4383 * config/aarch64/atomic.md (atomic_<atomic_optab><ALLI>):
4384 Fully expand LSE operations here.
4385 (atomic_fetch_<atomic_optab><ALLI>): Likewise.
4386 (atomic_<atomic_optab>_fetch<ALLI>): Likewise.
4387 (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse): Drop atomic_op iterator
4388 and use ATOMIC_LDOP instead; use register_operand for the input;
4389 drop the split and emit insns directly.
4390 (aarch64_atomic_fetch_<ATOMIC_LDOP><ALLI>_lse): Likewise.
4391 (aarch64_atomic_<atomic_op>_fetch<ALLI>_lse): Remove.
4392 (@aarch64_atomic_load<ATOMIC_LDOP><ALLI>): Remove.
4394 * config/aarch64/aarch64.c (aarch64_emit_atomic_swap): Remove.
4395 (aarch64_gen_atomic_ldop): Don't call it.
4396 * config/aarch64/atomics.md (atomic_exchange<ALLI>):
4397 Use aarch64_reg_or_zero.
4398 (aarch64_atomic_exchange<ALLI>): Likewise.
4399 (aarch64_atomic_exchange<ALLI>_lse): Remove split; remove & from
4400 operand 0; use aarch64_reg_or_zero for input; merge ...
4401 (@aarch64_atomic_swp<ALLI>): ... this and remove.
4403 * config/aarch64/aarch64.c (aarch64_gen_compare_reg_maybe_ze): New.
4404 (aarch64_split_compare_and_swap): Use it.
4405 (aarch64_expand_compare_and_swap): Likewise. Remove convert_modes;
4406 test oldval against the proper predicate.
4407 * config/aarch64/atomics.md (@atomic_compare_and_swap<ALLI>):
4408 Use nonmemory_operand for expected.
4409 (cas_short_expected_pred): New.
4410 (@aarch64_compare_and_swap<SHORT>): Use it; use "rn" not "rI" to match.
4411 (@aarch64_compare_and_swap<GPI>): Use "rn" not "rI" for expected.
4412 * config/aarch64/predicates.md (aarch64_plushi_immediate): New.
4413 (aarch64_plushi_operand): New.
4415 * config/aarch64/aarch64.c (aarch64_expand_compare_and_swap):
4416 Force oldval into the rval register for TARGET_LSE; emit the compare
4417 during initial expansion so that it may be deleted if unused.
4418 (aarch64_gen_atomic_cas): Remove.
4419 * config/aarch64/atomics.md (@aarch64_compare_and_swap<SHORT>_lse):
4420 Change =&r to +r for operand 0; use match_dup for operand 2;
4421 remove is_weak and mod_f operands as unused. Drop the split
4423 (@aarch64_atomic_cas<SHORT>): ... this pattern's output; remove.
4424 (@aarch64_compare_and_swap<GPI>_lse): Similarly.
4425 (@aarch64_atomic_cas<GPI>): Similarly.
4427 2018-10-31 Richard Biener <rguenther@suse.de>
4429 * tree-eh.c (replace_trapping_overflow): Simplify ABS_EXPR case
4432 2018-10-31 Sameera Deshpande <sameera.deshpande@linaro.org>
4434 * config/aarch64/aarch64-cores.def (saphira): Use saphira pipeline.
4435 * config/aarch64/aarch64.md: Include saphira.md
4436 * config/aarch64/saphira.md: New file for pipeline description.
4438 2018-10-30 Martin Sebor <msebor@redhat.com>
4441 * gimple-ssa-sprintf.c (format_directive): Use %G to include
4443 (sprintf_dom_walker::compute_format_length):
4444 Avoid setting POSUNDER4K here.
4445 (get_destination_size): Handle null argument values.
4446 (get_user_idx_format): New function.
4447 (sprintf_dom_walker::handle_gimple_call): Handle all printf-like
4448 functions, including user-defined with attribute format printf.
4449 Use %G to include inlining context.
4450 Set POSUNDER4K here.
4452 2018-10-30 Jan Hubicka <jh@suse.cz>
4454 * params.def (lto-partitions): Bump from 32 to 128.
4456 2018-10-30 Jan Hubicka <jh@suse.cz>
4459 (free_lang_data_d, add_tree_to_fld_list, fld_worklist_push): Move
4461 (free_lang_data_in_type): Forward declare.
4462 (fld_type_variant_equal_p): New function.
4463 (fld_type_variant): New function
4464 (fld_incomplete_types): New hash.
4465 (fld_incomplete_type_of): New function
4466 (fld_simplfied-type): New function.
4467 (free_lang_data_in_decl): Add fld parameter; simplify type of FIELD_DECL
4468 (free_lang_data): Allocate and free fld_incomplete_type; update call
4469 of free_lang_data_in_decl.
4471 2018-10-30 Eric Botcazou <ebotcazou@adacore.com>
4473 * gcov.c (output_lines): Remove duplicate line.
4475 2018-10-30 Aaron Sawdey <acsawdey@linux.ibm.com>
4477 * config/rs6000/rs6000.md (bswapdi2): Force address into register
4478 if not in indexed or indirect form.
4479 (bswapdi2_load): Change predicate to indexed_or_indirect_operand.
4480 (bswapdi2_store): Ditto.
4481 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem): New
4483 * config/rs6000/rs6000-protos.h (rs6000_force_indexed_or_indirect_mem):
4484 Prototype for helper function.
4486 2018-10-30 Martin Sebor <msebor@redhat.com>
4488 * doc/extend.texi (optimize): Clarify/expand attribute documentation.
4489 (target, pragma GCC optimize, pragma GCC target): Ditto.
4491 2018-10-30 Jonathan Wakely <jwakely@redhat.com>
4493 * doc/extend.texi: Fix prototype and description of
4494 __builtin_expect_with_probability.
4496 2018-10-30 Michael Ploujnikov <michael.ploujnikov@oracle.com>
4498 * cgraph.h (clone_function_name_1): Replaced by new
4499 clone_function_name_numbered that takes name as string; for
4500 privatize_symbol_name_1 use only.
4501 (clone_function_name): Renamed to
4502 clone_function_name_numbered to be explicit about numbering.
4503 (clone_function_name): New two-argument function that does
4504 not number its output.
4505 (clone_function_name): New three-argument function that
4506 takes a number to append to its output.
4507 * cgraphclones.c (duplicate_thunk_for_node):
4508 (clone_function_name_1): Renamed.
4509 (clone_function_name_numbered): Two new functions.
4510 (clone_function_name): Improved documentation.
4511 (cgraph_node::create_virtual_clone): Use clone_function_name_numbered.
4512 * config/rs6000/rs6000.c (make_resolver_func): Ditto.
4513 * final.c (final_scan_insn_1): Use the new clone_function_name
4515 * multiple_target.c (create_dispatcher_calls): Ditto.
4516 (create_target_clone): Ditto.
4517 * omp-expand.c (grid_expand_target_grid_body): Ditto.
4518 * omp-low.c (create_omp_child_function_name): Ditto.
4519 * omp-simd-clone.c (simd_clone_create): Ditto.
4520 * symtab.c (simd_symtab_node::noninterposable_alias): Use the
4521 new clone_function_name without numbering.
4523 2018-10-30 Richard Earnshaw <rearnsha@arm.com>
4525 * alloc-pool.h (base_pool_allocator <TBlockAllocator>::initialize):
4526 Assert that the allocation size is not zero.
4528 2018-10-30 Richard Biener <rguenther@suse.de>
4530 PR tree-optimization/87800
4531 * tree-vect-slp.c (vect_build_slp_tree_2): Reject any
4532 non-induction or reduction PHIs.
4534 2018-10-30 Sameera Deshpande <sameera.deshpande@linaro.org
4536 * config/aarch64/falkor-tag-collision-avoidance.c
4537 (execute_tag_collision_avoidance): Call df_note_add_problem.
4539 2018-10-30 Martin Liska <mliska@suse.cz>
4541 * doc/extend.texi: Fix typo in documentation
4542 of __builtin_expect_with_probability.
4544 2018-10-29 David Malcolm <dmalcolm@redhat.com>
4547 * input.c (get_substring_ranges_for_loc): Detect if
4548 linemap_resolve_location gives us a NULL map, and reject
4551 2018-10-29 Iain Buclaw <ibuclaw@gdcproject.org>
4553 * config.gcc (xstormy16-*-elf): Set tm_d_file.
4555 2018-10-29 David Malcolm <dmalcolm@redhat.com>
4556 Martin Sebor <msebor@redhat.com>
4557 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
4559 * doc/ux.texi (Quoting): New subsection, adapted from material at
4560 https://gcc.gnu.org/wiki/DiagnosticsGuidelines written by
4561 MartinSebor and ManuelLopezIbanez.
4562 (Fix-it hints): Note that fix-it hints shouldn't be marked for
4565 2018-10-29 Kugan Vivekanandarajah <kuganv@linaro.org>
4568 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Fix niter
4571 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
4573 * gcc.target/powerpc/ssse3-check.h: Remove inaccurate comment.
4575 2018-10-29 Paul Koning <ni1d@arrl.net>
4577 * doc/rtl.texi (CONST_WIDE_INT_ELT): Give correct macro name.
4579 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
4581 * gcc/config/rs6000/mmintrin.h (_mm_packs_pi16, _mm_packs_pi32,
4582 _mm_packs_pu16, _mm_unpackhi_pi8, _mm_unpacklo_pi8, _mm_add_pi8,
4583 _mm_add_pi16, _mm_add_pi32, _mm_sub_pi8, _mm_sub_pi16, _mm_sub_pi32,
4584 _mm_cmpgt_pi8, _mm_cmpeq_pi16, _mm_cmpgt_pi16, _mm_cmpeq_pi32,
4585 _mm_cmpgt_pi32, _mm_adds_pi8, _mm_adds_pi16, _mm_adds_pu8,
4586 _mm_adds_pu16, _mm_subs_pi8, _mm_subs_pi16, _mm_subs_pu8,
4587 _mm_subs_pu16, _mm_madd_pi16, _mm_mulhi_pi16, _mm_mullo_pi16,
4588 _mm_sll_pi16, _mm_sra_pi16, _mm_srl_pi16, _mm_set1_pi16, _mm_set1_pi8):
4589 Change 'vector' to '__vector'.
4590 * gcc/config/rs6000/xmmintrin.h (_mm_cvtps_pi32, _mm_cvttps_pi32,
4591 _mm_cvtps_pi16, _mm_cvtps_pi8, _mm_max_pi16, _mm_max_pu8, _mm_min_pi16,
4592 _mm_min_pu8, _mm_mulhi_pu16, _mm_shuffle_pi16, _mm_avg_pu8,
4593 _mm_avg_pu16): Likewise. And, whitespace corrections.
4595 2018-10-29 Richard Biener <rguenther@suse.de>
4597 PR tree-optimization/87785
4598 * tree-vect-slp.c (vect_gather_slp_loads): Only gather
4601 2018-10-29 Olivier Hainque <hainque@adacore.com>
4603 * VXWORKS_NET_LIBS_RTP: New macro, network part of VXWORKS_LIBS_RTP.
4605 2018-10-29 Olivier Hainque <hainque@adacore.com>
4607 * config/vxworks.h (VXWORKS_PERSONALITY): New VxWorks
4608 ports configuration macro, defaults to "gnu".
4609 (VXWORKS_OS_CPP_BUILTINS): Use it. Feed TOOL and
4610 TOOL_FAMILY instead of _VX_TOOL/_VXTOOL_FAMILY.
4612 2018-10-29 Olivier Hainque <hainque@adacore.com>
4614 * config/vxworks.h (CLEAR_INSN_CACHE): #undef before #define.
4616 2018-10-29 Richard Biener <rguenther@suse.de>
4618 PR tree-optimization/87790
4619 * tree-vect-slp.c (vect_mark_slp_stmts): Simplify.
4620 (vect_make_slp_decision): Adjust.
4621 (vect_slp_analyze_bb_1): Likewise.
4622 (vect_detect_hybrid_slp_stmts): Properly union SLP type over
4625 2018-10-29 Richard Biener <rguenther@suse.de>
4627 PR tree-optimization/87785
4628 * tree-vect-slp.c (vect_build_slp_tree_2): Remove loads argument
4630 (vect_build_slp_tree): Likewise.
4631 (vect_gather_slp_loads): New function.
4632 (vect_analyze_slp_instance): Gather loads separately from the
4635 2018-10-29 Martin Liska <mliska@suse.cz>
4637 * Makefile.in: Make dependency to json.o.
4638 * doc/gcov.texi: Document new JSON format, remove
4639 old intermediate format documentation.
4640 * gcov.c (struct function_info): Come up with m_name and
4642 (function_info::function_info): Initialize it.
4643 (function_info::~function_info): Release it.
4644 (main): Rename flag_intermediate_format to flag_json_format.
4645 (print_usage): Describe --json-format.
4646 (process_args): Set flag_json_format.
4647 (output_intermediate_line): Remove.
4648 (output_intermediate_json_line): Likewise.
4649 (get_gcov_intermediate_filename): Return new extension
4651 (output_intermediate_file): Implement JSON emission.
4652 (output_json_intermediate_file): Implement JSON emission.
4653 (generate_results): Use ::get_name for function name.
4654 Handle JSON output file.
4655 (read_graph_file): Use ::get_name instead of cplus_demangle.
4656 (read_count_file): Likewise.
4657 (solve_flow_graph): Likewise.
4658 (add_line_counts): Likewise.
4659 (accumulate_line_counts): Use new flag_json_format.
4660 (output_function_details): Use ::get_name instead of cplus_demangle.
4661 (output_lines): Likewise.
4662 * json.cc (test_writing_literals): Add new tests.
4663 * json.h (class literal): Add new boolean constructor.
4665 2018-10-29 Segher Boessenkool <segher@kernel.crashing.org>
4667 PR rtl-optimization/87701
4668 PR rtl-optimization/87780
4669 * combine.c (make_more_copies): Rewrite.
4671 2018-10-28 Kugan Vivekanandarajah <kuganv@linaro.org>
4673 * doc/generic.texi (ABSU_EXPR): Document.
4674 * match.pd (absu(x)*absu(x) -> x*x): Handle.
4675 (absu(absu(X)) -> absu(X)): Likewise.
4676 (absu(-X) -> absu(X)): Likewise.
4677 (absu(X) where X is nonnegative -> X): Likewise.
4679 2018-10-28 Iain Buclaw <ibuclaw@gdcproject.org>
4681 * Makefile.in (tm_d_file_list, tm_d_include_list): New variables.
4682 (TM_D_H, D_TARGET_DEF, D_TARGET_H, D_TARGET_OBJS): New variables.
4683 (tm_d.h, cs-tm_d.h, default-d.o): New rules.
4684 (d/d-target-hooks-def.h, s-d-target-hooks-def-h): New rules.
4685 (s-tm-texi): Also check timestamp on d-target.def.
4686 (generated_files): Add TM_D_H and d-target-hooks-def.h.
4687 (build/genhooks.o): Also depend on D_TARGET_DEF.
4688 * config.gcc (tm_d_file, d_target_objs, target_has_targetdm): New
4690 * config/aarch64/aarch64-d.c: New file.
4691 * config/aarch64/aarch64-linux.h (GNU_USER_TARGET_D_CRITSEC_SIZE):
4693 * config/aarch64/aarch64-protos.h (aarch64_d_target_versions): New
4695 * config/aarch64/aarch64.h (TARGET_D_CPU_VERSIONS): Define.
4696 * config/aarch64/t-aarch64 (aarch64-d.o): New rule.
4697 * config/arm/arm-d.c: New file.
4698 * config/arm/arm-protos.h (arm_d_target_versions): New prototype.
4699 * config/arm/arm.h (TARGET_D_CPU_VERSIONS): Define.
4700 * config/arm/linux-eabi.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
4701 * config/arm/t-arm (arm-d.o): New rule.
4702 * config/default-d.c: New file.
4703 * config/glibc-d.c: New file.
4704 * config/gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
4705 * config/i386/i386-d.c: New file.
4706 * config/i386/i386-protos.h (ix86_d_target_versions): New prototype.
4707 * config/i386/i386.h (TARGET_D_CPU_VERSIONS): Define.
4708 * config/i386/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
4709 (GNU_USER_TARGET_D_CRITSEC_SIZE): Define.
4710 * config/i386/t-i386 (i386-d.o): New rule.
4711 * config/kfreebsd-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
4712 * config/kopensolaris-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
4713 * config/linux-android.h (ANDROID_TARGET_D_OS_VERSIONS): Define.
4714 * config/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
4715 * config/mips/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
4716 * config/mips/mips-d.c: New file.
4717 * config/mips/mips-protos.h (mips_d_target_versions): New prototype.
4718 * config/mips/mips.h (TARGET_D_CPU_VERSIONS): Define.
4719 * config/mips/t-mips (mips-d.o): New rule.
4720 * config/powerpcspe/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
4721 * config/powerpcspe/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
4722 * config/powerpcspe/powerpcspe-d.c: New file.
4723 * config/powerpcspe/powerpcspe-protos.h (rs6000_d_target_versions):
4725 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
4726 Support GNU D by using 0 as the language type.
4727 * config/powerpcspe/powerpcspe.h (TARGET_D_CPU_VERSIONS): Define.
4728 * config/powerpcspe/t-powerpcspe (powerpcspe-d.o): New rule.
4729 * config/riscv/riscv-d.c: New file.
4730 * config/riscv/riscv-protos.h (riscv_d_target_versions): New
4732 * config/riscv/riscv.h (TARGET_D_CPU_VERSIONS): Define.
4733 * config/riscv/t-riscv (riscv-d.o): New rule.
4734 * config/rs6000/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
4735 * config/rs6000/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
4736 * config/rs6000/rs6000-d.c: New file.
4737 * config/rs6000/rs6000-protos.h (rs6000_d_target_versions): New
4739 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
4740 Support GNU D by using 0 as the language type.
4741 * config/rs6000/rs6000.h (TARGET_D_CPU_VERSIONS): Define.
4742 * config/rs6000/t-rs6000 (rs6000-d.o): New rule.
4743 * config/s390/s390-d.c: New file.
4744 * config/s390/s390-protos.h (s390_d_target_versions): New prototype.
4745 * config/s390/s390.h (TARGET_D_CPU_VERSIONS): Define.
4746 * config/s390/t-s390 (s390-d.o): New rule.
4747 * config/sparc/sparc-d.c: New file.
4748 * config/sparc/sparc-protos.h (sparc_d_target_versions): New
4750 * config/sparc/sparc.h (TARGET_D_CPU_VERSIONS): Define.
4751 * config/sparc/t-sparc (sparc-d.o): New rule.
4752 * config/t-glibc (glibc-d.o): New rule.
4753 * configure: Regenerated.
4754 * configure.ac (tm_d_file): New variable.
4755 (tm_d_file_list, tm_d_include_list, d_target_objs): Add substitutes.
4756 * doc/contrib.texi (Contributors): Add self for the D frontend.
4757 * doc/frontends.texi (G++ and GCC): Mention D as a supported language.
4758 * doc/install.texi (Configuration): Mention libphobos as an option for
4759 --enable-shared. Mention d as an option for --enable-languages.
4760 (Testing): Mention check-d as a target.
4761 * doc/invoke.texi (Overall Options): Mention .d, .dd, and .di as file
4762 name suffixes. Mention d as a -x option.
4763 * doc/sourcebuild.texi (Top Level): Mention libphobos.
4764 * doc/standards.texi (Standards): Add section on D language.
4765 * doc/tm.texi: Regenerated.
4766 * doc/tm.texi.in: Add @node for D language and ABI, and @hook for
4767 TARGET_CPU_VERSIONS, TARGET_D_OS_VERSIONS, and TARGET_D_CRITSEC_SIZE.
4768 * dwarf2out.c (is_dlang): New function.
4769 (gen_compile_unit_die): Use DW_LANG_D for D.
4770 (declare_in_namespace): Return module die for D, instead of adding
4771 extra declarations into the namespace.
4772 (gen_namespace_die): Generate DW_TAG_module for D.
4773 (gen_decl_die): Handle CONST_DECLSs for D.
4774 (dwarf2out_decl): Likewise.
4775 (prune_unused_types_walk_local_classes): Handle DW_tag_interface_type.
4776 (prune_unused_types_walk): Handle DW_tag_interface_type same as other
4777 kinds of aggregates.
4778 * gcc.c (default_compilers): Add entries for .d, .dd and .di.
4779 * genhooks.c: Include d/d-target.def.
4781 2018-10-28 Iain Sandoe <iain@sandoe.co.uk>
4784 * config/rs6000/darwin.h (STACK_BOUNDARY): New.
4785 (RS6000_STARTING_FRAME_OFFSET): Adjust to preserve 16byte alignment.
4786 (STACK_DYNAMIC_OFFSET): Likewise.
4788 2018-10-27 Sandra Loosemore <sandra@codesourcery.com>
4791 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Fix
4794 2018-10-26 Tulio Magno Quites Machado Filho <tuliom@linux.ibm.com>
4796 * doc/extend.texi (PowerPC builtins): Fix __builtin_unpack_ibm128
4797 return type and other typos.
4799 2018-10-26 Aaron Sawdey <acsawdey@linux.ibm.com>
4801 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Change to
4802 a shorter sequence with fewer branches.
4803 (emit_final_str_compare_gpr): Ditto.
4805 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
4807 * config/rs6000/tmmintrin.h: New file.
4808 * config.gcc (powerpc*-*-*): Add tmmintrin.h to extra_headers.
4810 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
4812 * config/rs6000/mmintrin.h: Enable 32bit compilation.
4813 * config/rs6000/xmmintrin.h: Likewise.
4815 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
4817 * config/rs6000/xmmintrin.h (_mm_extract_pi16): Fix for big-endian.
4819 2018-10-26 Richard Biener <rguenther@suse.de>
4821 * tree-vect-slp.c (vect_mark_slp_stmts): Add visited hash_set
4823 (vect_mark_slp_stmts_relevant): Likewise.
4824 (vect_detect_hybrid_slp_stmts): Likewise.
4825 (vect_bb_slp_scalar_cost): Likewise.
4826 (vect_remove_slp_scalar_calls): Likewise.
4828 2018-10-26 Jan Hubicka <jh@suse.cz>
4830 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix recursion.
4831 (warn_types_mismatch): Fix walk of DECL_NAME.
4832 (odr_types_equivalent_p): Fix overactive assert.
4834 2018-10-26 Richard Biener <rguenther@suse.de>
4836 PR tree-optimization/87105
4837 * tree-vectorizer.h (_slp_tree::refcnt): New member.
4838 * tree-vect-slp.c (vect_free_slp_tree): Decrement and honor
4840 (vect_create_new_slp_node): Initialize refcnt to one.
4842 (scalar_stmts_set_t, bst_fail): Remove.
4843 (vect_build_slp_tree_2): Add bst_map argument and adjust calls.
4844 (vect_build_slp_tree): Add bst_map argument and lookup
4845 already created SLP nodes.
4846 (vect_print_slp_tree): Handle a SLP graph, print SLP node
4848 (vect_slp_rearrange_stmts): Handle a SLP graph.
4849 (vect_analyze_slp_instance): Adjust and free SLP nodes from
4850 the CSE map. Fix indenting.
4851 (vect_schedule_slp_instance): Add short-cut.
4853 2018-10-26 Martin Liska <mliska@suse.cz>
4856 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Hash with
4857 addr_expr and not with pointers.
4859 2018-10-26 Jan Hubicka <jh@suse.cz>
4861 * tree.c (free_lang_data_in_type): Only check main variants.
4862 * ipa-devirt.c (warn_odr): Make static.
4863 (types_same_for_odr): Drop strict variant.
4864 (types_odr_comparable): Likewise.
4865 (odr_or_derived_type_p): Look for main variants.
4866 (odr_name_hasher::equal): Cleanup comment.
4867 (odr_subtypes_equivalent): Add warn and warned arguments; check main
4869 (type_variants_equivalent_p): break out from ...
4870 (odr_types_equivalent): ... here; go for main variants where needed.
4871 (warn_odr): ... here; turn static.
4872 (warn_types_mismatch): Compare mangled names of main variants.
4873 * ipa-utils.h (types_odr_comparable): Drop strict parameter.
4874 (type_with_linkage_p): Sanity check that we look at main variant.
4875 * lto.c (lto_read_decls): Only consider main variant to be ODR type.
4876 * tree.h (types_same_for_odr): Drop strict argument.
4878 2018-10-26 Richard Biener <rguenther@suse.de>
4880 PR tree-optimization/87746
4881 * tree-vect-data-refs.c (vect_update_misalignment_for_peel):
4882 Simplify and fix WRT strided store groups with size not
4883 equal to step in element count.
4884 (vect_analyze_group_access_1): Dump the whole group.
4886 2018-10-25 Carl Love <cel@us.ibm.com>
4888 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_VSCEDPGT,
4889 P9V_BUILTIN_VEC_VSCEDPLT, P9V_BUILTIN_VEC_VSCEDPEQ,
4890 P9V_BUILTIN_VEC_VSCEDPUO): Rename base overloaded name. Add quad
4891 precicion entry for each overloaded builtin.
4892 * config/rs6000/rs6000-builtin.def (VSCEDPGT, VSCEDPLT, VSCEDPEQ,
4893 VSCEDPUO): Rename overloaded name.
4894 (VSCEDPGT, VSCEQPGT, VSCEDPLT, VSCEQPLT, VSCEDPEQ, VSCEQPEQ,
4895 VSCEDPUO, VSCEQPUO): Add defitions for overloaded builtins.
4896 * config/rs6000/vsx.md (xscmpexpqp_<code>_<mode>): Add
4897 define_expand for xscmpexqp instruction.
4898 (*xscmpexpqp): Add define_insn for the xscmpexqp instruction.
4900 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
4901 Jinsong Ji <jji@us.ibm.com>
4903 * config/rs6000/emmintrin.h (_mm_slli_epi16): Replace deprecated
4904 function with vec_sl.
4905 (_mm_slli_epi32): Likewise.
4906 (_mm_slli_epi64): Likewise.
4907 (_mm_srai_epi16): Replace deprecated function with vec_sra.
4908 (_mm_srai_epi32): Likewise.
4909 (_mm_srli_epi16): Replace deprecated function with vec_sr.
4910 (_mm_srli_epi32): Likewise.
4911 (_mm_srli_epi64): Likewise.
4912 (_mm_sll_epi16): Replace deprecated function with vec_sl.
4913 (_mm_sll_epi32): Likewise.
4914 (_mm_sll_epi64): Likewise.
4915 (_mm_sra_epi16): Replace deprecated function with vec_sra.
4916 (_mm_sra_epi32): Likewise.
4917 (_mm_srl_epi16): Replace deprecated function with vec_sr.
4918 (_mm_srl_epi32): Likewise.
4919 (_mm_srl_epi64): Likewise.
4921 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
4922 Jinsong Ji <jji@us.ibm.com>
4924 * gcc/config/rs6000/emmintrin.h (_mm_sll_epi16): Replace
4925 comparison operators with vec_cmp* for compatibility due to
4926 unfortunate history; clean up formatting and use types more
4928 (_mm_sll_epi32): Likewise.
4929 (_mm_sll_epi64): Likewise.
4930 (_mm_srl_epi16): Likewise.
4931 (_mm_srl_epi32): Likewise.
4932 (_mm_srl_epi64): Likewise.
4934 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
4935 Jinsong Ji <jji@us.ibm.com>
4937 * config/rs6000/emmintrin.h (_mm_sll_epi64): Remove wrong cast.
4938 * config/rs6000/xmmintrin.h (_mm_min_ps): Change m's type to
4939 __vector __bool int. Use vec_cmpgt in preference to deprecated
4940 function vec_vcmpgtfp.
4941 (_mm_max_ps): Likewise.
4943 2018-10-25 Jeff Law <law@redhat.com>
4945 * config/rl78/rl78.c (insn_ok_now): Always re-recognize the insn
4948 2018-10-25 Martin Sebor <msebor@redhat.com>
4950 * doc/extend.texi (aligned): Expand attribute description.
4951 (Alignment): Rename section. Discuss function arguments.
4953 2018-10-25 Jan Hubicka <jh@suse.cz>
4955 * ipa-devirt.c (main_odr_variant): Remove.
4956 (hash_odr_name, types_same_for_odr, types_odr_comparable,
4957 odr_name_hasher::equal, odr_subtypes_equivalent_p):
4958 Drop use of main_odr_variant.
4959 (add_type_duplicate): Silence confused warnings on integer types.
4960 (get_odr_type): Always look for main variant.
4961 (register_odr_type): Simplify.
4963 2018-10-25 Richard Biener <rguenther@suse.de>
4965 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses):
4966 Initialize ng to silence error with release checking bootstrap.
4968 2018-10-25 Richard Biener <rguenther@suse.de>
4970 * tree-if-conv.c: Include tree-ssa-sccvn.h.
4971 (tree_if_conversion): Run CSE on the if-converted loop body.
4973 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
4975 * config/s390/constraints.md (ZL): New constraint.
4976 * config/s390/s390.c (legitimate_pic_operand_p): Accept LARL
4978 * config/s390/s390.md (movdi_larl): Remove.
4979 (movdi_64): Add the LARL alternative.
4981 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
4984 * rtl.c (RTX_CODE_HWINT_P_1): New helper macro.
4985 (RTX_CODE_HWINT_P): New macro.
4986 (rtx_code_size): Use RTX_CODE_HWINT_P ().
4988 2018-10-25 Jan Hubicka <jh@suse.cz>
4990 * ipa-devirt.c (odr_types_equivalent_p): Do not ICE if one of types
4993 2018-10-25 Richard Biener <rguenther@suse.de>
4995 PR tree-optimization/87665
4996 PR tree-optimization/87745
4997 * tree-vectorizer.h (get_earlier_stmt): Remove.
4998 (get_later_stmt): Pick up UID from the original non-pattern stmt.
5000 2018-10-25 Sam Tebbs <sam.tebbs@arm.com>
5002 * options.texi (Deprecated): Move list to Var section.
5004 2018-10-24 Bill Schmidt <wschmidt@linux.ibm.com>
5005 Jinsong Ji <jji@us.ibm.com>
5007 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Change deprecated
5008 __vector long to __vector long long.
5009 (_mm_cvtpd_ps): Likewise.
5010 (_mm_cvttpd_epi32): Likewise.
5011 (_mm_cvtpi32_pd): Likewise.
5012 (_mm_unpackhi_epi64): Likewise.
5013 (_mm_unpacklo_epi64): Likewise.
5015 2018-10-24 Segher Boessenkool <segher@kernel.crashing.org>
5017 PR rtl-optimization/87720
5018 * combine.c (make_more_copies): Skip if the dest is pc_rtx.
5020 2018-10-24 Alexandre Oliva <aoliva@redhat.com>
5022 * gimple-ssa-isolate-paths.c
5023 (find_implicit_erroneous_behavior): Do not change code if the
5024 pass is running for warnings only.
5025 (find_explicit_erroneous_behavior): Likewise.
5027 2018-10-24 Michael Meissner <meissner@linux.ibm.com>
5029 * config/rs6000/rs6000.c (TARGET_MANGLE_DECL_ASSEMBLER_NAME):
5030 Define as rs6000_mangle_decl_assembler_name.
5031 (rs6000_mangle_decl_assembler_name): If the user switched from IBM
5032 long double to IEEE long double, switch the names of the long
5033 double built-in functions to be <func>f128 instead of <func>l.
5035 2018-10-24 Martin Sebor <msebor@redhat.com>
5037 * doc/extend.texi (nonnull): List no-argument form. Reference
5038 -fno-delete-null-pointer-checks and -fisolate-erroneous-paths-attribute.
5040 2018-10-24 Richard Biener <rguenther@suse.de>
5042 * tree-ssa-sccvn.c (do_rpo_vn): Free rpo_state.
5044 2018-10-24 Martin Liska <mliska@suse.cz>
5046 PR tree-optimization/84436
5047 * tree-switch-conversion.c (switch_conversion::contains_same_values_p):
5049 (switch_conversion::contains_linear_function_p): New.
5050 (switch_conversion::build_one_array): Support linear
5051 transformation on input.
5052 * tree-switch-conversion.h (struct switch_conversion): Add
5053 contains_linear_function_p declaration.
5055 2018-10-24 Richard Biener <rguenther@suse.de>
5057 * varasm.c (const_hash_1): Return hash of ADDR_EXPR
5058 if its argument is CONSTANT_CLASS_P.
5060 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
5062 * ipa-utils.h (type_with_linkage_p): No longer check for TYPE_STUB_DECL;
5063 it is wrong for forward declarations.
5065 2018-10-24 Ilya Leoshkevich <iii@linux.ibm.com>
5067 * config/s390/s390.c (s390_check_qrst_address): Add the missing
5068 SYMBOL_REF_P () check.
5070 2018-10-24 Richard Biener <rguenther@suse.de>
5072 PR tree-optimization/87105
5073 * tree-vect-data-refs.c (vect_analyze_group_access_1): Adjust
5074 dump classification.
5075 (vect_analyze_data_ref_accesses): Handle duplicate loads and
5076 stores by splitting the affected group after the fact.
5077 * tree-vect-slp.c (vect_build_slp_tree_2): Dump when we
5078 fail the SLP build because of size constraints.
5080 2018-10-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5082 * configure.ac (gcc_cv_ld_aligned_shf_merge): New test.
5083 * configure: Regenerate.
5084 * config.in: Regenerate.
5085 * varasm.c (mergeable_string_section): Use readonly_data_section
5086 if linker doesn't support SHF_MERGE with alignment > 8.
5087 (mergeable_constant_section): Likewise.
5089 2018-10-24 Richard Biener <rguenther@suse.de>
5091 PR tree-optimization/84013
5092 * tree-ssa-structalias.c (struct msdi_data): New struct for
5093 marshalling data to walk_stmt_load_store_ops.
5094 (maybe_set_dependence_info): Refactor as callback for
5095 walk_stmt_load_store_ops.
5096 (compute_dependence_clique): Set restrict info on all stmt kinds.
5098 2018-10-24 Martin Liska <mliska@suse.cz>
5100 * cgraph.c (cgraph_node::dump):
5101 Remove reduntant dumps and make tp_first_run dump more compact.
5103 2018-10-24 Richard Biener <rguenther@suse.de>
5105 PR tree-optimization/87665
5106 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Adjust
5109 2018-10-12 Jeff Law <law@redhat.com>
5111 * config/h8300/h8300.c (h8300_expand_prologue): Fix stm generation
5114 2018-10-23 Richard Biener <rguenther@suse.de>
5116 * tree-vrp.c (add_assert_info): Guard dump_printf with
5118 * gimple-ssa-evrp-analyze.c
5119 (evrp_range_analyzer::record_ranges_from_incoming_edge):
5120 Use value_range::ignore_equivs_equal_p.
5122 2018-10-23 Richard Biener <rguenther@suse.de>
5124 PR tree-optimization/87105
5125 PR tree-optimization/87608
5126 * passes.def (pass_all_early_optimizations): Add early phi-opt
5128 * tree-ssa-phiopt.c (value_replacement): Ignore NOPs and predicts in
5129 addition to debug stmts.
5130 (tree_ssa_phiopt_worker): Add early_p argument, do only min/max
5131 and abs replacement early.
5132 * tree-cfg.c (gimple_empty_block_p): Likewise.
5134 2018-10-23 Richard Earnshaw <rearnsha@arm.com>
5137 * config.gcc (arm*-*-netbsdelf*): Default to StrongARM if no CPU
5138 specified to configure.
5139 (arm*-*-*): Use ARM7TDMI as the target CPU if no default provided.
5141 2018-10-23 Richard Biener <rguenther@suse.de>
5143 PR tree-optimization/87700
5144 * tree-ssa-copy.c (set_copy_of_val): Fix change detection logic.
5146 2018-10-23 Jakub Jelinek <jakub@redhat.com>
5149 * config/i386/avx512vlintrin.h (_mm_mask_mullo_epi32): Change type of
5150 second argument from __mmask16 to __mmask8.
5151 * config/i386/avx512vlbwintrin.h (_mm_mask_packus_epi32,
5152 _mm_mask_packs_epi32): Likewise.
5153 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i64scatter_ps):
5155 (_mm512_mask_prefetch_i64scatter_pd): Likewise. Formatting fix.
5157 2018-10-23 Richard Biener <rguenther@suse.de>
5159 * tree-vect-stmts.c (vect_analyze_stmt): Fix typo in comment.
5161 2018-10-23 Richard Biener <rguenther@suse.de>
5163 PR tree-optimization/86144
5164 * tree-vect-stmts.c (vect_analyze_stmt): Prefer -mveclibabi
5165 over simd attribute.
5167 2018-10-23 Richard Biener <rguenther@suse.de>
5169 PR tree-optimization/87693
5170 * tree-ssa-threadedge.c (thread_around_empty_blocks): Handle
5171 the case we do not find the taken edge.
5173 2018-10-22 Bill Schmidt <wschmidt@linux.ibm.com>
5174 Jinsong Ji <jji@us.ibm.com>
5176 * config/rs6000/emmintrin.h (_MM_SHUFFLE2): Comment cleanup.
5177 (_mm_store_pd): Use unaligned vector type for pointer cast.
5178 (_mm_maskmoveu_si128): Likewise.
5179 * config/rs6000/xmmintrin.h (__m128_u): New typedef.
5180 (_mm_store_ps): Use unaligned vector type for pointer cast.
5182 2018-10-22 Paul Koning <ni1d@arrl.net>
5184 * symtab.c (symtab_node::increase_alignment): Correct max
5187 2018-10-22 Yury Gribov <tetra2005@gmail.com>
5189 PR tree-optimization/87633
5190 * match.pd: Do not generate unordered integer comparisons.
5192 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
5194 PR rtl-optimization/87600
5195 * combine.c: Add include of expr.h.
5196 (cant_combine_insn_p): Do not combine moves from any hard non-fixed
5197 register to a pseudo.
5198 (make_more_copies): New function, add a copy to a new pseudo after
5199 the moves from hard registers into pseudos.
5200 (rest_of_handle_combine): Declare rebuild_jump_labels_after_combine
5201 later. Call make_more_copies.
5203 2018-10-22 Andrew Stubbs <ams@codesourcery.com>
5205 * lra-constraints.c (process_alt_operands): New local array,
5206 matching_early_clobber. Check matching_early_clobber before
5207 decrementing reject, and set matching_early_clobber after.
5209 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
5212 * config/rs6000/rs6000.c (print_operand_address): For unexpected RTL
5213 call output_addr_const and hope for the best.
5215 2018-10-22 Richard Biener <rguenther@suse.de>
5217 * gimple-ssa-evrp-analyze.c
5218 (evrp_range_analyzer::record_ranges_from_incoming_edge): Be
5219 smarter about what ranges to use.
5220 * tree-vrp.c (add_assert_info): Dump here.
5221 (register_edge_assert_for_2): Instead of here at multiple but
5224 * gcc.dg/tree-ssa/evrp12.c: New testcase.
5225 * gcc.dg/predict-6.c: Adjust.
5226 * gcc.dg/tree-ssa/vrp33.c: Disable EVRP.
5227 * gcc.dg/tree-ssa/vrp02.c: Likewise.
5228 * gcc.dg/tree-ssa/cunroll-9.c: Likewise.
5230 2018-10-22 Steven Bosscher <steven@gcc.gnu.org>
5231 Richard Biener <rguenther@suse.de>
5233 * bitmap.h: Update data structure documentation, including a
5234 description of bitmap views as either linked-lists or splay trees.
5235 (struct bitmap_element_def): Update comments for splay tree bitmaps.
5236 (struct bitmap_head_def): Likewise.
5237 (bitmap_list_view, bitmap_tree_view): New prototypes.
5238 (bitmap_initialize_stat): Initialize a bitmap_head's indx and
5240 (bmp_iter_set_init): Assert the iterated bitmaps are in list form.
5241 (bmp_iter_and_init, bmp_iter_and_compl_init): Likewise.
5242 * bitmap.c (bitmap_elem_to_freelist): Unregister overhead of a
5243 released bitmap element here.
5244 (bitmap_element_free): Remove.
5245 (bitmap_elt_clear_from): Work on splay tree bitmaps.
5246 (bitmap_list_link_element): Renamed from bitmap_element_link. Move
5247 this function similar ones such that linked-list bitmap implementation
5248 functions are grouped.
5249 (bitmap_list_unlink_element): Renamed from bitmap_element_unlink,
5250 and moved for grouping.
5251 (bitmap_list_insert_element_after): Renamed from
5252 bitmap_elt_insert_after, and moved for grouping.
5253 (bitmap_list_find_element): New function spliced from bitmap_find_bit.
5254 (bitmap_tree_link_left, bitmap_tree_link_right,
5255 bitmap_tree_rotate_left, bitmap_tree_rotate_right, bitmap_tree_splay,
5256 bitmap_tree_link_element, bitmap_tree_unlink_element,
5257 bitmap_tree_find_element): New functions for splay-tree bitmap
5259 (bitmap_element_link, bitmap_element_unlink, bitmap_elt_insert_after):
5260 Renamed and moved, see above entries.
5261 (bitmap_tree_listify_from): New function to convert part of a splay
5262 tree bitmap to a linked-list bitmap.
5263 (bitmap_list_view): Convert a splay tree bitmap to linked-list form.
5264 (bitmap_tree_view): Convert a linked-list bitmap to splay tree form.
5265 (bitmap_find_bit): Remove.
5266 (bitmap_clear, bitmap_clear_bit, bitmap_set_bit,
5267 bitmap_single_bit_set_p, bitmap_first_set_bit, bitmap_last_set_bit):
5268 Handle splay tree bitmaps.
5269 (bitmap_copy, bitmap_count_bits, bitmap_and, bitmap_and_into,
5270 bitmap_elt_copy, bitmap_and_compl, bitmap_and_compl_into,
5271 bitmap_compl_and_into, bitmap_elt_ior, bitmap_ior, bitmap_ior_into,
5272 bitmap_xor, bitmap_xor_into, bitmap_equal_p, bitmap_intersect_p,
5273 bitmap_intersect_compl_p, bitmap_ior_and_compl,
5274 bitmap_ior_and_compl_into, bitmap_set_range, bitmap_clear_range,
5275 bitmap_hash): Reject trying to act on splay tree bitmaps. Make
5276 corresponding changes to use linked-list specific bitmap_element
5277 manipulation functions as applicable for efficiency.
5278 (bitmap_tree_to_vec): New function.
5279 (debug_bitmap_elt_file): New function split out from ...
5280 (debug_bitmap_file): ... here. Handle splay tree bitmaps.
5281 (bitmap_print): Likewise.
5283 PR tree-optimization/63155
5284 * tree-ssa-propagate.c (ssa_prop_init): Use tree-view for the
5286 * tree-ssa-coalesce.c (coalesce_ssa_name): Populate used_in_copies
5289 2018-10-22 Martin Liska <mliska@suse.cz>
5291 PR tree-optimization/87686
5293 2018-08-29 Martin Liska <mliska@suse.cz>
5295 * tree-switch-conversion.c (switch_conversion::expand):
5296 Strenghten assumption about gswitch statements.
5298 2018-10-22 Martin Liska <mliska@suse.cz>
5300 * ipa-icf.c (sem_item::compare_attributes): Remove.
5301 (sem_item::compare_referenced_symbol_properties): Use
5302 attribute_list_equal instead.
5303 (sem_function::equals_wpa): Likewise.
5304 * ipa-icf.h: Remove compare_attributes.
5306 2018-10-22 Richard Biener <rguenther@suse.de>
5309 * mem-stats.h (mem_usage::operator==): Fix pasto.
5311 2018-10-22 Richard Biener <rguenther@suse.de>
5313 PR tree-optimization/87640
5314 * tree-vrp.c (set_value_range_with_overflow): Decompose
5316 (extract_range_from_binary_expr_1): Adjust.
5318 2018-10-22 Martin Jambor <mjambor@suse.cz>
5320 * tree-eh.h (stmt_could_throw_p): Add function parameter.
5321 (stmt_can_throw_external): Likewise.
5322 (stmt_can_throw_internal): Likewise.
5323 * tree-eh.c (lower_eh_constructs_2): Pass cfun to stmt_could_throw_p.
5324 (lower_eh_constructs_2): Likewise.
5325 (stmt_could_throw_p): Add fun parameter, use it instead of cfun.
5326 (stmt_can_throw_external): Likewise.
5327 (stmt_can_throw_internal): Likewise.
5328 (maybe_clean_eh_stmt_fn): Pass cfun to stmt_could_throw_p.
5329 (maybe_clean_or_replace_eh_stmt): Pass cfun to stmt_could_throw_p.
5330 (maybe_duplicate_eh_stmt_fn): Pass new_fun to stmt_could_throw_p.
5331 (maybe_duplicate_eh_stmt): Pass cfun to stmt_could_throw_p.
5332 (pass_lower_eh_dispatch::execute): Pass cfun to
5333 stmt_can_throw_external.
5334 (cleanup_empty_eh): Likewise.
5335 (verify_eh_edges): Pass cfun to stmt_could_throw_p.
5336 * cgraph.c (cgraph_edge::set_call_stmt): Pass a function to
5337 stmt_can_throw_external instead of pushing it to cfun.
5338 (symbol_table::create_edge): Likewise.
5339 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Pass cfun to
5340 stmt_can_throw_internal.
5341 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Pass cfun
5342 to stmt_could_throw_p.
5343 * gimple-ssa-store-merging.c (handled_load): Pass cfun to
5344 stmt_can_throw_internal.
5345 (pass_store_merging::execute): Likewise.
5346 * gimple-ssa-strength-reduction.c
5347 (find_candidates_dom_walker::before_dom_children): Pass cfun to
5349 * gimplify-me.c (gimple_regimplify_operands): Pass cfun to
5350 stmt_can_throw_internal.
5351 * ipa-pure-const.c (check_call): Pass cfun to stmt_could_throw_p and
5352 to stmt_can_throw_external.
5353 (check_stmt): Pass cfun to stmt_could_throw_p.
5354 (check_stmt): Pass cfun to stmt_can_throw_external.
5355 (pass_nothrow::execute): Likewise.
5356 * trans-mem.c (expand_call_tm): Pass cfun to stmt_can_throw_internal.
5357 * tree-cfg.c (is_ctrl_altering_stmt): Pass cfun to
5358 stmt_can_throw_internal.
5359 (verify_gimple_in_cfg): Pass cfun to stmt_could_throw_p.
5360 (stmt_can_terminate_bb_p): Pass cfun to stmt_can_throw_external.
5361 (gimple_purge_dead_eh_edges): Pass cfun to stmt_can_throw_internal.
5362 * tree-complex.c (expand_complex_libcall): Pass cfun to
5363 stmt_could_throw_p and to stmt_can_throw_internal.
5364 (expand_complex_multiplication): Pass cfun to stmt_can_throw_internal.
5365 * tree-inline.c (copy_edges_for_bb): Likewise.
5366 (maybe_move_debug_stmts_to_successors): Likewise.
5367 * tree-outof-ssa.c (ssa_is_replaceable_p): Pass cfun to
5369 * tree-parloops.c (oacc_entry_exit_ok_1): Likewise.
5370 * tree-sra.c (scan_function): Pass cfun to stmt_can_throw_external.
5371 * tree-ssa-alias.c (stmt_kills_ref_p): Pass cfun to
5372 stmt_can_throw_internal.
5373 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Likewise.
5374 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Pass cfun to
5376 (mark_aliased_reaching_defs_necessary_1): Pass cfun to
5377 stmt_can_throw_internal.
5378 * tree-ssa-forwprop.c (pass_forwprop::execute): Likewise.
5379 * tree-ssa-loop-im.c (movement_possibility): Pass cfun to
5381 * tree-ssa-loop-ivopts.c (find_givs_in_stmt_scev): Likewise.
5382 (add_autoinc_candidates): Pass cfun to stmt_can_throw_internal.
5383 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
5384 (convert_mult_to_fma_1): Likewise.
5385 (convert_to_divmod): Likewise.
5386 * tree-ssa-phiprop.c (propagate_with_phi): Likewise.
5387 * tree-ssa-pre.c (compute_avail): Pass cfun to stmt_could_throw_p.
5388 * tree-ssa-propagate.c
5389 (substitute_and_fold_dom_walker::before_dom_children): Likewise.
5390 * tree-ssa-reassoc.c (suitable_cond_bb): Likewise.
5391 (maybe_optimize_range_tests): Likewise.
5392 (linearize_expr_tree): Likewise.
5393 (reassociate_bb): Likewise.
5394 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
5395 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Likewise.
5396 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
5397 (handle_char_store): Likewise.
5398 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Pass cfun to
5399 stmt_can_throw_internal.
5400 * tree-vect-patterns.c (check_bool_pattern): Pass cfun to
5402 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
5403 (vectorizable_call): Pass cfun to stmt_can_throw_internal.
5404 (vectorizable_simd_clone_call): Likewise.
5405 * value-prof.c (gimple_ic): Pass cfun to stmt_could_throw_p.
5406 (gimple_stringop_fixed_value): Likewise.
5408 2018-10-22 Ilya Leoshkevich <iii@linux.ibm.com>
5410 * config/s390/s390.c (s390_loadrelative_operand_p): Accept
5411 literal pool references.
5412 (s390_check_qrst_address): Adapt to the new behavior of
5413 s390_loadrelative_operand_p ().
5415 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
5418 * config/i386/sse.md (*andnot<mode>3_bcst): New.
5420 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
5423 * config/i386/sse.md (*<code><mode>3_bcst): New.
5425 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
5428 * config/i386/sse.md (avx512bcst): Updated for V4SI, V2DI, V8SI,
5429 V4DI, V16SI and V8DI.
5430 (*sub<mode>3<mask_name>_bcst): New.
5431 (*add<mode>3<mask_name>_bcst): Likewise.
5433 2018-10-21 Bill Schmidt <wschmidt@linux.ibm.com>
5434 Jinsong Ji <jji@us.ibm.com>
5436 * config/rs6000/emmintrin.h (_mm_movemask_pd): Replace __vector
5437 __m64 with __vector unsigned long long for compatibility.
5438 (_mm_movemask_epi8): Likewise.
5439 * config/rs6000/xmmintrin.h (_mm_cvtps_pi32): Likewise.
5440 (_mm_cvttps_pi32): Likewise.
5441 (_mm_cvtpi32_ps): Likewise.
5442 (_mm_cvtps_pi16): Likewise.
5443 (_mm_loadh_pi): Likewise.
5444 (_mm_storeh_pi): Likewise.
5445 (_mm_movehl_ps): Likewise.
5446 (_mm_movelh_ps): Likewise.
5447 (_mm_loadl_pi): Likewise.
5448 (_mm_storel_pi): Likewise.
5449 (_mm_movemask_ps): Likewise.
5450 (_mm_shuffle_pi16): Likewise.
5452 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
5455 * config/i386/avx512fintrin.h (_mm512_fnmsub_round_pd): Use
5456 __builtin_ia32_vfnmsubpd512_mask.
5457 (_mm512_mask_fnmsub_round_pd): Likewise.
5458 (_mm512_fnmsub_pd): Likewise.
5459 (_mm512_mask_fnmsub_pd): Likewise.
5460 (_mm512_maskz_fnmsub_round_pd): Use
5461 __builtin_ia32_vfnmsubpd512_maskz.
5462 (_mm512_maskz_fnmsub_pd): Likewise.
5463 (_mm512_fnmsub_round_ps): Use __builtin_ia32_vfnmsubps512_mask.
5464 (_mm512_mask_fnmsub_round_ps): Likewise.
5465 (_mm512_fnmsub_ps): Likewise.
5466 (_mm512_mask_fnmsub_ps): Likewise.
5467 (_mm512_maskz_fnmsub_round_ps): Use
5468 __builtin_ia32_vfnmsubps512_maskz.
5469 (_mm512_maskz_fnmsub_ps): Likewise.
5470 * config/i386/avx512vlintrin.h (_mm256_mask_fnmsub_pd): Use
5471 __builtin_ia32_vfnmsubpd256_mask.
5472 (_mm256_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256_maskz.
5473 (_mm_mask_fnmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
5474 (_mm_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd128_maskz.
5475 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
5476 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
5477 (_mm256_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_maskz.
5478 (_mm_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_mask.
5479 (_mm_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_maskz.
5480 * config/i386/fmaintrin.h (_mm_fnmsub_pd): Use
5481 __builtin_ia32_vfnmsubpd.
5482 (_mm256_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256.
5483 (_mm_fnmsub_ps): Use __builtin_ia32_vfnmsubps.
5484 (_mm256_fnmsub_ps): Use __builtin_ia32_vfnmsubps256.
5485 (_mm_fnmsub_sd): Use __builtin_ia32_vfnmsubsd3.
5486 (_mm_fnmsub_ss): Use __builtin_ia32_vfnmsubss3.
5487 * config/i386/i386-builtin.def: Add
5488 __builtin_ia32_vfnmsubpd256_mask,
5489 __builtin_ia32_vfnmsubpd256_maskz,
5490 __builtin_ia32_vfnmsubpd128_mask,
5491 __builtin_ia32_vfnmsubpd128_maskz,
5492 __builtin_ia32_vfnmsubps256_mask,
5493 __builtin_ia32_vfnmsubps256_maskz,
5494 __builtin_ia32_vfnmsubps128_mask,
5495 __builtin_ia32_vfnmsubps128_maskz,
5496 __builtin_ia32_vfnmsubpd512_mask,
5497 __builtin_ia32_vfnmsubpd512_maskz,
5498 __builtin_ia32_vfnmsubps512_mask,
5499 __builtin_ia32_vfnmsubps512_maskz, __builtin_ia32_vfnmsubss3,
5500 __builtin_ia32_vfnmsubsd3, __builtin_ia32_vfnmsubps,
5501 __builtin_ia32_vfnmsubpd, __builtin_ia32_vfnmsubps256 and.
5502 __builtin_ia32_vfnmsubpd256.
5503 * config/i386/sse.md (fma4i_fnmsub_<mode>): New.
5504 (<avx512>_fnmsub_<mode>_maskz<round_expand_name>): Likewise.
5505 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_1):
5507 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_2):
5509 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_3):
5511 (fmai_vmfnmsub_<mode><round_name>): Likewise.
5513 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
5516 * config/i386/avx512fintrin.h (_mm512_fnmadd_round_pd): Use
5517 __builtin_ia32_vfnmaddpd512_mask.
5518 (_mm512_mask_fnmadd_round_pd): Likewise.
5519 (_mm512_fnmadd_pd): Likewise.
5520 (_mm512_mask_fnmadd_pd): Likewise.
5521 (_mm512_maskz_fnmadd_round_pd): Use
5522 __builtin_ia32_vfnmaddpd512_maskz.
5523 (_mm512_maskz_fnmadd_pd): Likewise.
5524 (_mm512_fnmadd_round_ps): Use __builtin_ia32_vfnmaddps512_mask.
5525 (_mm512_mask_fnmadd_round_ps): Likewise.
5526 (_mm512_fnmadd_ps): Likewise.
5527 (_mm512_mask_fnmadd_ps): Likewise.
5528 (_mm512_maskz_fnmadd_round_ps): Use
5529 __builtin_ia32_vfnmaddps512_maskz.
5530 (_mm512_maskz_fnmadd_ps): Likewise.
5531 * config/i386/avx512vlintrin.h (_mm256_mask_fnmadd_pd): Use
5532 __builtin_ia32_vfnmaddpd256_mask.
5533 (_mm256_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256_maskz.
5534 (_mm_mask_fnmadd_pd): Use __builtin_ia32_vfmaddpd128_mask
5535 (_mm_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd128_maskz.
5536 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
5537 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
5538 (_mm256_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_maskz.
5539 (_mm_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_mask.
5540 (_mm_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_maskz.
5541 * config/i386/fmaintrin.h (_mm_fnmadd_pd): Use
5542 __builtin_ia32_vfnmaddpd.
5543 (_mm256_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256.
5544 (_mm_fnmadd_ps): Use __builtin_ia32_vfnmaddps.
5545 (_mm256_fnmadd_ps): Use __builtin_ia32_vfnmaddps256.
5546 (_mm_fnmadd_sd): Use __builtin_ia32_vfnmaddsd3.
5547 (_mm_fnmadd_ss): Use __builtin_ia32_vfnmaddss3.
5548 * config/i386/i386-builtin.def: Add
5549 __builtin_ia32_vfnmaddpd256_mask,
5550 __builtin_ia32_vfnmaddpd256_maskz,
5551 __builtin_ia32_vfnmaddpd128_mask,
5552 __builtin_ia32_vfnmaddpd128_maskz,
5553 __builtin_ia32_vfnmaddps256_mask,
5554 __builtin_ia32_vfnmaddps256_maskz,
5555 __builtin_ia32_vfnmaddps128_mask,
5556 __builtin_ia32_vfnmaddps128_maskz,
5557 __builtin_ia32_vfnmaddpd512_mask,
5558 __builtin_ia32_vfnmaddpd512_maskz,
5559 __builtin_ia32_vfnmaddps512_mask,
5560 __builtin_ia32_vfnmaddps512_maskz, __builtin_ia32_vfnmaddss3,
5561 __builtin_ia32_vfnmaddsd3, __builtin_ia32_vfnmaddps,
5562 __builtin_ia32_vfnmaddpd, __builtin_ia32_vfnmaddps256 and.
5563 __builtin_ia32_vfnmaddpd256.
5564 * config/i386/sse.md (fma4i_fnmadd_<mode>): New.
5565 (<avx512>_fnmadd_<mode>_maskz<round_expand_name>): Likewise.
5566 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_1):
5568 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_2):
5570 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_3):
5572 (fmai_vmfnmadd_<mode><round_name>): Likewise.
5574 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
5577 * config/i386/avx512fintrin.h (_mm512_fmsub_round_pd): Use
5578 __builtin_ia32_vfmsubpd512_mask.
5579 (_mm512_mask_fmsub_round_pd): Likewise.
5580 (_mm512_fmsub_pd): Likewise.
5581 (_mm512_mask_fmsub_pd): Likewise.
5582 (_mm512_maskz_fmsub_round_pd): Use
5583 __builtin_ia32_vfmsubpd512_maskz.
5584 (_mm512_maskz_fmsub_pd): Likewise.
5585 (_mm512_fmsub_round_ps): Use __builtin_ia32_vfmsubps512_mask.
5586 (_mm512_mask_fmsub_round_ps): Likewise.
5587 (_mm512_fmsub_ps): Likewise.
5588 (_mm512_mask_fmsub_ps): Likewise.
5589 (_mm512_maskz_fmsub_round_ps): Use
5590 __builtin_ia32_vfmsubps512_maskz.
5591 (_mm512_maskz_fmsub_ps): Likewise.
5592 * config/i386/avx512vlintrin.h (_mm256_mask_fmsub_pd): Use
5593 __builtin_ia32_vfmsubpd256_mask.
5594 (_mm256_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd256_maskz.
5595 (_mm_mask_fmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
5596 (_mm_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd128_maskz.
5597 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
5598 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
5599 (_mm256_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps256_maskz.
5600 (_mm_mask_fmsub_ps): Use __builtin_ia32_vfmsubps128_mask.
5601 (_mm_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps128_maskz.
5602 * config/i386/fmaintrin.h (_mm_fmsub_pd): Use
5603 __builtin_ia32_vfmsubpd.
5604 (_mm256_fmsub_pd): Use __builtin_ia32_vfmsubpd256.
5605 (_mm_fmsub_ps): Use __builtin_ia32_vfmsubps.
5606 (_mm256_fmsub_ps): Use __builtin_ia32_vfmsubps256.
5607 (_mm_fmsub_sd): Use __builtin_ia32_vfmsubsd3.
5608 (_mm_fmsub_ss): Use __builtin_ia32_vfmsubss3.
5609 * config/i386/i386-builtin.def: Add
5610 __builtin_ia32_vfmsubpd256_mask,
5611 __builtin_ia32_vfmsubpd256_maskz,
5612 __builtin_ia32_vfmsubpd128_mask,
5613 __builtin_ia32_vfmsubpd128_maskz,
5614 __builtin_ia32_vfmsubps256_mask,
5615 __builtin_ia32_vfmsubps256_maskz,
5616 __builtin_ia32_vfmsubps128_mask,
5617 __builtin_ia32_vfmsubps128_maskz,
5618 __builtin_ia32_vfmsubpd512_mask,
5619 __builtin_ia32_vfmsubpd512_maskz,
5620 __builtin_ia32_vfmsubps512_mask,
5621 __builtin_ia32_vfmsubps512_maskz, __builtin_ia32_vfmsubss3,
5622 __builtin_ia32_vfmsubsd3, __builtin_ia32_vfmsubps,
5623 __builtin_ia32_vfmsubpd, __builtin_ia32_vfmsubps256 and.
5624 __builtin_ia32_vfmsubpd256.
5625 * config/i386/sse.md (fma4i_fmsub_<mode>): New.
5626 (<avx512>_fmsub_<mode>_maskz<round_expand_name>): Likewise.
5627 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_1):
5629 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_2):
5631 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_3):
5633 (fmai_vmfmsub_<mode><round_name>): Likewise.
5635 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
5637 * config/i386/sse.md (*<plusminus_insn><mode>3<mask_name>_bcst_1):
5638 Remove plus. Renamed to ...
5639 (*sub<mode>3<mask_name>_bcst): This.
5640 (*add<mode>3<mask_name>_bcst_2): Renamede to ...
5641 (*add<mode>3<mask_name>_bcst): This.
5643 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
5646 * config/i386/sse.md (*mul<mode>3<mask_name>_bcst): New.
5648 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
5651 * i386/avx512vlintrin.h (_mm256_or_epi32): New.
5652 (_mm_or_epi32): Likewise.
5653 (_mm256_xor_epi32): Likewise.
5654 (_mm_xor_epi32): Likewise.
5655 (_mm256_or_epi64): Likewise.
5656 (_mm_or_epi64): Likewise.
5657 (_mm256_xor_epi64): Likewise.
5658 (_mm_xor_epi64): Likewise.
5660 2018-10-20 H.J. Lu <hongjiu.lu@intel.com>
5663 * config/i386/sse.md (*<avx512>_div<mode>3<mask_name>_bcst): New.
5665 2018-10-20 Jakub Jelinek <jakub@redhat.com>
5668 * varasm.c (decode_addr_const): Handle COMPOUND_LITERAL_EXPR.
5670 2018-10-20 Andreas Schwab <schwab@linux-m68k.org>
5672 * doc/ux.texi: Move @section directly after @node.
5674 2018-10-19 Jakub Jelinek <jakub@redhat.com>
5678 * omp-low.c (check_omp_nesting_restrictions): Diagnose ordered without
5679 depend closely nested inside of loop with ordered clause with
5682 2018-10-19 David Malcolm <dmalcolm@redhat.com>
5684 * Makefile.in (TEXI_GCCINT_FILES): Add ux.texi.
5685 * doc/gccint.texi: Include ux.texi and use it in top-level menu.
5686 * doc/ux.texi: New file.
5688 2018-10-19 Segher Boessenkool <segher@kernel.crashing.org>
5690 * config/rs6000/rs6000.h (REG_ALLOC_ORDER): Move 68 (that is, CR0) to
5691 be the first CR field allocated.
5693 2018-10-19 Richard Biener <rguenther@suse.de>
5696 * config/i386/i386.c (ix86_builtin_vectorization_cost): Use
5697 TYPE_VECTOR_SUBPARTS and avoid relying on vector mode.
5699 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
5702 * config/i386/sse.md
5703 (*<plusminus_insn><mode>3<mask_name>_bcst_1): New.
5704 (*add<mode>3<mask_name>_bcst_2): Likewise.
5706 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
5708 * config/i386/sse.md
5709 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
5710 Replace nonimmediate_operand with register_operand.
5711 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
5713 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
5716 2018-10-19 Ilya Leoshkevich <iii@linux.ibm.com>
5718 PR rtl-optimization/87596
5719 * lra-constraints.c (spill_hard_reg_in_range): Use INSN_P () +
5720 lra_get_insn_recog_data () instead of lra_insn_recog_data[]
5721 for instructions in FROM..TO range.
5723 2018-10-19 Eric Botcazou <ebotcazou@adacore.com>
5725 * cfgexpand.c (expand_one_var): Use specific wording in error message
5726 for non-local frame variables.
5727 * stor-layout.c (layout_decl): Do not issue a warning for them.
5729 2018-10-19 Robin Dapp <rdapp@linux.ibm.com>
5731 * haifa-sched.c (priority): Add force_recompute parameter.
5732 (apply_replacement): Call priority () with force_recompute = true.
5733 (restore_pattern): Likewise.
5735 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
5737 * simplify-rtx.c (simplify_subreg): Limit mask of vec_merge to
5738 HOST_BITS_PER_WIDE_INT.
5739 (test_vector_ops_duplicate): Likewise.
5741 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
5744 * config/i386/sse.md (VF_AVX512): New.
5745 (avx512bcst): Likewise.
5746 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
5748 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
5750 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
5753 2018-10-18 Jonathan Wakely <jwakely@redhat.com>
5755 * doc/invoke.texi (-dumpversion): Improve grammar.
5756 (-dumpfullversion): Make more consistent with -dumpversion.
5758 2018-10-18 Uros Bizjak <ubizjak@gmail.com>
5760 * config/i386/i386.c (ix86_emit_fp_unordered_jump):
5761 Set JUMP_LABEL to the jump insn.
5762 (ix86_emit_i387_log1p): Use ix86_expand_branch to expand branch.
5763 Predict emitted jump and add label to jump insn.
5765 2018-10-18 David Malcolm <dmalcolm@redhat.com>
5767 PR tree-optimization/87562
5768 * input.c (get_substring_ranges_for_loc): Use
5769 LRK_SPELLING_LOCATION rather than LRK_MACRO_EXPANSION_POINT when
5770 getting the linemap for the endpoint. Verify that it's either
5771 in the same linemap as the start point's spelling location, or
5772 at least in the same file.
5774 2018-10-18 Richard Biener <rguenther@suse.de>
5776 * config/i386/i386.c (ix86_builtin_vectorization_cost): Do not
5777 feed width-specific load/store costs through ix86_vec_cost.
5778 * config/i386/x86-tune-costs.h (athlon_cost): Adjust.
5779 (k8_cost): Likewise.
5780 (bdver_cost): Likewise.
5781 (znver1_cost): Likewise.
5782 (btver1_cost): Likewise.
5783 (btver2_cost): Likewise.
5785 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
5787 * simplify-rtx.c (simplify_subreg): Call simplify_gen_subreg
5788 to simplify subreg of vec_merge.
5790 2018-10-18 Richard Biener <rguenther@suse.de>
5792 * config/i386/i386.c: Fix costing of vector FMA.
5794 2018-10-18 Richard Biener <rguenther@suse.de>
5796 * config/i386/i386.c (ix86_vec_cost): Remove !parallel path
5798 (ix86_builtin_vectorization_cost): For vec_construct properly
5799 cost insertion into SSE regs.
5800 (...): Adjust calls to ix86_vec_cost.
5802 2018-10-18 Richard Biener <rguenther@suse.de>
5806 2018-02-07 Richard Biener <rguenther@suse.de>
5808 PR tree-optimization/84204
5809 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
5812 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
5815 * simplify-rtx.c (simplify_subreg): Simplify subreg of vec_merge
5817 (test_vector_ops_duplicate): Add test for a scalar subreg of a
5818 VEC_MERGE of a VEC_DUPLICATE.
5820 2018-10-17 Joseph Myers <joseph@codesourcery.com>
5822 * doc/cpp.texi (__STDC_VERSION__): Document C2X handling.
5823 * doc/invoke.texi (-std=c2x, -std=gnu2x): Document new options.
5824 * doc/standards.texi (C Language): Document C2X.
5825 * dwarf2out.c (highest_c_language), config/rl78/rl78.c
5826 (rl78_option_override): Handle "GNU C2X" language name.
5828 2018-10-17 Joseph Myers <joseph@codesourcery.com>
5830 * doc/invoke.texi (-std=c17), doc/standards.texi (C Language):
5831 Document C17 as published in 2018.
5833 2018-10-17 Eric Botcazou <ebotcazou@adacore.com>
5836 * fold-const.c (fold_truth_andor_1): If the right side is not constant,
5837 bail out if both sides do not have the same storage order.
5839 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
5841 * bitmap.c (bitmap_head::dump): New.
5842 * bitmap.h (bitmap_head): Add dump().
5843 * gimple-ssa-evrp-analyze.c
5844 (evrp_range_analyzer::try_find_new_range): Adjust for value_range API.
5845 (evrp_range_analyzer::set_ssa_range_info): Same.
5846 (evrp_range_analyzer::record_ranges_from_phis): Same.
5847 (evrp_range_analyzer::record_ranges_from_stmt): Same.
5848 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Same.
5849 * gimple-ssa-sprintf.c (get_int_range): Same.
5850 (format_integer): Same.
5851 (sprintf_dom_walker::handle_gimple_call): Same.
5852 * ipa-cp.c (ipcp_vr_lattice::meet_with_1): Same.
5853 (ipcp_vr_lattice::top_p): Same.
5854 (ipcp_vr_lattice::bottom_p): Same.
5855 (ipcp_vr_lattice::set_to_bottom): Same.
5856 (ipa_vr_operation_and_type_effects): Same.
5857 (propagate_vr_across_jump_function): Same.
5858 (ipcp_store_vr_results): Same.
5859 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Same.
5860 (ipa_print_node_jump_functions_for_edge): Same.
5861 (ipa_get_value_range): Same.
5862 (ipa_compute_jump_functions_for_edge): Same.
5863 (ipa_write_jump_function): Same.
5864 * tree-ssa-dom.c (simplify_stmt_for_jump_threading): Same.
5865 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
5867 * vr-values.c (set_value_range_to_nonnegative): Same.
5868 (set_value_range_to_truthvalue): Same.
5869 (vr_values::get_value_range): Same.
5870 (vr_values::set_defs_to_varying): Same.
5871 (vr_values::update_value_range): Same.
5872 (symbolic_range_based_on_p): Same.
5873 (vr_values::op_with_boolean_value_range_p): Same.
5874 (vr_values::extract_range_for_var_from_comparison_expr): Same.
5875 (vr_values::extract_range_from_ssa_name): Same.
5876 (vr_values::extract_range_from_binary_expr): Same.
5877 (vr_values::extract_range_from_unary_expr): Same.
5878 (vr_values::extract_range_from_cond_expr): Same.
5879 (vr_values::extract_range_from_comparison): Same.
5880 (vr_values::check_for_binary_op_overflow): Same.
5881 (vr_values::extract_range_basic): Same.
5882 (vr_values::extract_range_from_assignment): Same.
5883 (compare_ranges): Same.
5884 (compare_range_with_value): Same.
5885 (vr_values::adjust_range_with_scev): Same.
5886 (vrp_valueize): Same.
5887 (vrp_valueize_1): Same.
5888 (vr_values::get_vr_for_comparison): Same.
5889 (vr_values::compare_name_with_value): Same.
5890 (vr_values::compare_names): Same.
5891 (vr_values::vrp_evaluate_conditional): Same.
5892 (find_case_label_ranges): Same.
5893 (vr_values::vrp_visit_switch_stmt): Same.
5894 (vr_values::extract_range_from_phi_node): Same.
5895 (vr_values::simplify_div_or_mod_using_ranges): Same.
5896 (vr_values::simplify_bit_ops_using_ranges): Same.
5897 (test_for_singularity): Same.
5898 (range_fits_type_p): Same.
5899 (vr_values::simplify_cond_using_ranges_1): Same.
5900 (vr_values::simplify_switch_using_ranges): Same.
5901 (vr_values::simplify_float_conversion_using_ranges): Same.
5902 (vr_values::two_valued_val_range_p): Same.
5903 (vr_values::add_equivalence): Move to value_range::equiv_add.
5904 * vr-values.h (vr_values::add_equivalence): Remove.
5905 (VR_INITIALIZER): Remove.
5906 * tree-vrp.c (value_range::set): New.
5907 (value_range::equiv_add): New.
5908 (value_range::value_range): New.
5909 (value_range::deep_copy): New.
5910 (value_range::check): New.
5911 (value_range::equal_p): New.
5912 (value_range::ignore_equivs_equal_p): New.
5913 (value_range::operator==): New.
5914 (value_range::operator!=): New.
5915 (value_range::symbolic_p): New.
5916 (value_range::numeric_p): New.
5917 (value_range::set_undefined): New.
5918 (value_range::set_varying): New.
5919 (value_range::may_contain_p): New.
5920 (value_range::equiv_clear): New.
5921 (value_range::singleton_p): New.
5922 (value_range::intersect): New.
5923 (value_range::dump): New.
5924 (value_range::set_and_canonicalize): New.
5925 (set_value_range): Adjust for value_range API.
5926 (set_value_range_to_undefined): Same.
5927 (set_value_range_to_varying): Same.
5928 (set_and_canonicalize_value_range): Same.
5929 (set_value_range_to_nonnull): Same.
5930 (set_value_range_to_null): Same.
5931 (range_is_null): Same.
5932 (range_is_nonnull): Same.
5933 (range_int_cst_p): Same.
5934 (range_int_cst_singleton_p): Same.
5935 (symbolic_range_p): Same.
5936 (range_includes_zero_p): Same.
5937 (value_range_constant_singleton): Same.
5938 (vrp_set_zero_nonzero_bits): Same.
5939 (ranges_from_anti_range): Same.
5940 (extract_range_into_wide_ints): Same.
5941 (extract_range_from_multiplicative_op): Same.
5942 (set_value_range_with_overflow): Same.
5943 (extract_range_from_binary_expr_1): Same.
5944 (extract_range_from_unary_expr): Same.
5945 (dump_value_range): Same.
5946 (debug_value_range): Same.
5947 (vrp_prop::check_array_ref): Same.
5948 (vrp_prop::check_mem_ref): Same.
5949 (vrp_prop::vrp_initialize): Same.
5950 (vrp_prop::visit_stmt): Same.
5951 (intersect_ranges): Same.
5952 (vrp_prop::visit_phi): Same.
5953 (vrp_prop::vrp_finalize): Same.
5954 (determine_value_range_1): Same.
5955 (determine_value_range): Same.
5956 (vrp_intersect_ranges_1): Rename to...
5957 (vrp_intersect_1): this.
5958 (vrp_intersect_ranges): Rename to...
5959 (value_range::intersect_helper): ...this.
5960 (vrp_meet_1): Rename to...
5961 (value_range::union_helper): ...this.
5962 (vrp_meet): Rename to...
5963 (value_range::union_): ...this.
5964 (copy_value_range): Remove.
5965 * tree-vrp.h (struct value_range): Rewrite into a proper class.
5966 (value_range::vrtype): New.
5967 (value_range::type): New.
5968 (value_range::equiv): New.
5969 (value_range::min): New.
5970 (value_range::max): New.
5971 (value_range::varying_p): New.
5972 (value_range::undefined_p): New.
5973 (value_range::null_p): New.
5974 (value_range::equiv_add): New.
5975 (copy_value_range): Remove.
5977 2018-10-17 David Malcolm <dmalcolm@redhat.com>
5979 * Makefile.in (SELFTEST_TARGETS): New.
5980 (selftest) Change from s-selftest-c to $(SELFTEST_TARGETS).
5981 (C_SELFTEST_FLAGS, C_SELFTEST_DEPS, s-selftest-c, selftest-c-gdb)
5982 (selftest-gdb, selftest-c-valgrind, selftest-valgrind): Move to
5984 (CPP_SELFTEST_FLAGS, CPP_SELFTEST_DEPS, s-selftest-c++)
5985 (selftest-c++-gdb, selftest-c++-valgrind): Move to
5987 * configure: Regenerate.
5988 * configure.ac (selftest_languages): New.
5990 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
5992 * tree-vrp.c (extract_range_from_multiplicative_op): Remove
5993 overflow wraps argument.
5994 (extract_range_from_binary_expr_1): Do not pass overflow wraps to
5995 wide_int_range_multiplicative_op.
5996 * wide-int-range.cc (wide_int_range_mult_wrapping): Remove
5997 overflow wraps argument.
5998 (wide_int_range_multiplicative_op): Same.
5999 (wide_int_range_lshift): Same.
6000 (wide_int_range_div): Same.
6001 * wide-int-range.h (wide_int_range_multiplicative_op): Same.
6002 (wide_int_range_lshift): Same.
6003 (wide_int_range_div): Same.
6005 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
6007 * wide-int-range.h (wide_int_range_shift_undefined_p): Adjust to
6008 use sign as argument.
6009 * tree-vrp.c (extract_range_from_binary_expr_1): Pass sign to
6010 wide_int_range_shift_undefined_p.
6012 2018-10-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6014 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>):
6016 (@despeculate_copy<ALLI_TI:mode>): ... This.
6017 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): Remove
6020 2018-10-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6022 * config.gcc: Obsolete *-*-solaris2.10*.
6023 * doc/install.texi (Specific, *-*-solaris2*): Document it.
6025 2018-10-12 Jeff Law <law@redhat.com>
6027 * config/ft32/ft32.md (ft32_general_movsrc_operand): Disable
6028 reg + sym +- const_int addressing modes.
6030 2018-10-15 David Malcolm <dmalcolm@redhat.com>
6032 * common.opt (fdiagnostics-minimum-margin-width=): New option.
6033 * diagnostic-show-locus.c (layout::layout): Apply the minimum
6035 (layout::start_annotation_line): Only print up to 3 of the
6036 margin character, to avoid touching the left-hand side.
6037 (selftest::test_diagnostic_show_locus_fixit_lines): Update for
6038 minimum margin width, as set by test_diagnostic_context's ctor.
6039 (selftest::test_fixit_insert_containing_newline): Likewise.
6040 (selftest::test_fixit_insert_containing_newline_2): Likewise.
6041 (selftest::test_line_numbers_multiline_range): Clear
6042 dc.min_margin_width.
6043 * diagnostic.c (diagnostic_initialize): Initialize
6045 * diagnostic.h (struct diagnostic_context): Add field
6047 * doc/invoke.texi: Add -fdiagnostics-minimum-margin-width=.
6048 * opts.c (common_handle_option): Handle
6049 OPT_fdiagnostics_minimum_margin_width_.
6050 * selftest-diagnostic.c
6051 (selftest::test_diagnostic_context::test_diagnostic_context):
6052 Initialize min_margin_width to 6.
6053 * toplev.c (general_init): Initialize global_dc->min_margin_width.
6055 2018-10-15 David Malcolm <dmalcolm@redhat.com>
6057 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
6058 Fix usage of "error_at_rich_loc" in the comment.
6060 2018-10-15 Renlin Li <renlin.li@arm.com>
6063 * tree-vectorizer.c (try_vectorize_loop_1): Don't use
6064 if-conversioned loop when it contains ifn with types not
6065 supported by backend.
6066 * internal-fn.c (expand_direct_optab_fn): Add an assert.
6067 (direct_internal_fn_supported_p): New helper function.
6068 * internal-fn.h (direct_internal_fn_supported_p): Declare.
6070 2018-10-15 Jakub Jelinek <jakub@redhat.com>
6073 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
6074 Remove OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
6075 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
6077 2018-10-15 Bin Cheng <bin.cheng@linux.alibaba.com>
6079 PR tree-optimization/87022
6080 * tree-loop-distribution.c (pg_add_dependence_edges): Check all
6081 bits in dist vector rather than the first one.
6083 2018-10-15 Richard Biener <rguenther@suse.de>
6086 * tree-ssa-structalias.c (struct vls_data): Add escaped_p member.
6087 (visit_loadstore): When a used restrict tag escaped verify that
6088 the points-to solution of "other" pointers do not include
6090 (compute_dependence_clique): If a used restrict tag escaped
6091 communicated that down to visit_loadstore.
6093 2018-10-15 Andreas Krebbel <krebbel@linux.ibm.com>
6095 * config/s390/s390.c (s390_expand_vec_init): Force vector element
6096 into reg if it isn't a general operand.
6098 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
6101 * config/i386/sse.md (*vec_dupv2di): Add register source to
6104 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
6107 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
6108 Add OPTION_MASK_ISA_AVX512IFMA_UNSET,
6109 OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
6110 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
6112 2018-10-13 Eric Botcazou <ebotcazou@adacore.com>
6114 * dwarf2cfi.c (struct dw_trace_info): Add args_size_defined_for_eh.
6115 (notice_args_size): Set it in the current trace if no insn that can
6116 throw internally has been seen yet.
6117 (connect_traces): When connecting args_size between traces, allow the
6118 incoming values not to match if there is an insn setting it before the
6119 first insn that can throw internally; in that case, force the creation
6120 of a CFI note on this latter insn.
6122 2018-10-13 Jonathan Wakely <jwakely@redhat.com>
6124 * opt-problem.h (opt_wrapper): Use template-argument-list when naming
6125 the base class, because using the injected-class-name was not clearly
6126 specified until DR 176.
6128 2018-10-12 Paul Koning <ni1d@arrl.net>
6130 * config/pdp11/pdp11.md (doloop_end): New expander.
6131 (doloop_end_insn): renamed from "doloop_end".
6132 (addqi3): New pattern.
6133 (subqi3): New pattern.
6134 * config/pdp11/predicates.md (incdec_operand): New predicate.
6136 2018-10-12 Yury Gribov <tetra2005@gmail.com>
6139 * real.c (format_helper::can_represent_integral_type_p): New function
6140 * real.h (format_helper::can_represent_integral_type_p): Ditto.
6141 * match.pd: New pattern.
6143 2018-10-12 Alexandre Oliva <oliva@adacore.com>
6145 * configure.ac: Introduce --enable-large-address-aware
6146 to define MINGW_DEFAULT_LARGE_ADDR_AWARE.
6147 * doc/install.texi: Document it.
6148 * configure, config.in: Rebuilt.
6149 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Define,
6150 based on MINGW_DEFAULT_LARGE_ADDR_AWARE.
6151 (LINK_SPEC): Insert it.
6152 * config/i386/mingw-w64.h: Likewise.
6154 * cygming.h (HAVE_GAS_ALIGNED_COMM): Fallback-define.
6156 2018-10-12 Peter Bergner <bergner@linux.ibm.com>
6158 PR rtl-optimization/87600
6159 * ira-lives (non_conflicting_reg_copy_p): Disable for non LRA targets.
6161 2018-10-12 Paul Koning <ni1d@arrl.net>
6163 * doc/md.texi (doloop_end): Document that the pattern code may
6164 need to check operand mode.
6166 2018-10-12 Wilco Dijkstra <wdijkstr@arm.com>
6168 * config/aarch64/aarch64.md (zero_extendsidi2_aarch64): Add alternatives
6169 to zero-extend between int and floating-point registers.
6170 (load_pair_zero_extendsidi2_aarch64): Add alternative for zero-extended
6171 ldp into floating-point registers. Add type and arch attributes.
6172 (zero_extend<SHORT:mode><GPI:mode>2_aarch64): Add arch attribute.
6173 Use f_loads for type attribute.
6175 2018-10-11 Martin Sebor <msebor@redhat.com>
6177 * doc/extend.texi (attribute packed): Correct typos.
6179 2018-10-11 Martin Sebor <msebor@redhat.com>
6181 * doc/extend.texi (attribute flatten): Mention interaction with
6184 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
6187 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
6188 Set new_decl virtual flag to zero.
6190 2018-10-11 Martin Sebor <msebor@redhat.com>
6193 * doc/extend.texi (attribute format_arg): Discuss using multiple
6194 attributes on a single function.
6196 2018-10-11 Giuliano Belinassi <giuliano.belinassi@usp.br>
6198 PR tree-optimization/86829
6199 * match.pd (sin (atan (x))): New simplification rules.
6200 (cos (atan (x))): Likewise.
6201 * real.c (build_sinatan_real): New function.
6202 * real.h (build_sinatan_real): Prototype.
6204 2018-10-11 Will Schmidt <will_schmidt@vnet.ibm.com>
6206 * config/rs6000/rs6000.c (map_to_integral_tree_type): New helper
6208 (fold_mergeeo_helper): New helper function.
6209 (rs6000_gimple_fold_builtin): Add hooks for vec_mergee and vec_mergeo
6210 intrinsics. Correct some whitespace indentation issues.
6212 2018-10-11 Wilco Dijkstra <wdijkstr@arm.com>
6215 * config/aarch64/aarch64.c (aarch64_mask_and_shift_for_ubfiz_p):
6216 Use HOST_WIDE_INT_1U for shift.
6218 2018-10-11 Doug Rupp <rupp@adacore.com>
6219 Olivier Hainque <hainque@adacore.com>
6221 * config/rs6000/vxworks.h (VXWORKS_RELAX_LINK_SPEC): New macro.
6222 Pass --relax to the linker for RTPs.
6223 (LINK_SPEC): Append VXWORKS_RELAX_LINK_SPEC.
6225 2018-10-11 Andrew Stubbs <ams@codesourcery.com>
6226 Jan Hubicka <jh@suse.cz>
6227 Martin Jambor <mjambor@suse.cz>
6229 * print-rtl.c (print_rtx_operand_codes_E_and_V): Print how many times
6230 the same elements are repeated rather than printing all of them.
6231 * read-rtl.c (rtx_reader::read_rtx_operand): Recognize and expand
6232 "repeated" elements.
6233 * read-rtl-function.c (test_loading_repeat): New function.
6234 (read_rtl_function_c_tests): Call test_loading_repeat.
6235 * rtl-tests.c (test_dumping_repeat): New function.
6236 (rtl_tests_c_tests): Call test_dumping_repeat.
6238 2018-10-11 Richard Biener <rguenther@suse.de>
6240 * config/i386/x86-tune-costs.h (bdver?_memcpy, bdver?_memset,
6241 bdver?_cost): Unify to ...
6242 (bdver_memcpy, bdver_memset, bdver_cost): ... this.
6243 * config/i386/i386.c (processor_cost_table): Adjust.
6245 2018-10-10 Eric Botcazou <ebotcazou@adacore.com>
6248 * cgraphunit.c (cgraph_node::expand_thunk): Force DECL_IGNORED_P on
6249 the thunk when expanding to GIMPLE.
6251 2018-10-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
6253 * varasm.c (mergeable_string_section): Don't try to move zero-length
6254 strings to the merge section.
6256 2018-10-10 Uros Bizjak <ubizjak@gmail.com>
6259 * config/i386/mmx.md (const_vector 0 -> mem splitter): New splitter.
6261 2018-10-10 Jakub Jelinek <jakub@redhat.com>
6264 * config/i386/i386-builtin.def (IX86_BUILTIN_RDPMC): Move from args set
6265 to special_args set.
6267 2018-10-10 Richard Biener <rguenther@suse.de>
6269 * config/i386/sse.md (reduc_plus_scal_v8df, reduc_plus_scal_v4df,
6270 reduc_plus_scal_v2df, reduc_plus_scal_v16sf, reduc_plus_scal_v8sf,
6271 reduc_plus_scal_v4sf): Merge into pattern reducing to half width
6272 and recursing and pattern terminating the recursion on SSE
6273 vector width using ix86_expand_reduc.
6274 (reduc_sminmax_scal_<mode>): Split into part reducing to half
6275 width and recursing and SSE2 vector variant doing the final
6276 reduction with ix86_expand_reduc.
6277 (reduc_uminmax_scal_<mode>): Likewise for the AVX512 variants
6278 with terminating the recursion at AVX level, splitting that
6281 2018-10-09 David Malcolm <dmalcolm@redhat.com>
6283 * genmatch.c (error_cb): Rename to...
6284 (diagnostic_cb): ...this, converting int params to enums.
6285 (fatal_at): Update for renaming.
6286 (warning_at): Likewise.
6288 * input.c (selftest::ebcdic_execution_charset::apply):
6289 Update for renaming of...
6290 (selftest::ebcdic_execution_charset::on_error): ...this, renaming
6292 (selftest::ebcdic_execution_charset::on_diagnostic): ...this,
6293 converting level and reason to enums.
6294 (class selftest::lexer_error_sink): Rename to...
6295 (class selftest::lexer_test_options): ...this, renaming field
6296 "m_errors" to "m_diagnostics".
6297 (selftest::lexer_test_options::apply): Update for renaming of...
6298 (selftest::lexer_test_options::on_error): ...this, renaming to...
6299 (selftest::lexer_test_options::on_diagnostic): ...this
6300 converting level and reason to enums.
6301 (selftest::test_lexer_string_locations_raw_string_unterminated):
6302 Update for renamings.
6303 * opth-gen.awk (struct cpp_reason_option_codes_t): Use enum for
6306 2018-10-09 Paul A. Clarke <pc@us.ibm.com>
6308 * config.gcc (powerpc*-*-*): Add pmmintrin.h to extra_headers.
6309 * config/rs6000/pmmintrin.h: New file.
6311 2018-10-09 Eric Botcazou <ebotcazou@adacore.com>
6313 PR tree-optimization/86659
6314 * gimple-match.h (gimple_match_op constructors): Initialize reverse.
6316 2018-10-09 Richard Biener <rguenther@suse.de>
6318 PR tree-optimization/63155
6319 * tree-ssa-structalias.c: Include tree-ssa.h.
6320 (get_constraint_for_ssa_var): For undefs return nothing_id.
6321 (find_func_aliases): Cleanup PHI handling.
6323 2018-10-09 Richard Biener <rguenther@suse.de>
6325 * tree-vectorizer.c (dump_stmt_cost): Fix cut&paste missing
6328 2018-10-09 Martin Liska <mliska@suse.cz>
6330 * asan.c (asan_emit_stack_protection): If a stack variable
6331 is located in a same file as current function, then emit
6332 line info into variable definition string.
6334 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
6336 * print-rtl.c (rtx_writer::print_rtx_operand_code_i): Print column
6339 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
6341 * cgraphunit.c (cgraph_node::expand_thunk): Do not force DECL_IGNORED_P
6344 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
6346 PR tree-optimization/86659
6347 * gimple-match.h (struct gimple_match_op): Add reverse field.
6348 (gimple_match_op::set_op): New overloaded method.
6349 * gimple-match-head.c (maybe_build_generic_op) <BIT_FIELD_REF>: Set
6350 the REF_REVERSE_STORAGE_ORDER flag on the value.
6351 (gimple_simplify) <GIMPLE_ASSIGN>: For BIT_FIELD_REF, propagate the
6352 REF_REVERSE_STORAGE_ORDER flag and avoid simplifying if it is set.
6354 2018-10-08 Richard Sandiford <richard.sandiford@arm.com>
6357 * gimple-ssa-backprop.c (backprop::intersect_uses): Use
6358 FOR_EACH_IMM_USE_FAST instead of FOR_EACH_IMM_USE_STMT.
6360 2018-10-08 H.J. Lu <hongjiu.lu@intel.com>
6363 * config/i386/avx512fintrin.h (_mm512_mask_fmaddsub_round_pd):
6364 Defined with __builtin_ia32_vfmaddsubpd512_mask.
6366 2018-10-08 Richard Biener <rguenther@suse.de>
6368 * config/i386/x86-tune-costs.h (znver1_cost): Make AVX256 vector loads
6369 cost the same as AVX128 ones.
6371 2018-10-08 Paul Koning <ni1d@arrl.net>
6373 * config/pdp11/pdp11-protos.h (output_block_move): Remove.
6374 (expand_block_move): New function.
6375 * config/pdp11/pdp11.c (output_block_move): Remove.
6376 (expand_block_move): New function.
6377 * config/pdp11/pdp11.h (MOVE_RATIO): New definition.
6378 * config/pdp11/pdp11.md (movmemhi): Use expand_block_move.
6379 (*movmemhi1): Remove.
6381 2018-10-08 Robin Dapp <rdapp@linux.ibm.com>
6383 * config/s390/2827.md: Increase latencies for some FP instructions.
6385 2018-10-08 Richard Biener <rguenther@suse.de>
6387 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
6389 * tree-vectorizer.c (dump_stmt_cost): Add cost param and dump it.
6390 * tree-vectorizer.h (dump_stmt_cost): Adjust.
6391 (add_stmt_cost): Dump return value of the hook.
6393 2018-10-08 Richard Biener <rguenther@suse.de>
6395 PR tree-optimization/63155
6396 * tree-ssa-propagate.c (add_ssa_edge): Do cheap check first.
6397 (ssa_propagation_engine::ssa_propagate): Remove redundant
6398 bitmap bit clearing.
6400 2018-10-05 Peter Bergner <bergner@linux.ibm.com>
6402 PR rtl-optimization/86939
6403 PR rtl-optimization/87479
6404 * ira.h (non_conflicting_reg_copy_p): New prototype.
6405 * ira-lives.c (ignore_reg_for_conflicts): New static variable.
6406 (make_hard_regno_dead): Don't add conflicts for register
6407 ignore_reg_for_conflicts.
6408 (make_object_dead): Likewise.
6409 (non_conflicting_reg_copy_p): New function.
6410 (process_bb_node_lives): Set ignore_reg_for_conflicts for copies.
6411 Remove special conflict handling of REAL_PIC_OFFSET_TABLE_REGNUM.
6412 * lra-lives.c (ignore_reg_for_conflicts): New static variable.
6413 (make_hard_regno_dead): Don't add conflicts for register
6414 ignore_reg_for_conflicts. Remove special conflict handling of
6415 REAL_PIC_OFFSET_TABLE_REGNUM. Remove now unused argument
6416 check_pic_pseudo_p and update callers.
6417 (mark_pseudo_dead): Don't add conflicts for register
6418 ignore_reg_for_conflicts.
6419 (process_bb_lives): Set ignore_reg_for_conflicts for copies.
6421 2018-10-05 Andrew Waterman <andrew@sifive.com>
6422 Jim Wilson <jimw@sifive.com>
6424 * config/riscv/riscv.md (f<quiet_pattern>_quiet<ANYF:mode><X:mode>4):
6425 Add define_expand. Add ! HONOR_SNANS check to current pattern. Add
6426 new pattern using HONOR_SNANS that emits one extra instruction.
6428 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
6430 * config/rs6000/rs6000.md (unnamed mfcr scc_comparison_operator
6431 patterns): Merge SI and DI patterns to a GPR pattern.
6432 (unnamed define_insn and define_split for record form of that): Merge
6433 to a single define_insn_and_split pattern.
6435 2018-10-05 David Malcolm <dmalcolm@redhat.com>
6438 * input.c (expand_location_to_spelling_point): Add param "aspect"
6439 and use rather than hardcoding LOCATION_ASPECT_CARET.
6440 (get_substring_ranges_for_loc): Handle the case of a single token
6441 within a macro expansion.
6442 * input.h (expand_location_to_spelling_point): Add "aspect" param,
6443 defaulting to LOCATION_ASPECT_CARET.
6445 2018-10-05 Paul Koning <ni1d@arrl.net>
6447 * config/pdp11/pdp11.c (TARGET_CXX_GUARD_TYPE): Define.
6448 (TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT): Define.
6449 (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
6450 (pdp11_guard_type): New function.
6452 2018-10-05 Paul Koning <ni1d@arrl.net>
6454 * config/pdp11/pdp11.h (FLOAT_TYPE_SIZE): Always 32.
6455 * config/pdp11/pdp11.opt (mfloat32): Remove.
6457 * doc/invoke.texi (pdp11 -mfloat32): Remove:
6458 (pdp11 -mfloat64): Remove.
6460 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
6462 * config/i386/i386.md (*cmpxf_cc_i387): Remove pattern.
6463 (*cmp<mode>_cc_i387): Ditto.
6464 (*cmpu<mode>_cc_i387): Ditto.
6465 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
6466 * config/i386/i386.c (ix86_expand_fp_compare): Remove
6468 <case IX86_FPCMP_SAHF>: Do not generate pattern with HImode clobber.
6469 Emit x86_sahf_1 pattern.
6470 (ix86_expand_compare): Update call to ix86_expand_fp_compare.
6471 (ix86_expand_carry_flag_compare): Ditto.
6473 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
6475 * config/i386/i386.md (*cmpxf_i387): Change operand 2 predicate
6476 to reg_or_0_operand. Add "C" constraint.
6477 (*cmpxf_cc_i387): Ditto.
6478 (*cmp<mode>_i387): Change operand 2 predicate
6479 to nonimm_or_0_operand. Add "C" constraint.
6480 (*cmp<mode>_cc_i387): Ditto.
6481 (*cmp<mode>_0_i387): Remove insn pattern.
6482 (*cmp<mode>_0_cc_i387): Ditto.
6484 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
6486 * config/i386/constraints.md ("C"): Do not depend on TARGET_SSE.
6487 * config/i386/predicates.md (nonimm_or_0_operand): Rename
6488 from vector_move_operand. Update all uses.
6490 2018-10-05 Martin Sebor <msebor@redhat.com>
6492 PR tree-optimization/87490
6493 * builtins.c (expand_builtin_strnlen): Handle a null data.decl
6496 2018-10-05 Richard Biener <rguenther@suse.de>
6498 PR tree-optimization/63155
6499 * tree-ssa-ccp.c (ccp_propagate::visit_phi): Avoid excess
6500 vertical space in dumpfiles.
6501 * tree-ssa-propagate.h
6502 (ssa_propagation_engine::process_ssa_edge_worklist): Remove.
6503 * tree-ssa-propagate.c (cfg_blocks_back): New global.
6504 (ssa_edge_worklist_back): Likewise.
6505 (curr_order): Likewise.
6506 (cfg_blocks_get): Remove abstraction.
6507 (cfg_blocks_add): Likewise.
6508 (cfg_blocks_empty_p): Likewise.
6509 (add_ssa_edge): Add to current or next worklist based on
6511 (add_control_edge): Likewise.
6512 (ssa_propagation_engine::process_ssa_edge_worklist): Fold
6514 (ssa_propagation_engine::ssa_propagate): ... here. Unify
6515 iteration from CFG and SSA edge worklist so we process
6516 everything in RPO order, prioritizing forward progress
6518 (ssa_prop_init): Allocate new worklists, do not dump
6520 (ssa_prop_fini): Free new worklists.
6522 2018-10-05 Richard Biener <rguenther@suse.de>
6524 * tree-core.h (tree_block::abstract_flag): Remove.
6525 (tree_block::block_num): Make full 32bits.
6526 * tree.def (BLOCK): Remove docs about BLOCK_ABSTRACT.
6527 * tree.h (BLOCK_ABSTRACT): Remove.
6528 * dwarf2out.c (gen_lexical_block_die): Remove dead code
6529 resulting from BLOCK_ABSTRACT being always false.
6530 (gen_inlined_subroutine_die): Likewise.
6531 (gen_block_die): Likewise.
6532 * tree.c (block_ultimate_origin): Likewise.
6533 * tree-pretty-print.c (dump_block_node): Remove code dealing
6534 with BLOCK_ABSTRACT.
6535 * tree-ssa-live.c (dump_scope_block): Likewise.
6536 * tree-streamer-in.c (unpack_ts_block_value_fields): Likewise.
6537 * tree-streamer-out.c (pack_ts_block_value_fields): Likewise.
6539 2018-10-05 Richard Biener <rguenther@suse.de>
6541 * config/i386/i386.c (ix86_add_stmt_cost): When scalar cost
6542 is asked for initialize mode to the component mode of the
6545 2018-10-05 H.J. Lu <hongjiu.lu@intel.com>
6548 * config/i386/gnu-user.h (ASM_SPEC): Don't pass -msse2avx to
6549 assembler for -mavx.
6550 * config/i386/gnu-user64.h (ASM_SPEC): Likewise.
6552 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
6555 * config/rs6000/rs6000-builtin.def (RS6000_BUILTIN_SET_FPSCR_DRN): Use
6557 * config/rs6000/rs6000.md (rs6000_set_fpscr_rn): Require the operand
6558 to be DImode. When using mffscrn, force the operand to a register.
6560 2018-10-04 Uros Bizjak <ubizjak@gmail.com>
6562 * config/i386/i386.md (*fop_<X87MODEF:mode>_2_i387): Macroize insn
6563 from *fop_<MODEF:mode>_2_i387 and *fop_xf_2_i387 using
6564 X87MODEF mode iterator.
6565 (*fop_<X87MODEF:mode>_3_i387): Macroize insn from
6566 *fop_<MODEF:mode>_3_i387 and *fop_xf_3_i387 using
6567 X87MODEF mode iterator.
6569 2018-10-04 Vinay Kumar <vinay.kumar@blackfigtech.com>
6571 * doc/invoke.texi (-Wno-prio-ctor-dtor): Document new warning
6572 -Wno-prio-ctor-dtor.
6574 2018-10-04 David Malcolm <dmalcolm@redhat.com>
6576 * Makefile.in (OBJS): Add opt-problem.o.
6577 * dump-context.h: Include "selftest.h.
6578 (selftest::temp_dump_context): New forward decl.
6579 (class dump_context): Make friend of class
6580 selftest::temp_dump_context.
6581 (dump_context::dump_loc_immediate): New decl.
6582 (class dump_pretty_printer): Move here from dumpfile.c.
6583 (class temp_dump_context): Move to namespace selftest.
6584 (temp_dump_context::temp_dump_context): Add param
6585 "forcibly_enable_dumping".
6586 (selftest::verify_dumped_text):
6587 (ASSERT_DUMPED_TEXT_EQ): Move here from dumpfile.c.
6588 (selftest::verify_item):
6589 (ASSERT_IS_TEXT): Move here from dumpfile.c.
6590 (ASSERT_IS_TREE): Likewise.
6591 (ASSERT_IS_GIMPLE): Likewise.
6592 * dumpfile.c (dump_context::dump_loc): Move immediate dumping
6594 (dump_context::dump_loc_immediate): ...this new function.
6595 (class dump_pretty_printer): Move to dump-context.h.
6596 (dump_switch_p_1): Don't enable MSG_PRIORITY_REEMITTED.
6597 (opt_info_switch_p_1): Enable MSG_PRIORITY_REEMITTED.
6598 (temp_dump_context::temp_dump_context): Move to "selftest"
6599 namespace. Add param "forcibly_enable_dumping", and use it to
6600 conditionalize the use of m_pp;
6601 (selftest::verify_dumped_text): Make non-static.
6602 (ASSERT_DUMPED_TEXT_EQ): Move to dump-context.h.
6603 (selftest::verify_item): Make non-static.
6604 (ASSERT_IS_TEXT): Move to dump-context.h.
6605 (ASSERT_IS_TREE): Likewise.
6606 (ASSERT_IS_GIMPLE): Likewise.
6607 (selftest::test_capture_of_dump_calls): Pass "true" for new
6608 param of temp_dump_context.
6609 * dumpfile.h (enum dump_flag): Add MSG_PRIORITY_REEMITTED, adding
6610 it to MSG_ALL_PRIORITIES. Update values of TDF_COMPARE_DEBUG and
6612 * opt-problem.cc: New file.
6613 * opt-problem.h: New file.
6614 * optinfo-emit-json.cc
6615 (selftest::test_building_json_from_dump_calls): Pass "true" for
6616 new param of temp_dump_context.
6617 * optinfo.cc (optinfo_kind_to_dump_flag): New function.
6618 (optinfo::emit_for_opt_problem): New function.
6619 (optinfo::emit): Clarity which emit_item is used.
6620 * optinfo.h (optinfo::get_dump_location): New accessor.
6621 (optinfo::emit_for_opt_problem): New decl.
6622 (optinfo::emit): Make const.
6623 * selftest-run-tests.c (selftest::run_tests): Call
6624 selftest::opt_problem_cc_tests.
6625 * selftest.h (selftest::opt_problem_cc_tests): New decl.
6626 * tree-data-ref.c (dr_analyze_innermost): Convert return type from
6627 bool to opt_result, converting fprintf messages to
6628 opt_result::failure_at calls. Add "stmt" param for use by the
6630 (create_data_ref): Pass "stmt" to the dr_analyze_innermost call.
6631 (runtime_alias_check_p): Convert return type from bool to
6632 opt_result, converting dump_printf calls to
6633 opt_result::failure_at, using the statement DDR_A for their
6635 (find_data_references_in_stmt): Convert return type from bool to
6636 opt_result, converting "return false" to opt_result::failure_at
6638 * tree-data-ref.h: Include "opt-problem.h".
6639 (dr_analyze_innermost): Convert return type from bool to opt_result,
6640 and add a const gimple * param.
6641 (find_data_references_in_stmt): Convert return type from bool to
6643 (runtime_alias_check_p): Likewise.
6644 * tree-predcom.c (find_looparound_phi): Pass "init_stmt" to
6645 dr_analyze_innermost.
6646 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test):
6647 Convert return type from bool to opt_result, adding a message for
6648 the PARAM_VECT_MAX_VERSION_FOR_ALIAS_CHECKS zero case.
6649 (vect_analyze_data_ref_dependence): Convert return type from bool
6650 to opt_result. Change sense of return type from "false"
6651 effectively meaning "no problems" to "false" meaning a problem,
6652 so that "return false" becomes "return opt_result::success".
6653 Convert "return true" calls to opt_result::failure_at, using
6654 the location of statement A rather than vect_location.
6655 (vect_analyze_data_ref_dependences): Convert return type from bool
6657 (verify_data_ref_alignment): Likewise, converting dump_printf_loc
6658 calls to opt_result::failure_at, using the stmt location rather
6660 (vect_verify_datarefs_alignment): Convert return type from bool
6662 (vect_enhance_data_refs_alignment): Likewise. Split local "stat"
6663 into multiple more-tightly-scoped copies.
6664 (vect_analyze_data_refs_alignment): Convert return type from bool
6666 (vect_analyze_data_ref_accesses): Likewise, converting a
6667 "return false" to a "return opt_result::failure_at", adding a
6669 (vect_prune_runtime_alias_test_list): Convert return type from
6670 bool to opt_result, converting dump_printf_loc to
6671 opt_result::failure_at. Add a %G to show the pertinent statement,
6672 and use the stmt's location rather than vect_location.
6673 (vect_find_stmt_data_reference): Convert return type from
6674 bool to opt_result, converting dump_printf_loc to
6675 opt_result::failure_at, using stmt's location.
6676 (vect_analyze_data_refs): Convert return type from bool to
6677 opt_result. Convert "return false" to "return
6678 opt_result::failure_at", adding messages as needed.
6679 * tree-vect-loop.c (vect_determine_vf_for_stmt_1): Convert return
6680 type from bool to opt_result.
6681 (vect_determine_vf_for_stmt): Likewise.
6682 (vect_determine_vectorization_factor): Likewise, converting
6683 dump_printf_loc to opt_result::failure_at, using location of phi
6684 rather than vect_location.
6685 (vect_analyze_loop_form_1): Convert return type from bool to
6686 opt_result, converting dump_printf_loc calls, retaining the use of
6688 (vect_analyze_loop_form): Convert return type from loop_vec_info
6689 to opt_loop_vec_info.
6690 (vect_analyze_loop_operations): Convert return type from bool to
6691 opt_result, converting dump_printf_loc calls, using the location
6692 of phi/stmt rather than vect_location where available. Convert
6693 various "return false" to "return opt_result::failure_at" with
6694 "unsupported phi" messages.
6695 (vect_get_datarefs_in_loop): Convert return type from bool to
6696 opt_result. Add a message for the
6697 PARAM_LOOP_MAX_DATAREFS_FOR_DATADEPS failure.
6698 (vect_analyze_loop_2): Convert return type from bool to
6699 opt_result. Ensure "ok" is set to a opt_result::failure_at before
6700 each "goto again;", adding new messages where needed.
6701 Add "unsupported grouped {store|load}" messages.
6702 (vect_analyze_loop): Convert return type from loop_vec_info to
6704 * tree-vect-slp.c (vect_analyze_slp): Convert return type from
6706 * tree-vect-stmts.c (process_use): Likewise, converting
6707 dump_printf_loc call and using stmt location, rather than
6709 (vect_mark_stmts_to_be_vectorized): Likeise.
6710 (vect_analyze_stmt): Likewise, adding a %G.
6711 (vect_get_vector_types_for_stmt): Convert return type from bool to
6712 opt_result, converting dump_printf_loc calls and using stmt
6713 location, rather than vect_location.
6714 (vect_get_mask_type_for_stmt): Convert return type from tree to
6715 opt_tree, converting dump_printf_loc calls and using stmt location.
6716 * tree-vectorizer.c: Include "opt-problem.h.
6717 (try_vectorize_loop_1): Flag "Analyzing loop at" dump message as
6718 MSG_PRIORITY_INTERNALS. Convert local "loop_vinfo" from
6719 loop_vec_info to opt_loop_vec_info. If if fails, and dumping is
6720 enabled, use it to report at the top level "couldn't vectorize
6721 loop" followed by the problem.
6722 * tree-vectorizer.h (opt_loop_vec_info): New typedef.
6723 (vect_mark_stmts_to_be_vectorized): Convert return type from bool
6725 (vect_analyze_stmt): Likewise.
6726 (vect_get_vector_types_for_stmt): Likewise.
6727 (tree vect_get_mask_type_for_stmt): Likewise.
6728 (vect_analyze_data_ref_dependences): Likewise.
6729 (vect_enhance_data_refs_alignment): Likewise.
6730 (vect_analyze_data_refs_alignment): Likewise.
6731 (vect_verify_datarefs_alignment): Likewise.
6732 (vect_analyze_data_ref_accesses): Likewise.
6733 (vect_prune_runtime_alias_test_list): Likewise.
6734 (vect_find_stmt_data_reference): Likewise.
6735 (vect_analyze_data_refs): Likewise.
6736 (vect_analyze_loop): Convert return type from loop_vec_info to
6738 (vect_analyze_loop_form): Likewise.
6739 (vect_analyze_slp): Convert return type from bool to opt_result.
6741 2018-10-04 David Malcolm <dmalcolm@redhat.com>
6743 * doc/invoke.texi (-fopt-info): Document new "internals"
6745 * dump-context.h (dump_context::apply_dump_filter_p): New decl.
6746 * dumpfile.c (dump_options): Update for renaming of MSG_ALL to
6748 (optinfo_verbosity_options): Add "internals".
6749 (kind_as_string): Update for renaming of MSG_ALL to MSG_ALL_KINDS.
6750 (dump_context::apply_dump_filter_p): New member function.
6751 (dump_context::dump_loc): Use apply_dump_filter_p rather than
6752 explicitly masking the dump_kind.
6753 (dump_context::begin_scope): Increment the scope depth first. Use
6754 apply_dump_filter_p rather than explicitly masking the dump_kind.
6755 (dump_context::emit_item): Use apply_dump_filter_p rather than
6756 explicitly masking the dump_kind.
6757 (dump_dec): Likewise.
6758 (dump_hex): Likewise.
6759 (dump_switch_p_1): Default to MSG_ALL_PRIORITIES.
6760 (opt_info_switch_p_1): Default to MSG_PRIORITY_USER_FACING.
6761 (opt_info_switch_p): Update handling of default
6762 MSG_OPTIMIZED_LOCATIONS to cope with default of
6763 MSG_PRIORITY_USER_FACING.
6764 (dump_basic_block): Use apply_dump_filter_p rather than explicitly
6765 masking the dump_kind.
6766 (selftest::test_capture_of_dump_calls): Update test_dump_context
6767 instances to use MSG_ALL_KINDS | MSG_PRIORITY_USER_FACING rather
6768 than MSG_ALL. Generalize scope test to be run at all four
6769 combinations of with/without MSG_PRIORITY_USER_FACING and
6770 MSG_PRIORITY_INTERNALS, adding examples of explicit priority
6771 for each of the two values.
6772 * dumpfile.h (enum dump_flag): Add comment about the MSG_* flags.
6773 Rename MSG_ALL to MSG_ALL_KINDS. Add MSG_PRIORITY_USER_FACING,
6774 MSG_PRIORITY_INTERNALS, and MSG_ALL_PRIORITIES, updating the
6775 values for TDF_COMPARE_DEBUG and TDF_ALL_VALUES.
6776 (AUTO_DUMP_SCOPE): Add a note to the comment about the interaction
6777 with MSG_PRIORITY_*.
6778 * tree-vect-loop-manip.c (vect_loop_versioning): Mark versioning
6779 dump messages as MSG_PRIORITY_USER_FACING.
6780 * tree-vectorizer.h (DUMP_VECT_SCOPE): Add a note to the comment
6781 about the interaction with MSG_PRIORITY_*.
6783 2018-10-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
6785 * varasm.c (output_constant): Add new parameter merge_strings.
6786 Make strings properly zero terminated in merge string sections.
6787 (mergeable_string_section): Don't fail if the last char is non-zero.
6788 (assemble_variable_contents): Handle merge string sections.
6789 (assemble_variable): Likewise.
6790 (assemble_constant_contents): Likewise.
6791 (output_constant_def_contents): Likewise.
6792 (output_constructor_array_range,
6793 output_constructor_regular_field): Adjust call to output_constant.
6794 (output_object_block): Adjust call to assemble_constant_contents
6795 and assemble_variable_contents.
6797 2018-10-04 Martin Liska <mliska@suse.cz>
6800 * cgraphunit.c (process_function_and_variable_attributes):
6801 Warn about a function with alias attribute and a body.
6803 2018-10-04 Martin Liska <mliska@suse.cz>
6806 * multiple_target.c (redirect_to_specific_clone): New function.
6807 (ipa_target_clone): Use it.
6808 * tree-inline.c: Fix comment.
6810 2018-10-04 David Malcolm <dmalcolm@redhat.com>
6812 * dumpfile.c (gcc::dump_manager::dump_manager): Initialize new
6814 (gcc::dump_manager::~dump_manager): Free m_optinfo_filename.
6815 (gcc::dump_manager::register_pass): New member function, adapted
6816 from loop body in gcc::pass_manager::register_pass, adding a
6817 call to update_dfi_for_opt_info.
6818 (gcc::dump_manager::opt_info_enable_passes): Store the
6819 -fopt-info options into the new fields. Move the loop
6821 (gcc::dump_manager::update_dfi_for_opt_info): ...this new member
6823 * dumpfile.h (struct opt_pass): New forward decl.
6824 (gcc::dump_manager::register_pass): New decl.
6825 (gcc::dump_manager::update_dfi_for_opt_info): New decl.
6826 (class gcc::dump_manager): Add fields "m_optgroup_flags",
6827 "m_optinfo_flags", and "m_optinfo_filename".
6828 * passes.c (gcc::pass_manager::register_pass): Move all of the
6829 dump-handling code to gcc::dump_manager::register_pass.
6831 2018-10-04 Peter Bergner <bergner@linux.ibm.com>
6833 PR rtl-optimization/87466
6834 * target.def (setjmp_preserves_nonvolatile_regs_p): New target hook.
6835 * doc/tm.texi.in (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P): New hook.
6836 * doc/tm.texi: Regenerate.
6837 * ira-lives.c (process_bb_node_lives): Use the new target hook.
6838 * lra-lives.c (process_bb_lives): Likewise.
6839 * config/rs6000/rs6000.c (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P):
6842 2018-10-04 Tamar Christina <tamar.christina@arm.com>
6844 * params.c (add_params): Fix initialization.
6846 2018-10-04 Martin Liska <mliska@suse.cz>
6848 PR gcov-profile/84107
6849 * tree-profile.c (init_ic_make_global_vars):
6850 Remove ic_void_ptr_var and ic_gcov_type_ptr_var.
6851 Come up with new ic_tuple* variables. Emit
6852 __gcov_indirect_call{,_topn} variables.
6853 (gimple_gen_ic_profiler): Access the variable
6855 (gimple_gen_ic_func_profiler): Access
6856 __gcov_indirect_call.callee field.
6857 (gimple_init_gcov_profiler): Use ptr_type_node.
6858 * value-prof.c (gimple_ic): Use ptr_type_node.
6860 2018-10-04 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6862 PR tree-optimization/85787
6863 * ipa-pure-const.c (malloc_candidate_p_1): Move most of
6864 malloc_candidate_p into this function and add support for
6865 detecting multiple phis.
6866 (DUMP_AND_RETURN): Move from malloc_candidate_p into top-level macro.
6868 2018-10-04 Martin Liska <mliska@suse.cz>
6871 * ipa-inline.c (inline_to_all_callers_1):
6872 Call ultimate_alias_target for node being inlined.
6874 2018-10-03 Jeff Law <law@redhat.com>
6876 * gimple-ssa-sprintf.c (format_string): Do not hardcode size of
6878 * tree.c (get_typenode_from_name): Moved from fortran/trans-types.c.
6879 * tree.h (get_typenode_from_name): Prototype.
6881 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
6883 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
6884 Change operand 2 predicate to nonimmediate_operand.
6885 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
6887 2018-10-03 Martin Sebor <msebor@redhat.com>
6888 Jeff Law <law@redhat.com>
6890 * gimple-ssa-sprintf.c (struct fmtresult): Add new member and
6892 (get_string_length): Detect unterminated arrays.
6893 (format_string): Same.
6894 (format_directive): Warn about unterminated arrays.
6895 (handle_gimple_call): Mark statements with no_warning as needed.
6897 2018-10-03 Jim Wilson <jimw@sifive.com>
6899 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): For ABI_ILP32E,
6900 also define __riscv_abi_rve. Delete trailing white space.
6902 2018-10-03 Paul Koning <ni1d@arrl.net>
6904 Enable LRA register allocator for PDP11.
6905 * config/pdp11/constraints.md (Q): Use define_memory_constraint.
6908 * config/pdp11/pdp11.c (pdp11_lra_p): New function.
6909 * config/pdp11/pdp11.opt (-mlra): New option.
6910 * doc/invoke.texi (PDP-11 Options): Document -mlra.
6912 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
6914 * config/i386/i386.md (*<absneg:code>extendsfdf2): Remove.
6915 (*<absneg:code>extend<mode>xf2): Ditto.
6917 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
6919 PR tree-optimization/87415
6920 * tree-vrp.c (set_value_range_with_overflow): Special case one bit
6923 2018-10-02 Jeff Law <law@redhat.com>
6925 * gimple-fold.c (get_range_strlen): Only set *nonstr when
6926 an unterminated string is discovered. Bubble up range
6927 even for unterminated strings.
6928 (gimple_fold_builtin_strlen): Do not fold if get_range_strlen
6929 indicates the string was not terminated via NONSTR.
6931 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
6933 * tree-vrp.c (extract_range_from_unary_expr): Special case all
6934 pointer conversions.
6935 Do not do anything special for anti-ranges.
6937 2018-10-03 Jérôme Lambourg <lambourg@adacore.com>
6939 * config/arm/vxworks.h (ARM_TARGET2_DWARF_FORMAT): Adjust to
6940 DW_EH_PE_pcrel | DW_EH_PE_indirect for RTPs.
6942 2018-10-03 Martin Liska <mliska@suse.cz>
6944 PR gcov-profile/86109
6945 * coverage.c (coverage_begin_function): Do not
6946 mark lambdas as artificial.
6947 * tree-core.h (struct GTY): Remove tm_clone_flag
6948 and introduce new lambda_function.
6949 * tree.h (DECL_LAMBDA_FUNCTION): New macro.
6951 2018-10-02 Aaron Sawdey <acsawdey@linux.ibm.com>
6954 * config/rs6000/rs6000-string.c (expand_strn_compare): Check that both
6955 P8_VECTOR and VSX are enabled.
6957 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
6959 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
6960 0x3907 as CPU model number.
6962 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
6964 * common/config/s390/s390-common.c: Rename PF_ARCH12 to PF_Z14.
6965 * config/s390/s390.h (enum processor_flags): Rename PF_ARCH12 to
6966 PF_Z14. Rename TARGET_CPU_ARCH12 to TARGET_CPU_Z14,
6967 TARGET_CPU_ARCH12_P to TARGET_CPU_Z14_P, TARGET_ARCH12 to
6968 TARGET_Z14, and TARGET_ARCH12_P to TARGET_Z14_P.
6969 * config/s390/s390.md: Likewise. Rename also the cpu attribute
6970 value from arch12 to z14.
6972 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
6974 * config/i386/i386.md (fxam<mode>2_i387_with_temp): Remove.
6976 (isinf<mode>2): Ditto.
6978 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
6980 * config/i386/i386.c (ix86_emit_i387_round): Extend op1 to XFmode
6981 before emitting fxam. Perform calculations in XFmode.
6983 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
6985 * match.pd (((X /[ex] A) +- B) * A): New transformation.
6987 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
6990 * fold-const.c (fold_plusminus_mult_expr): Handle complex and vectors.
6991 * tree.c (signed_or_unsigned_type_for): Handle complex.
6993 2018-10-02 Jeff Law <law@redhat.com>
6995 * gimple-fold.c (get_range_strlen): Remove dead code.
6997 2018-10-02 Martin Sebor <msebor@redhat.com>
6998 Jeff Law <law@redhat.com>
7000 * builtins.c (unterminated_array): Add new arguments.
7001 If argument is not terminated, bubble up size and exact
7003 (expand_builtin_strnlen): Detect, avoid expanding
7004 and diagnose unterminated arrays.
7005 (c_strlen): Fill in offset of start of unterminated strings.
7006 * builtins.h (unterminated_array): Update prototype.
7008 2018-10-02 Richard Biener <rguenther@suse.de>
7010 * config/i386/sse.md (reduc_plus_scal_v4df): Avoid the use
7011 of haddv4df, first reduce to SSE width and exploit the fact
7012 that we only need element zero with the reduction result.
7013 (reduc_plus_scal_v2df): Likewise.
7015 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
7017 * dojump.h (do_jump): Delete.
7018 (do_jump_1): Likewise.
7019 (split_comparison): Move around.
7020 * dojump.c (do_jump): Make static.
7021 (do_jump_1): Likewise.
7022 (jumpifnot): Move around.
7023 (jumpifnot_1): Likewise.
7025 (jumpif_1): Likewise.
7026 * expr.c (expand_expr_real_1): Call jumpif[not] instead of do_jump.
7028 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
7030 * reorg.c (make_return_insns): Use emit_copy_of_insn_after for the
7031 insns in the delay slot and add_insn_after for the jump insn.
7033 2018-10-02 Richard Biener <rguenther@suse.de>
7035 * tree-inline.c (expand_call_inline): Use the location of
7036 the callee declaration for the inline-entry marker.
7037 * final.c (notice_source_line): Remove special-casing of
7038 NOTE_INSN_INLINE_ENTRY.
7040 2018-10-01 Carl Love <cel@us.ibm.com>
7043 * config/rs6000/rs6000-builtin.def (__builtin_mffsl): New.
7044 (__builtin_mtfsb0): New.
7045 (__builtin_mtfsb1): New.
7046 ( __builtin_set_fpscr_rn): New.
7047 (__builtin_set_fpscr_drn): New.
7048 * config/rs6000/rs6000.c (rs6000_expand_mtfsb_builtin): Add.
7049 (rs6000_expand_set_fpscr_rn_builtin): Add.
7050 (rs6000_expand_set_fpscr_drn_builtin): Add.
7051 (rs6000_expand_builtin): Add case statement entries for
7052 RS6000_BUILTIN_MTFSB0, RS6000_BUILTIN_MTFSB1,
7053 RS6000_BUILTIN_SET_FPSCR_RN, RS6000_BUILTIN_SET_FPSCR_DRN,
7054 RS6000_BUILTIN_MFFSL.
7055 (rs6000_init_builtins): Add ftype initialization and def_builtin
7056 calls for __builtin_mffsl, __builtin_mtfsb0, __builtin_mtfsb1,
7057 __builtin_set_fpscr_rn, __builtin_set_fpscr_drn.
7058 * config/rs6000.md (rs6000_mtfsb0, rs6000_mtfsb1, rs6000_mffscrn,
7059 rs6000_mffscdrn): Add define_insn.
7060 (rs6000_set_fpscr_rn, rs6000_set_fpscr_drn): Add define_expand.
7061 * doc/extend.texi: Add documentation for the builtins.
7063 2018-10-01 Richard Biener <rguenther@suse.de>
7065 PR tree-optimization/87465
7066 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Fix typo
7067 causing branch miscounts.
7069 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7071 * common/config/aarch64/aarch64-common.c (TARGET_OPTION_DEFAULT_PARAM,
7072 aarch64_option_default_param): New.
7073 (params.h): Include.
7074 (TARGET_OPTION_VALIDATE_PARAM, aarch64_option_validate_param): New.
7075 * config/aarch64/aarch64.c (aarch64_override_options_internal): Simplify
7076 stack-clash protection validation code.
7078 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7080 * params.c (validate_param): New.
7081 (add_params): Use it.
7082 (set_param_value): Refactor param validation into validate_param.
7083 (diagnostic.h): Include.
7084 * diagnostic.h (diagnostic_ready_p): New.
7086 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7088 * params.c (set_param_value):
7089 Add index of parameter being validated.
7090 * common/common-target.def (option_validate_param): New.
7091 * common/common-targhooks.h (default_option_validate_param): New.
7092 * common/common-targhooks.c (default_option_validate_param): New.
7093 * doc/tm.texi.in (TARGET_OPTION_VALIDATE_PARAM): New.
7094 * doc/tm.texi: Regenerate.
7096 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7099 * config/aarch64/aarch64.c (aarch64_override_options_internal):
7100 Add validation for stack-clash parameters and set defaults.
7102 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7105 * configure.ac: Add stack-clash-protection-guard-size.
7106 * doc/install.texi: Document it.
7107 * config.in (DEFAULT_STK_CLASH_GUARD_SIZE): New.
7108 * params.def: Update comment for guard-size.
7109 (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE,
7110 PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Update description.
7111 * configure: Regenerate.
7113 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7116 * config/aarch64/aarch64.h (STACK_CLASH_MIN_BYTES_OUTGOING_ARGS,
7117 STACK_DYNAMIC_OFFSET): New.
7118 * config/aarch64/aarch64.c (aarch64_layout_frame):
7119 Update outgoing args size.
7120 (aarch64_stack_clash_protection_alloca_probe_range,
7121 TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
7123 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7126 * explow.c (anti_adjust_stack_and_probe_stack_clash): Support custom
7128 * target.def (stack_clash_protection_alloca_probe_range): New.
7129 (stack_clash_protection_final_dynamic_probe): Remove.
7130 * targhooks.h (default_stack_clash_protection_alloca_probe_range) New.
7131 (default_stack_clash_protection_final_dynamic_probe): Remove.
7132 * targhooks.c: Likewise.
7133 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
7134 (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE): Remove.
7135 * doc/tm.texi: Regenerate.
7137 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7140 * config/aarch64/aarch64-protos.h (aarch64_output_probe_sve_stack_clash): New.
7141 * config/aarch64/aarch64.c (aarch64_output_probe_sve_stack_clash,
7142 aarch64_clamp_to_uimm12_shift): New.
7143 (aarch64_allocate_and_probe_stack_space): Add SVE specific section.
7144 * config/aarch64/aarch64.md (probe_sve_stack_clash): New.
7146 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7149 * config/aarch64/aarch64.c (aarch64_layout_frame): Add assert.
7151 2018-10-01 Jeff Law <law@redhat.com>
7152 Richard Sandiford <richard.sandiford@linaro.org>
7153 Tamar Christina <tamar.christina@arm.com>
7156 * config/aarch64/aarch64.md
7157 (probe_stack_range): Add k (SP) constraint.
7158 * config/aarch64/aarch64.h (STACK_CLASH_CALLER_GUARD,
7159 STACK_CLASH_MAX_UNROLL_PAGES): New.
7160 * config/aarch64/aarch64.c (aarch64_output_probe_stack_range): Emit
7161 stack probes for stack clash.
7162 (aarch64_allocate_and_probe_stack_space): New.
7163 (aarch64_expand_prologue): Use it.
7164 (aarch64_expand_epilogue): Likewise and update IP regs re-use criteria.
7165 (aarch64_sub_sp): Add emit_move_imm optional param.
7167 2018-10-01 MCC CS <deswurstes@users.noreply.github.com>
7169 PR tree-optimization/87261
7170 * match.pd: Remove trailing whitespace.
7171 Add (x & y) | ~(x | y) -> ~(x ^ y),
7172 (~x | y) ^ (x ^ y) -> x | ~y and (x ^ y) | ~(x | y) -> ~(x & y)
7174 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
7176 * config/arc/arc.md (*add_n): Clean up pattern, update instruction
7178 (ashlsi3_insn): Update instruction constraints.
7179 (ashrsi3_insn): Likewise.
7180 (rotrsi3): Likewise.
7181 (add_shift): Likewise.
7182 * config/arc/constraints.md (Csz): New 32 bit constraint. It
7183 avoids placing in the limm field small constants which, otherwise,
7184 could end into a small instruction.
7186 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
7188 * config/arc/arc.md (maddsidi4_split): Don't use dmac if the
7189 destination register is not odd-even.
7190 (umaddsidi4_split): Likewise.
7192 2018-10-01 Richard Biener <rguenther@suse.de>
7194 * tree-inline.c (expand_call_inline): Store origin of fn
7195 in BLOCK_ABSTRACT_ORIGIN for the inline BLOCK.
7196 * tree.c (block_ultimate_origin): Simplify and do some
7199 2018-09-30 Uros Bizjak <ubizjak@gmail.com>
7201 * config/i386/mmx.md (EMMS): New int iterator.
7202 (emms): New int attribute.
7203 (mmx_<emms>): Macroize insn from *mmx_emms and *mmx_femms using
7204 EMMS int iterator. Explicitly declare clobbers.
7205 (mmx_emms): Remove expander.
7207 * config/i386/predicates.md (emms_operation): Remove predicate.
7208 (vzeroall_pattern): New predicate.
7209 (vzeroupper_pattern): Rename from vzeroupper_operation.
7210 * config/i386/i386.c (ix86_avx_u128_mode_after): Use
7211 vzeroupper_pattern and vzeroall_pattern predicates.
7213 2018-09-30 Peter Bergner <bergner@linux.ibm.com>
7215 PR rtl-optimization/86939
7216 * ira-lives.c (make_hard_regno_born): Rename from this...
7217 (make_hard_regno_live): ... to this. Remove update to conflict
7218 information. Update function comment.
7219 (make_hard_regno_dead): Add conflict information update. Update
7221 (make_object_born): Rename from this...
7222 (make_object_live): ... to this. Remove update to conflict information.
7223 Update function comment.
7224 (make_object_dead): Add conflict information update. Update function
7226 (mark_pseudo_regno_live): Call make_object_live.
7227 (mark_pseudo_regno_subword_live): Likewise.
7228 (mark_hard_reg_dead): Update function comment.
7229 (mark_hard_reg_live): Call make_hard_regno_live.
7230 (process_bb_node_lives): Likewise.
7231 * lra-lives.c (make_hard_regno_born): Rename from this...
7232 (make_hard_regno_live): ... to this. Remove update to conflict
7233 information. Remove now uneeded check_pic_pseudo_p argument.
7234 Update function comment.
7235 (make_hard_regno_dead): Add check_pic_pseudo_p argument and add update
7236 to conflict information. Update function comment.
7237 (mark_pseudo_live): Remove update to conflict information. Update
7239 (mark_pseudo_dead): Add conflict information update.
7240 (mark_regno_live): Call make_hard_regno_live.
7241 (mark_regno_dead): Call make_hard_regno_dead with new arguement.
7242 (process_bb_lives): Call make_hard_regno_live and make_hard_regno_dead.
7244 2018-09-29 H.J. Lu <hongjiu.lu@intel.com>
7247 * config/i386/i386.c (construct_container): Use TImode for
7248 BLKmode values in 2 integer registers.
7250 2018-09-29 Jeff Law <law@redhat.com>
7252 * builtins.c (unterminated_array): Pass in c_strlen_data * to
7253 c_strlen rather than just a tree *.
7254 (c_strlen): Change NONSTR argument to a c_strlen_data pointer.
7255 Update recursive calls appropriately. If caller did not provide a
7256 suitable data pointer, create a local one. When a non-terminated
7257 string is discovered, bubble up information about the string via the
7258 c_strlen_data object.
7259 * builtins.h (c_strlen): Update prototype.
7260 (c_strlen_data): New structure.
7261 * gimple-fold.c (get_range_strlen): Update calls to c_strlen.
7262 For a type 2 call, if c_strlen indicates a non-terminated string
7263 use the length of the non-terminated string.
7264 (gimple_fold_builtin_stpcpy): Update calls to c_strlen.
7266 2018-09-29 Jakub Jelinek <jakub@redhat.com>
7269 * config/i386/avx512fintrin.h (_mm512_abs_pd, _mm512_mask_abs_pd): Use
7270 __m512d type for __A argument rather than __m512.
7272 2018-09-28 John David Anglin <danglin@gcc.gnu.org>
7274 * match.pd (simple_comparison): Don't optimize if either operand is
7275 a function pointer when target needs function pointer canonicalization.
7277 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
7279 * config/rs6000/driver-rs6000.c (asm_names): Adjust the entries for
7280 power5 .. power9 to remove indirection.
7281 * config/rs6000/rs6000.h (ASM_CPU_POWER5_SPEC, ASM_CPU_POWER6_SPEC,
7282 ASM_CPU_POWER7_SPEC, ASM_CPU_POWER8_SPEC, ASM_CPU_POWER9_SPEC,
7283 ASM_CPU_476_SPEC): Delete.
7284 (ASM_CPU_SPEC): Adjust.
7285 (EXTRA_SPECS): Delete asm_cpu_power5, asm_cpu_power6, asm_cpu_power7,
7286 asm_cpu_power8, asm_cpu_power9, asm_cpu_476.
7288 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
7290 * config.in: Delete HAVE_AS_DCI.
7291 * config/powerpcspe/powerpcspe.h: Treat HAVE_AS_DCI as always true.
7292 * config/rs6000/rs6000.h: Ditto.
7293 * configure.ac: Delete HAVE_AS_DCI.
7294 * configure: Regenerate.
7296 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
7298 * config.in (HAVE_AS_LWSYNC): Delete.
7299 * config/powerpcspe/powerpcspe.h (TARGET_LWSYNC_INSTRUCTION): Delete.
7300 * config/powerpcspe/sync.md (*lwsync): Always generate lwsync, never
7302 * config/rs6000/rs6000.h (TARGET_LWSYNC_INSTRUCTION): Delete.
7303 * config/rs6000/sync.md (*lwsync): Always generate lwsync, never do it
7305 * configure.ac: Delete HAVE_AS_LWSYNC.
7306 * configure: Regenerate.
7308 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
7309 Pierre-Marie de Rodat <derodat@adacore.com>
7311 * calls.c (expand_call): Try to do a tail call for thunks at -O0 too.
7312 * cgraph.h (struct cgraph_thunk_info): Add indirect_offset.
7313 (cgraph_node::create_thunk): Add indirect_offset parameter.
7314 (thunk_adjust): Likewise.
7315 * cgraph.c (cgraph_node::create_thunk): Add indirect_offset parameter
7316 and initialize the corresponding field with it.
7317 (cgraph_node::dump): Dump indirect_offset field.
7318 * cgraphclones.c (duplicate_thunk_for_node): Deal with indirect_offset.
7319 * cgraphunit.c (cgraph_node::analyze): Be prepared for external thunks.
7320 (thunk_adjust): Add indirect_offset parameter and deal with it.
7321 (cgraph_node::expand_thunk): Deal with the indirect_offset field and
7322 pass it to thunk_adjust. Do not call the target hook if it's non-zero
7323 or if the thunk is external or local. Fix formatting. Do not chain
7324 the RESULT_DECL to BLOCK_VARS. Pass the static chain to the target,
7325 if any, in the GIMPLE representation.
7326 * ipa-icf.c (sem_function::equals_wpa): Deal with indirect_offset.
7327 * lto-cgraph.c (lto_output_node): Write indirect_offset field.
7328 (input_node): Read indirect_offset field.
7329 * tree-inline.c (expand_call_inline): Pass indirect_offset field in the
7330 call to thunk_adjust.
7331 * tree-nested.c (struct nesting_info): Add thunk_p field.
7332 (create_nesting_tree): Set it.
7333 (convert_all_function_calls): Copy static chain from targets to thunks.
7334 (finalize_nesting_tree_1): Return early for thunks.
7335 (unnest_nesting_tree_1): Do not finalize thunks.
7336 (gimplify_all_functions): Do not gimplify thunks.
7338 2018-09-28 David Malcolm <dmalcolm@redhat.com>
7340 * opt-suggestions.c (option_proposer::build_option_suggestions):
7341 Release "option_values".
7343 2018-09-28 David Malcolm <dmalcolm@redhat.com>
7345 * coverage.c (get_coverage_counts): Convert problem-reporting dump
7346 messages from MSG_OPTIMIZED_LOCATIONS to MSG_MISSED_OPTIMIZATION.
7347 * dumpfile.c (kind_as_string): New function.
7348 (dump_loc): Rather than a hardcoded prefix of "note: ", use
7349 kind_as_string to vary the prefix based on dump_kind.
7350 (selftest::test_capture_of_dump_calls): Update for above.
7352 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
7354 * config/i386/i386.h (SSE_REGNO): Fix check for FIRST_REX_SSE_REG.
7355 (GET_SSE_REGNO): Rename from SSE_REGNO. Update all uses for rename.
7357 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
7359 * config/i386/i386.h (CC_REGNO): Remove FPSR_REG.
7360 * config/i386/i386.c (ix86_fixed_condition_code_regs): Use
7361 INVALID_REGNUM instead of FPSR_REG.
7362 (ix86_md_asm_adjust): Do not clobber FPSR_REG.
7363 * config/i386/i386.md: Update comment of FP compares.
7364 (fldenv): Do not clobber FPSR_REG.
7366 2018-09-28 Richard Biener <rguenther@suse.de>
7368 * tree.h (BLOCK_ORIGIN): New.
7369 * omp-expand.c (grid_expand_target_grid_body): Assign
7370 BLOCK_ORIGIN to BLOCK_ABSTRACT_ORIGIN.
7371 * tree-inline.c (remap_block): Likewise.
7372 * auto-profile.c (get_function_decl_from_block): Simplify
7373 by eliding the BLOCK_ABSTRACT_ORIGIN chasing.
7374 * langhooks.c (lhd_print_error_function): Likewise.
7375 * optinfo-emit-json.cc (optrecord_json_writer::inlining_chain_to):
7377 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
7378 * tree.c (block_nonartificial_location): Likewise.
7379 (block_ultimate_origin): Likewise.
7380 * tree-pretty-print.c (percent_K_format): Likewise. Remove
7381 no longer needed LTO case.
7383 2018-09-28 Andrew Stubbs <ams@codesourcery.com>
7384 Jan Hubicka <jh@suse.cz>
7385 Martin Jambor <mjambor@suse.cz>
7387 * simplify-rtx.c (simplify_merge_mask): New function.
7388 (simplify_ternary_operation): Use it, also see if VEC_MERGEs with the
7389 same masks are used in op1 or op2.
7390 (test_vec_merge): New function.
7391 (test_vector_ops): Call test_vec_merge.
7393 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
7395 * config/sparc/sparc-protos.h (sparc_branch_cost): Declare.
7396 * config/sparc/sparc.h (BRANCH_COST): Call sparc_branch_cost.
7397 * config/sparc/sparc.c (struct processor_costs): Add branch_cost field.
7398 (cypress_costs): Set it.
7399 (supersparc_costs): Likewise.
7400 (hypersparc_costs): Likewise.
7401 (leon_cost): Likewise.
7402 (leon3_costs): Likewise.
7403 (sparclet_costs): Likewise.
7404 (ultrasparc_costs): Likewise.
7405 (ultrasparc_costs): Likewise.
7406 (niagara_costs): Likewise.
7407 (niagara2_costs): Likewise.
7408 (niagara3_costs): Likewise.
7409 (niagara4_costs): Likewise.
7410 (niagara7_costs): Likewise.
7411 (m8_costs): Likewise.
7412 (TARGET_CAN_FOLLOW_JUMP): Define.
7413 (pass_work_around_errata::gate): Minor tweak.
7414 (sparc_option_override): Remove MASK_FSMULD mask for V7 processors.
7415 Do not set both MASK_VIS4 and MASK_VIS4B for M8 processor.
7416 Automaitcally clear MASK_FSMULD mask for V7 processors.
7417 (sparc_can_follow_jump): New static function.
7418 (output_ubranch): Deal with CROSSING_JUMP_P.
7419 (sparc_use_sched_lookahead): Rewrite using switch statement.
7420 (sparc_issue_rate): Reorder.
7421 (sparc_branch_cost): New function.
7423 2018-09-27 Martin Sebor <msebor@redhat.com>
7425 * tree.h (tree_to_shwi): Add attribute nonnull and pure.
7426 (tree_to_poly_int64, tree_to_uhwi, tree_to_poly_uint64): Same.
7427 (int_fits_type_p): Same.
7429 2018-09-27 Uros Bizjak <ubizjak@gmail.com>
7431 * config/i386/i386.md (FPCR_REG): Remove.
7432 (UNSPEC_FLDCW): Remove.
7433 (x86_fnstcw_1): Use (const_int 0) instead of FPCR_REG.
7434 (x86_fldcw_1): Remove insn pattern.
7435 (fnstenv): Do not clobber FPCR_REG.
7437 * config/i386/i386.h (FIXED_REGISTERS) Remove fpsr register.
7438 (CALL_USED_REGISTERS): Ditto.
7439 (REG_ALLOC_ORDER): Ditto.
7440 (REG_CLASS_CONTENTS): Ditto.
7441 (HI_REGISTER_NAMES): Ditto.
7442 (ADDITIONAL_REGISTER_NAMES): Use defines instead
7443 of numerical constants.
7444 * config/i386/i386.c (regclass_map): Remove fpsr register.
7445 (dbx_register_map): Ditto.
7446 (dbx64_register_map): Ditto.
7447 (svr4_dbx_register_map): Ditto.
7448 (print_reg): Do not handle FPCR_REG.
7450 2018-09-27 Segher Boessenkool <segher@kernel.crashing.org>
7453 * config.in (HAVE_AS_CMPB, HAVE_AS_DFP, HAVE_AS_FPRND, HAVE_AS_MFPGPR,
7454 HAVE_AS_POPCNTB, HAVE_AS_POPCNTD, HAVE_AS_POWER8, HAVE_AS_POWER9):
7455 Delete, always treat as true.
7456 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
7457 Ditto. Simplify remaining code.
7458 * config/powerpcspe/powerpcspe.h: Ditto.
7459 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ditto.
7460 Simplify remaining code.
7461 (rs6000_expand_builtin): Ditto.
7462 * config/rs6000/rs6000.h: Ditto.
7463 * configure.ac: Ditto.
7464 * configure: Regenerate.
7466 2018-09-27 Martin Liska <mliska@suse.cz>
7468 * coverage.c (get_coverage_counts): Revert the formatting
7469 of missing profile opt info.
7471 2018-09-27 Richard Biener <rguenther@suse.de>
7475 * dwarf2out.c (set_block_origin_self): Do not mark outermost
7476 block as we do not output that.
7477 (gen_inlined_subroutine_die): Elide the originally outermost
7478 block, matching what we do for concrete instances.
7479 (decls_for_scope): Add parameter specifying whether to recurse
7482 2018-09-27 Andrew Stubbs <ams@codesourcery.com>
7483 Tom de Vries <tom@codesourcery.com>
7487 * expmed.c (emit_cstore): Fix handling of result_mode == BImode and
7488 STORE_FLAG_VALUE == 1.
7490 2018-09-27 Andreas Krebbel <krebbel@linux.ibm.com>
7492 * config/s390/s390.md (PPA_TX_ABORT, PPA_OOO_BARRIER): New
7493 constant definitions.
7494 ("tx_assist"): Replace magic number with PPA_TX_ABORT.
7495 ("*ppa"): Enable pattern also for -march=zEC12 -mno-htm.
7496 ("speculation_barrier"): New expander definition.
7498 2018-09-26 Indu Bhagat <indu.bhagat@oracle.com>
7500 PR gcov-profile/86957
7501 * common.opt: New warning option -Wmissing-profile.
7502 * coverage.c (get_coverage_counts): Add warning for missing .gcda file.
7503 * doc/invoke.texi: Document -Wmissing-profile.
7505 2018-09-26 Jim Wilson <jimw@sifive.com>
7507 * config/riscv/riscv.md (subsi3_extended2): Add J constraint.
7508 (negdi2, negsi2, negsi2_extended, negsi2_extended2): New.
7510 2018-09-26 Martin Sebor <msebor@redhat.com>
7512 * tree.c (zerop): Change return type to bool.
7513 (integer_zerop, integer_onep, integer_each_onep): Same.
7514 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
7515 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
7516 (real_onep, real_minus_onep, chain_index): Same.
7517 (print_type_hash_statistics, type_list_equal): Same.
7518 * tree.h (zerop): Same.
7519 (zerop, integer_zerop, integer_onep, integer_each_onep): Same.
7520 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
7521 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
7522 (real_onep, real_minus_onep, chain_index): Same.
7523 (print_type_hash_statistics, type_list_equal): Same.
7525 2018-09-26 Jim Wilson <jimw@sifive.com>
7527 * config/riscv/riscv.h (FUNCTION_ARG_REGNO_P): Fix comment.
7529 2018-09-26 Jakub Jelinek <jakub@redhat.com>
7532 * config/i386/i386.c: Include debug.h and dwarf2out.h.
7533 (output_indirect_thunk): Emit DW_CFA_def_cfa_offset after the
7536 2018-09-25 Andrew Stubbs <ams@codesourcery.com>
7538 * builtins.c (get_builtin_sync_mem): Force address mode conversion.
7540 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
7542 * config/i386/i386.h (enum reg_class): Remove FP_TOP_SSE_REGS
7543 and FP_SECOND_SSE_REGS.
7544 (REG_CLASS_NAMES): Ditto.
7545 (REG_CLASS_CONTENTS): Ditto.
7546 * config/i386/i386.c (ix86_preferred_reload_class) Do not handle
7547 FP_TOP_SSE_REGS and FP_SECOND_SSE_REGS classes.
7548 (ix86_preferred_output_reload_class): Ditto.
7549 * config/i386/i386.md (fix_trunc<mode>_i387_fisttp): Change "=&1f"
7550 clobber constraint to "=&f".
7551 (fix_truncdi_i387): Ditto.
7552 (lrintxfdi2): Ditto.
7553 (fistdi2_<rounding>): Ditto.
7554 (fpremxf4_i387): Change "=u" constraint to "=f".
7555 (fprem1xf4_i387): Ditto.
7557 (fptanxf4_i387): Ditto.
7558 (fxtractxf3_i387): Ditto.
7559 (fscalexf4_i387): Ditto.
7560 (atan2xf3): Change "u" constraint to "f".
7561 (fyl2xxf3_i387): Ditto.
7562 (fyl2xp1xf3_i387): Ditto.
7564 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
7567 * config/i386/i386.h (NUM_MODES_FOR_MODE_SWITCHING): Update
7568 for removed I387_MASK_PM entity.
7571 2018-09-26 Jeff Law <law@redhat.com>
7573 2018-09-26 Alexey Neyman <stilor@att.net>
7575 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
7576 headers are no longer pulled in by <isl/val.h>.
7578 2018-09-26 Richard Biener <rguenther@suse.de>
7581 * dwarf2out.c (gen_lexical_block_die): Do not equate inline
7582 or concrete instance DIE to the tree. Create abstract origin
7583 attributes also for concrete instances.
7585 2018-09-26 Alexey Neyman <stilor@att.net>
7587 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
7588 headers are no longer pulled in by <isl/val.h>.
7590 2018-09-26 Matthew Malcomson <matthew.malcomson@arm.com>
7592 * config/arm/arm.c (arm_split_compare_and_swap, arm_split_atomic_op):
7593 Use new helper functions.
7594 * config/arm/sync.md (atomic_load<mode>, atomic_store<mode>):
7595 Use new helper functions.
7596 * config/arm/aarch-common-protos.h (aarch_mm_needs_acquire,
7597 aarch_mm_needs_release): New declarations.
7598 * config/arm/aarch-common.c (aarch_mm_needs_acquire,
7599 aarch_mm_needs_release): New.
7601 2018-09-26 Eric Botcazou <ebotcazou@adacore.com>
7603 * config/arm/arm.c (arm_reorg): Skip Thumb reorg pass for thunks.
7604 (arm32_output_mi_thunk): Deal with long calls.
7606 2018-09-26 Richard Biener <rguenther@suse.de>
7610 * tree-inline.c (expand_call_inline): When the location
7611 of the call is UNKNOWN_LOCATION use DECL_SOURCE_LOCATION
7612 or BUILTINS_LOCATION for the BLOCK_SOURCE_LOCATION of
7613 the inserted BLOCK to make inlined_function_outer_scope_p
7615 * dwarf2out.c (add_call_src_coords_attributes): Do not add
7616 coords for reserved locations.
7618 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
7620 * config/rs6000/rs6000.md (*movcc_internal1): Use set_attr_alternative.
7621 (*call_indirect_nonlocal_sysv<mode>): Ditto.
7622 (*call_value_indirect_nonlocal_sysv<mode>): Ditto.
7623 (*sibcall_nonlocal_sysv<mode>): Ditto.
7624 (*sibcall_value_nonlocal_sysv<mode>): Ditto.
7625 (<bd>_<mode>): Ditto.
7626 (<bd>tf_<mode>): Ditto.
7628 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
7630 * config/rs6000/altivec.md (*altivec_mov<mode>): Write the output
7631 control string as a list of templates instead of as C code.
7632 (*altivec_movti): Ditto.
7633 * config/rs6000/darwin.md (movdf_low_di): Ditto.
7635 2018-09-25 Jim Wilson <jimw@sifive.com>
7637 * config/riscv/riscv.c (riscv_split_symbol): Mark auipc label as weak
7638 when target symbol is weak.
7640 2018-09-25 Bernd Edlinger <bernd.edlinger@hotmail.de>
7643 * builtins.c (unterminated_array): Simplify.
7644 * expr.c (string_constant): Handle SSA_NAME. Add more exceptions
7645 where pointer arithmetic is safe.
7647 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
7650 * config/rs6000/altivec.md (altivec_vspltb): Use
7651 const_0_to_15_operand instead of u5bit_cint_operand.
7652 (*altivec_vspltb_internal): Ditto.
7653 (altivec_vspltb_direct): Ditto.
7654 (altivec_vsplth): Use const_0_to_7_operand instead of
7656 (*altivec_vsplth_internal): Ditto.
7657 (altivec_vsplth_direct): Ditto.
7658 (altivec_vspltw): Use const_0_to_3_operand instead of
7660 (*altivec_vspltw_internal): Ditto.
7661 (altivec_vspltw_direct): Ditto.
7662 (altivec_vspltsf): Ditto.
7663 (*altivec_vspltsf_internal): Ditto.
7664 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Handle the
7665 various splats with the proper size immediate. Reorder the various
7666 cases by ascending size of immediate, and put all such together.
7668 2018-09-25 Richard Biener <rguenther@suse.de>
7671 * dwarf2out.c (add_AT_external_die_ref): Remove now redundant
7673 (maybe_create_die_with_external_ref): Do not create
7674 DW_TAG_imported_unit here.
7675 (add_abstract_origin_attribute): Handle external BLOCK refs.
7676 (dwarf2out_abstract_function): Simplify LTO case.
7677 (dwarf2out_early_finish): Create DW_TAG_imported_unit explicitely
7678 rather than using maybe_create_die_with_external_ref.
7680 2018-09-25 Uros Bizjak <ubizjak@gmail.com>
7683 * config/i386/i386.md (frndintxf2_mask_pm): Remove.
7684 (frndintxf2_mask_pm_i387): Ditto.
7685 (nearbyintxf2): Rewrite expander pattern to match rintxf2.
7686 Enable for !flag_trapping_math.
7687 (nearbyint<mode>2): Enable x87 modes for !flag_trapping_math.
7688 Enable SSE modes for TARGET_SSE4_1 and expand them with round insn.
7689 Change operand 1 predicate to nonimmediate_operand.
7690 (attr "i387_cw"): Remove mask_pm.
7691 * config/i386/i386.h (enum ix86_stack_slot): Remove SLOT_CW_MASK_PM.
7692 (enum ix86_entity): Remove I387_MASK_PM.
7693 * config/i386/i386.c (ix86_i387_mode_needed): Do not
7694 handle I387_MASK_PM.
7695 (ix86_mode_needed): Ditto.
7696 (ix86_mode_after): Ditto.
7697 (ix86_mode_entry): Ditto.
7698 (ix86_mode_exit): Ditto.
7699 (emit_i387_cw_initialization): Do not handle I387_CW_MASK_PM.
7701 2018-09-25 Jakub Jelinek <jakub@redhat.com>
7703 * vr-values.c (vr_values::vr_values): Initialize to_remove_edges and
7704 to_update_switch_stmts to vNULL instead of calling create on them
7707 2018-09-25 Richard Biener <rguenther@suse.de>
7709 PR tree-optimization/87402
7710 * tree-ssa-sccvn.c (SSA_VISITED): Remove unused function.
7711 (visit_phi): Re-instantiate handling of supposed to be VARYING
7712 but non-VARYING backedge value.
7714 2018-09-25 Richard Biener <rguenther@suse.de>
7717 * dwarf2out.c (struct sym_off_pair): New.
7718 (external_die_map): New global.
7719 (lookup_decl_die): When in LTO create DIEs lazily from the
7721 (lookup_block_die): New function, create DIEs lazily in LTO.
7722 (equate_block_to_die): New function.
7723 (dwarf2out_die_ref_for_decl): During WPA get the association
7724 from the external DIE map.
7725 (dwarf2out_register_external_die): Record mapping into the
7727 (maybe_create_die_with_external_ref): New function split out from
7728 DIE generation part of old dwarf2out_register_external_die.
7729 (add_abstract_origin_attribute): Do not return the DIE. When
7730 in LTO reference externals directly.
7731 (dwarf2out_abstract_function): When in LTO ignore calls for
7732 decls with external DIEs (already present abstract instances).
7733 (gen_call_site_die): Adjust.
7734 (add_high_low_attributes): Likewise.
7735 (gen_lexical_block_die): Likewise.
7736 (gen_inlined_subroutine_die): Likewie.
7737 (gen_block_die): Likewise.
7738 (dwarf2out_inline_entry): Likewise.
7739 (dwarf2out_early_finish): In LTRANS phase create DW_TAG_imported_unit
7742 2018-09-25 Martin Liska <mliska@suse.cz>
7744 * ipa-fnsummary.c (estimate_node_size_and_time): Scale by two
7745 integers and not by a float value.
7747 2018-09-25 Martin Liska <mliska@suse.cz>
7750 * dbgcnt.c (dbg_cnt_process_single_pair): Return false
7752 * dumpfile.c (dump_enable_all): Remove extra parenthesis.
7753 * gcov-tool.c: Declare the function with ATTRIBUTE_NORETURN.
7754 * godump.c (go_format_type): Remove extra parenthesis.
7756 2018-09-25 Martin Liska <mliska@suse.cz>
7758 * alias.c (set_dest_equal_p): Remove unused function.
7759 * config/i386/i386.c (def_builtin_pure2): Likewise.
7760 * diagnostic-show-locus.c (class layout): Remove
7762 (layout::layout): Likewise here.
7763 * dump-context.h (class temp_dump_context): Likewise.
7764 * dwarf2out.c (add_AT_fde_ref): Remove unused function.
7765 (add_AT_loclistsptr): Likewise.
7766 (add_AT_offset): Likewise.
7767 (get_AT_hi_pc): Likewise.
7768 (is_comdat_die): Likewise.
7769 (type_is_enum): Likewise.
7770 (ceiling): Likewise.
7771 (add_AT_vms_delta): Likewise.
7772 (is_class_die): Likewise.
7773 * edit-context.c (class line_event): Remove unused field.
7774 * graphite-sese-to-poly.c (tree_int_to_gmp): Remove
7776 * ipa-cp.c (ipa_get_vr_lat): Likewise.
7777 * lra-constraints.c (ok_for_index_p_nonstrict): Likewise.
7778 (ok_for_base_p_nonstrict): Likewise.
7779 * tree-chrec.c (is_not_constant_evolution): Likewise.
7780 (chrec_fold_poly_cst): Likewise.
7781 * tree-if-conv.c (has_pred_critical_p): Likewise.
7782 * tree-ssa-coalesce.c (print_exprs): Likewise.
7783 * tree-ssa-pre.c (bitmap_set_contains_expr): Likewise.
7784 * tree-ssa-uninit.c (is_and_or_or_p): Likewise.
7785 * tree-vrp.c (value_ranges_intersect_p): Likewise.
7786 (value_range_nonnegative_p): Likewise.
7788 2018-09-25 Martin Liska <mliska@suse.cz>
7790 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
7791 Do not handle "GNU Pascal".
7792 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
7794 * config/sparc/sparc.c (sparc_pass_by_reference): Remove Pascal
7795 from documentation. Likewise.
7796 * dbxout.c (dbxout_range_type): Likewise.
7797 * doc/cpp.texi: Likewise.
7798 * doc/extend.texi: Likewise.
7799 * doc/frontends.texi: Likewise.
7800 * doc/invoke.texi: Remove Pascal entry.
7801 * tree.def (CLEANUP_POINT_EXPR): Likewise.
7802 * doc/rtl.texi (MODE_FUNCTION): Remove not used entry.
7804 2018-09-25 Martin Liska <mliska@suse.cz>
7807 * doc/invoke.texi: Document all parameters and remove default
7810 2018-09-25 Ilya Leoshkevich <iii@linux.ibm.com>
7813 * rtl.c (rtx_code_size): Take into account that EQ_ATTR_ALT
7814 contains HOST_WIDE_INTs when computing its size.
7816 2018-09-24 Jim Wilson <jimw@sifive.com>
7819 * config/riscv/riscv.h (STACK_BOUNDARY): Test riscv_abi == ABI_ILP32E
7821 (ABI_STACK_BOUNDARY, MAX_ARGS_IN_REGISTERS): Likewise.
7823 2018-09-24 Andrew Pinski <apinski@marvell.com>
7825 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Don't
7826 access prev before checking it for NULLness in the
7827 AARCH64_FUSE_CMP_BRANCH case.
7829 2018-09-24 H.J. Lu <hongjiu.lu@intel.com>
7832 * config/i386/i386.c (rest_of_insert_endbranch): Set
7833 endbr_queued_at_entrance to true and don't insert ENDBR if
7834 x86_function_profiler will be called.
7835 (x86_function_profiler): Insert ENDBR if endbr_queued_at_entrance
7837 * config/i386/i386.h (machine_function): Add
7838 endbr_queued_at_entrance.
7840 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
7842 * genattrtab.c (mk_attr_alt): Use alternative_mask.
7843 (attr_rtx_1): Adjust caching to match the new EQ_ATTR_ALT field
7845 (check_attr_test): Use alternative_mask.
7846 (get_attr_value): Likewise.
7847 (compute_alternative_mask): Use alternative_mask and XWINT.
7848 (make_alternative_compare): Use alternative_mask.
7849 (attr_alt_subset_p): Use XWINT.
7850 (attr_alt_subset_of_compl_p): Likewise.
7851 (attr_alt_intersection): Use alternative_mask and XWINT.
7852 (attr_alt_union): Likewise.
7853 (attr_alt_complement): Use HOST_WIDE_INT and XWINT.
7854 (mk_attr_alt): Use alternative_mask and HOST_WIDE_INT.
7855 (simplify_test_exp): Use alternative_mask and XWINT.
7856 (write_test_expr): Use alternative_mask and XWINT, adjust bit
7857 number calculation to support 64 bits. Generate code that
7858 checks 64-bit masks.
7859 (main): Use alternative_mask.
7860 * rtl.def (EQ_ATTR_ALT): Change field types from ii to ww.
7862 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
7865 * config/s390/s390.c (s390_emit_epilogue): Do not use PARALLEL
7866 RETURN+USE when returning via %r14.
7868 2018-09-24 Martin Liska <mliska@suse.cz>
7870 * gcov.c (output_lines): Print colorization legend
7871 for both flag_use_colors and flag_use_hotness_colors.
7874 2018-09-24 Martin Liska <mliska@suse.cz>
7876 * coverage.c (get_coverage_counts): Use warning_at
7877 with current_function_decl location. Use %qD in warning
7880 2018-09-24 Martin Liska <mliska@suse.cz>
7882 * memory-block.h (memory_block_pool::release): Annotate with
7883 valgrind that the memory is not accessible.
7885 2018-09-24 Martin Liska <mliska@suse.cz>
7888 * asan.c: Make asan_handled_variables extern.
7890 * cfgexpand.c (expand_stack_vars): Make sure
7891 a representative is unpoison if another
7892 variable in the partition is handled by
7893 use-after-scope sanitization.
7895 2018-09-24 Richard Biener <rguenther@suse.de>
7897 PR tree-optimization/63155
7898 * tree-ssa-propagate.c (add_ssa_edge): Avoid adding PHIs to
7899 the worklist when the edge of the respective argument isn't
7902 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
7904 * config/i386/i386.h (enum reg_class): Rename MASK_REGS to
7905 ALL_MASK_REGS and MASK_EVEX_REGS to MASK_REGS.
7906 (MASK_CLASS_P): Update for rename.
7907 (MAYBE_MASK_CLASS_P): Ditto.
7908 (REG_CLASS_NAMES): Update.
7909 (REG_CLASS_CONTENT): Update.
7910 * config/i386/i386.c (regclass_map): Update for MASK_REG
7911 and ALL_MASK_REGS rename.
7912 * config/i386/constraints.md (Yk): Update for rename.
7915 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
7917 * config/i386/i386.h (enum reg_class): Remove
7918 EVEX_SSE_REGS and MOD4_SSE_REGS.
7919 (REG_CLASS_NAMES): Update.
7920 (REG_CLASS_CONTENT): Update.
7921 * config/i386/i386.c (regclass_map): Declare AVX-512 SSE
7922 registers as ALL_SSE_REGS.
7923 (ix86_additional_allocno_class_p): Remove.
7924 (TARGET_ADDITIONAL_ALLOCNO_CLASS_P): Remove.
7925 (ix86_register_priority): Lower priority of EVEX SSE registers.
7926 Use IN_RANGE macro where appropriate.
7927 (ix86_hard_regno_mode_ok): Merge AVX-5124FMAPS and
7928 AVX-5124VNNIW checks.
7929 (ix86_modes_tieable_p): Tie 512-bit SSE modes.
7930 * config/i386/sse.md (avx5124fmaddps_4fmaddps)
7931 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddps_maskz)
7932 (avx5124fmaddps_4fmaddss, avx5124fmaddps_4fmaddss_mask)
7933 (avx5124fmaddps_4fmaddss_maskz, avx5124fmaddps_4fnmaddps)
7934 (avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddps_maskz)
7935 (avx5124fmaddps_4fnmaddss, avx5124fmaddps_4fnmaddss_mask)
7936 (avx5124fmaddps_4fnmaddss_maskz, avx5124vnniw_vp4dpwssd)
7937 (avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssd_maskz)
7938 (avx5124vnniw_vp4dpwssds, avx5124vnniw_vp4dpwssds_mask)
7939 (avx5124vnniw_vp4dpwssds_maskz): Use "v" instead of "Yh" constraint.
7940 * config/i386/constraints.md (Yh): Remove.
7942 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
7944 * config/i386/i386.c (regclass_map): Declare integer REX registers
7947 2018-09-23 Gerald Pfeifer <gerald@pfeifer.com>
7949 * doc/service.texi (Service): Switch the fsf.org link to https.
7951 2018-09-22 Chung-Ju Wu <jasonwucj@gmail.com>
7954 * config/nds32/nds32.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7955 Define to speculation_safe_value_not_needed.
7957 2018-09-21 Florian Weimer <fweimer@redhat.com>
7960 * doc/extend.texi (Common Function Attributes): Mention that
7961 noreturn suppresses tail call optimization.
7963 2018-09-21 Jeff Law <law@redhat.com>
7965 * gimple-ssa-evrp.c (evrp_dom_walker::cleanup): Call
7966 vr_values::cleanup_edges_and_switches.
7967 * tree-vrp.c (to_remove_edges, to_update_switch_stmts): Moved into
7969 (identify_jump_threads): Remove EDGE_IGNORE handling.
7970 (execute_vrp): Move handling of to_remove_edges and
7971 to_update_switch_stmts into vr_values class member functions.
7972 * tree-vrp.h (switch_update, to_remove_edges): Remove declarations.
7973 (to_update_switch_stmts): Likewise.
7974 * vr-values.c: Include cfghooks.h.
7975 (vr_values::vr_values): Initialize to_remove_edges and
7976 to_update_switch_stmts.
7977 (vr_values::~vr_values): Verify to_remove_edges and
7978 to_update_switch_stmts are empty.
7979 (vr_values::simplify_switch_using_ranges): Set EDGE_IGNORE as needed.
7980 (vr_values::cleanup_edges_and_switches): New member function.
7981 * vr-values.h (vr_values): Add cleanup_edges_and_switches member
7982 function. Add new data members.
7984 2018-09-21 David Malcolm <dmalcolm@redhat.com>
7986 PR tree-optimization/87309
7987 * dumpfile.c (dump_context::begin_scope): Filter the dump_loc
7988 calls with pflags and alt_flags.
7989 (selftest::test_capture_of_dump_calls): Add test of interaction of
7990 MSG_OPTIMIZED_LOCATIONS with AUTO_DUMP_SCOPE.
7992 2018-09-21 Olivier Hainque <hainque@adacore.com>
7994 * config.gcc: Factorize and comment inclusion of vxworks-dummy.h.
7996 2018-09-21 Olivier Hainque <hainque@adacore.com>
7998 * config/vxworks.h (CLEAR_INSN_CACHE): #define to 1.
8000 2018-09-21 Olivier Hainque <hainque@adacore.com>
8002 * config/vxworks.h (VXWORKS_LIBGCC_SPEC): Remove -lc_internal.
8003 Merge block comment with the one ahead of VXWORKS_LIBS_RTP. Then:
8004 (VXWORKS_LIBS_RTP): Minor reordering.
8006 2018-09-21 Olivier Hainque <hainque@adacore.com>
8008 * config/vxworks.h (STARTFILE_PREFIX_SPEC): Define.
8009 (VXWORKS_LIBS_DIR_RTP): Remove definition and use.
8011 2018-09-21 Olivier Hainque <hainque@adacore.com>
8013 * config/vxworks.h (SIZE_TYPE): Account for TARGET_VXWORKS64.
8014 (PTRDIFF_TYPE): Likewise.
8016 2018-09-21 Olivier Hainque <hainque@adacore.com>
8018 * config.gcc: Enforce def of TARGET_VXWORKS64 to 1 from
8019 triplet, similar to support for VxWorks7.
8020 * config/vxworks-dummy.h: Provide a default definition
8021 of TARGET_VXWORKS64 to 0.
8023 2018-09-21 Olivier Hainque <hainque@adacore.com>
8025 * config/vxworks.h (TARGET_VXWORKS7): Move default definition ...
8026 * config/vxworks-dummy.h: here.
8028 2018-09-21 Olivier Hainque <hainque@adacore.com>
8030 * config.gcc: Prepend vxworks-dummy.h to tm_file for powerpc*
8032 2018-09-21 Shaokun Zhang <zhangshaokun@hisilicon.com>
8033 Bo Zhou <zbo.zhou@hisilicon.com>
8035 * config/aarch64/aarch64-cores.def (tsv110): New CPU.
8036 * config/aarch64/aarch64-tune.md: Regenerated.
8037 * doc/invoke.texi (AArch64 Options/-mtune): Add "tsv110".
8038 * config/aarch64/aarch64.c (tsv110_tunings): New tuning table.
8039 * config/aarch64/aarch64-cost-tables.h: Add "tsv110" extra costs.
8041 2018-09-21 Andrew Stubbs <ams@codesourcery.com>
8042 Julian Brown <julian@codesourcery.com>
8044 * builtins.c (get_builtin_sync_mem): Handle address spaces.
8046 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
8048 * config/rs6000/rs6000.c (rs6000_function_ok_for_sibcall): Return false
8049 if the call takes a static chain.
8051 2018-09-21 Martin Liska <mliska@suse.cz>
8053 * auto-profile.c (autofdo_source_profile::read): Do not
8055 (read_profile): Do not add working sets.
8056 (read_autofdo_file): Remove sum_all.
8057 (afdo_callsite_hot_enough_for_early_inline): Remove const
8059 * coverage.c (struct counts_entry): Remove gcov_summary.
8060 (read_counts_file): Read new GCOV_TAG_OBJECT_SUMMARY,
8061 do not support GCOV_TAG_PROGRAM_SUMMARY.
8062 (get_coverage_counts): Remove summary and expected
8064 * coverage.h (get_coverage_counts): Likewise.
8065 * doc/gcov-dump.texi: Remove -w option.
8066 * gcov-dump.c (dump_working_sets): Remove.
8067 (main): Do not support '-w' option.
8068 (print_usage): Likewise.
8069 (tag_summary): Likewise.
8070 * gcov-io.c (gcov_write_summary): Do not dump
8072 (gcov_read_summary): Likewise.
8073 (gcov_histo_index): Remove.
8074 (gcov_histogram_merge): Likewise.
8075 (compute_working_sets): Likewise.
8076 * gcov-io.h (GCOV_TAG_OBJECT_SUMMARY): Mark
8078 (GCOV_TAG_PROGRAM_SUMMARY): Mark it obsolete.
8079 (GCOV_TAG_SUMMARY_LENGTH): Adjust.
8080 (GCOV_HISTOGRAM_SIZE): Remove.
8081 (GCOV_HISTOGRAM_BITVECTOR_SIZE): Likewise.
8082 (struct gcov_summary): Simplify rapidly just
8083 to runs and sum_max fields.
8084 (gcov_histo_index): Remove.
8085 (NUM_GCOV_WORKING_SETS): Likewise.
8086 (compute_working_sets): Likewise.
8087 * gcov-tool.c (print_overlap_usage_message): Remove
8088 trailing empty line.
8089 * gcov.c (read_count_file): Read GCOV_TAG_OBJECT_SUMMARY.
8090 (output_lines): Remove program related line.
8091 * ipa-profile.c (ipa_profile): Do not consider GCOV histogram.
8092 * lto-cgraph.c (output_profile_summary): Do not stream GCOV
8094 (input_profile_summary): Do not read it.
8095 (merge_profile_summaries): And do not merge it.
8096 (input_symtab): Do not call removed function.
8097 * modulo-sched.c (sms_schedule): Do not print sum_max.
8098 * params.def (HOT_BB_COUNT_FRACTION): Reincarnate param that was
8099 removed when histogram method was invented.
8100 (HOT_BB_COUNT_WS_PERMILLE): Mention that it's used only in LTO
8102 * postreload-gcse.c (eliminate_partially_redundant_load): Fix
8104 * predict.c (get_hot_bb_threshold): Use HOT_BB_COUNT_FRACTION
8105 and dump selected value.
8106 * profile.c (add_working_set): Remove.
8107 (get_working_sets): Likewise.
8108 (find_working_set): Likewise.
8109 (get_exec_counts): Do not work with working sets.
8110 (read_profile_edge_counts): Do not inform as sum_max is removed.
8111 (compute_branch_probabilities): Likewise.
8112 (compute_value_histograms): Remove argument for call of
8113 get_coverage_counts.
8114 * profile.h: Do not make gcov_summary const.
8116 2018-09-21 Monk Chiang <sh.chiang04@gmail.com>
8118 * config.gcc (nds32*-*-*): Set TARGET_DEFAULT_TLSDESC_TRAMPOLINE=0.
8120 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
8122 PR tree-optimization/86990
8123 * gimple-ssa-store-merging.c (imm_store_chain_info:coalesce_immediate):
8124 Check that the entire merged store group is made of constants only for
8127 2018-09-20 Allan Sandfeld Jensen <allan.jensen@qt.io>
8129 * gcc.c (LINK_COMMAND_SPEC): Handle -r like -nostdlib.
8130 (VTABLE_VERIFICATION_SPEC): Likewise.
8131 (SANITIZER_EARLY_SPEC): Likewise.
8132 (SANITIZER_SPEC): Likewise.
8133 * config/darwin.h (LINK_COMMAND_SPEC): Likewise.
8134 * doc/invoke.texi (Link Options): Document -r.
8136 2018-09-20 Richard Biener <rguenther@suse.de>
8139 * gimplify.c (gimplify_expr): Retain alignment of
8140 addressable lvalue in dereference.
8142 2018-09-20 Alexandre Oliva <aoliva@redhat.com>
8145 * configure.ac: Check for .loc is_stmt support.
8146 * configure, config.in: Rebuilt.
8147 * dwarf2out.c (dwarf2out_source_line): Skip is_stmt
8150 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
8152 * config/rs6000/rs6000.opt (misel=no, misel=yes): Delete.
8153 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -misel=yes and
8156 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
8158 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
8160 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete
8163 2018-09-20 Marek Polacek <polacek@redhat.com>
8165 * doc/invoke.texi: Add -Wno-init-list-lifetime to C++ Language Options.
8167 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
8169 PR tree-optimization/87288
8170 * tree-vect-loop.c (vect_analyze_loop_2): Take PEELING_FOR_GAPS
8171 into account when determining PEELING_FOR_NITERS.
8173 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
8175 PR tree-optimization/86877
8176 * tree-vect-loop.c (vect_analyze_loop_2): Call
8177 vect_verify_datarefs_alignment.
8179 2018-09-19 Marek Polacek <polacek@redhat.com>
8181 * doc/invoke.texi: Document -Wclass-conversion.
8183 2018-09-19 John David Anglin <danglin@gcc.gnu.org>
8185 * config/pa/pa.c (pa_adjust_priority): Delete.
8186 (TARGET_SCHED_ADJUST_PRIORITY): Delete define.
8188 * config/pa/pa.md (atomic_storeqi): Restore deleted expander.
8189 (atomic_storehi): Likewise.
8190 (atomic_storesi): Likewise.
8191 (atomic_loaddi): Restore compare and swap exchange loop code.
8193 2018-09-19 Segher Boessenkool <segher@kernel.crashing.org>
8195 PR rtl-optimization/86902
8196 * combine.c (try_combine): When changing the CC mode used, don't change
8197 an unrelated mode in other_insn to that new CC mode.
8199 2018-09-19 David Malcolm <dmalcolm@redhat.com>
8201 * tree-data-ref.c (runtime_alias_check_p): Use formatted printing
8202 with %T in place of calls to dump_generic_expr.
8203 (prune_runtime_alias_test_list): Likewise.
8204 (create_runtime_alias_checks): Likewise.
8205 * tree-vect-data-refs.c (vect_check_nonzero_value): Likewise.
8206 (vect_analyze_data_ref_dependence): Likewise.
8207 (vect_slp_analyze_data_ref_dependence): Likewise.
8208 (vect_record_base_alignment): Likewise. Use %G in place of call
8209 to dump_gimple_stmt.
8210 (vect_compute_data_ref_alignment): Likewise.
8211 (verify_data_ref_alignment): Likewise.
8212 (vect_find_same_alignment_drs): Likewise.
8213 (vect_analyze_group_access_1): Likewise.
8214 (vect_analyze_data_ref_accesses): Likewise.
8215 (dependence_distance_ge_vf): Likewise.
8216 (dump_lower_bound): Likewise.
8217 (vect_prune_runtime_alias_test_list): Likewise.
8218 (vect_find_stmt_data_reference): Likewise.
8219 (vect_analyze_data_refs): Likewise.
8220 (vect_create_addr_base_for_vector_ref): Likewise.
8221 (vect_create_data_ref_ptr): Likewise.
8222 * tree-vect-loop-manip.c (vect_set_loop_condition): Likewise.
8223 (vect_can_advance_ivs_p): Likewise.
8224 (vect_update_ivs_after_vectorizer): Likewise.
8225 (vect_gen_prolog_loop_niters): Likewise.
8226 (vect_prepare_for_masked_peels): Likewise.
8227 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
8228 (vect_determine_vectorization_factor): Likewise.
8229 (vect_is_simple_iv_evolution): Likewise.
8230 (vect_analyze_scalar_cycles_1): Likewise.
8231 (vect_analyze_loop_operations): Likewise.
8232 (report_vect_op): Likewise.
8233 (vect_is_slp_reduction): Likewise.
8234 (check_reduction_path): Likewise.
8235 (vect_is_simple_reduction): Likewise.
8236 (vect_create_epilog_for_reduction): Likewise.
8237 (vect_finalize_reduction:): Likewise.
8238 (vectorizable_induction): Likewise.
8239 (vect_transform_loop_stmt): Likewise.
8240 (vect_transform_loop): Likewise.
8241 (optimize_mask_stores): Likewise.
8242 * tree-vect-patterns.c (vect_pattern_detected): Likewise.
8243 (vect_split_statement): Likewise.
8244 (vect_recog_over_widening_pattern): Likewise.
8245 (vect_recog_average_pattern): Likewise.
8246 (vect_determine_min_output_precision_1): Likewise.
8247 (vect_determine_precisions_from_range): Likewise.
8248 (vect_determine_precisions_from_users): Likewise.
8249 (vect_mark_pattern_stmts): Likewise.
8250 (vect_pattern_recog_1): Likewise.
8251 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
8252 (vect_record_max_nunits): Likewise.
8253 (vect_build_slp_tree_1): Likewise.
8254 (vect_build_slp_tree_2): Likewise.
8255 (vect_print_slp_tree): Likewise.
8256 (vect_analyze_slp_instance): Likewise.
8257 (vect_detect_hybrid_slp_stmts): Likewise.
8258 (vect_detect_hybrid_slp_1): Likewise.
8259 (vect_slp_analyze_operations): Likewise.
8260 (vect_slp_analyze_bb_1): Likewise.
8261 (vect_transform_slp_perm_load): Likewise.
8262 (vect_schedule_slp_instance): Likewise.
8263 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
8264 (vect_mark_stmts_to_be_vectorized): Likewise.
8265 (vect_init_vector_1): Likewise.
8266 (vect_get_vec_def_for_operand): Likewise.
8267 (vect_finish_stmt_generation_1): Likewise.
8268 (vect_check_load_store_mask): Likewise.
8269 (vectorizable_call): Likewise.
8270 (vectorizable_conversion): Likewise.
8271 (vectorizable_operation): Likewise.
8272 (vectorizable_load): Likewise.
8273 (vect_analyze_stmt): Likewise.
8274 (vect_is_simple_use): Likewise.
8275 (vect_get_vector_types_for_stmt): Likewise.
8276 (vect_get_mask_type_for_stmt): Likewise.
8277 * tree-vectorizer.c (increase_alignment): Likewise.
8279 2018-09-19 Andrew Stubbs <ams@codesourcery.com>
8281 * doc/rtl.texi: Adjust vec_select description.
8282 * simplify-rtx.c (simplify_binary_operation_1): Allow VEC_SELECT to use
8283 non-constant selectors.
8285 2018-09-19 Matthew Malcomson <matthew.malcomson@arm.com>
8287 * config/aarch64/aarch64-protos.h
8288 (aarch64_offset_9bit_signed_unscaled_p): New declaration.
8289 * config/aarch64/aarch64.md (arches): New "rcpc8_4" attribute value.
8290 (arch_enabled): Add check for "rcpc8_4" attribute value of "arch".
8291 * config/aarch64/aarch64.h (AARCH64_FL_RCPC8_4): New bitfield.
8292 (AARCH64_FL_FOR_ARCH8_4): Include AARCH64_FL_RCPC8_4.
8293 (AARCH64_FL_PROFILE): Move index so flags are ordered.
8294 (AARCH64_ISA_RCPC8_4): New flag.
8295 * config/aarch64/aarch64.c (offset_9bit_signed_unscaled_p): Renamed
8296 to aarch64_offset_9bit_signed_unscaled_p.
8297 * config/aarch64/atomics.md (atomic_store<mode>): Allow offset
8299 * config/aarch64/constraints.md (Ust): New constraint.
8300 * config/aarch64/predicates.md.
8301 (aarch64_9bit_offset_memory_operand): New predicate.
8302 (aarch64_rcpc_memory_operand): New predicate.
8304 2018-09-19 Eric Botcazou <ebotcazou@adacore.com>
8306 PR rtl-optimization/87361
8307 * rtlanal.c (nonzero_bits1): Revert accidental change.
8309 2018-09-19 Richard Biener <rguenther@suse.de>
8311 PR tree-optimization/87349
8312 PR tree-optimization/87342
8313 * tree-ssa-sccvn.c (do_rpo_vn): Iterate max_rpo computation.
8315 2018-09-18 Marek Polacek <polacek@redhat.com>
8317 P1064R0 - Allowing Virtual Function Calls in Constant Expressions
8318 * gimple-fold.c (gimple_get_virt_method_for_vtable): Adjust assert.
8320 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
8322 * config/rs6000/rs6000.md: Remove old "Cygnus sibcall" comment.
8324 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
8326 PR rtl-optimization/86882
8327 * rtlanal.c (reg_overlap_mentioned_p): Handle CLOBBER.
8329 2018-09-18 Uros Bizjak <ubizjak@gmail.com>
8331 * config/i386/i386.md (*<code>extend<mode>xf2): Macroize insn from
8332 *<code>extendsfxf2 and *<code>extenddfxf2 using MODEF mode iterator.
8334 2018-09-18 Jonathan Wakely <jwakely@redhat.com>
8337 * doc/invoke.texi (Link Options): Fix formatting and grammar.
8339 2018-09-18 Richard Biener <rguenther@suse.de>
8342 * tree-ssa-coalesce.c (tree_int_map_hasher): Remove.
8343 (compute_samebase_partition_bases): Likewise.
8344 (coalesce_ssa_name): Always use compute_optimized_partition_bases.
8345 (gimple_can_coalesce_p): Simplify.
8347 2018-09-18 Hans-Peter Nilsson <hp@bitrange.com>
8349 Handle a library implementation of ffs calling __builtin_ffs.
8350 * config/mmix/mmix.c (TARGET_INIT_LIBFUNCS): Override with...
8351 (mmix_init_libfuncs): New function: make __builtin_ffs expand
8354 2018-09-17 David Malcolm <dmalcolm@redhat.com>
8356 * diagnostic-show-locus.c (class layout_range): Add field
8358 (layout_range::layout_range): Add "original_idx" param and use it
8359 to initialize new field.
8360 (make_range): Use 0 for original_idx.
8361 (layout::layout): Pass in index to calls to
8362 maybe_add_location_range.
8363 (layout::maybe_add_location_range): Add param "original_idx" and
8364 pass it on to layout_range.
8365 (layout::print_any_labels): Pass on range->m_original_idx to
8367 (gcc_rich_location::add_location_if_nearby): Use 0 for
8369 * gcc-rich-location.h (text_range_label::get_text): Update for new
8371 (range_label_for_type_mismatch::get_text): Likewise.
8373 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
8375 * config/i386/i386.c (ix86_emit_i387_log1p): Emit fldln2 earlier.
8377 2018-09-17 David Malcolm <dmalcolm@redhat.com>
8379 * gimple-ssa-sprintf.c (fmtwarn): Update for introduction of
8380 format_string_diagnostic_t.
8381 (fmtwarn_n): Likewise.
8382 * substring-locations.c
8383 (format_string_diagnostic_t::format_string_diagnostic_t) New ctor.
8384 (format_warning_n_va): Convert to...
8385 (format_string_diagnostic_t::emit_warning_n_va): ...this.
8386 (format_warning_va): Convert to...
8387 (format_string_diagnostic_t::emit_warning_va): ...this.
8388 (format_warning_at_substring): Convert to...
8389 (format_string_diagnostic_t::emit_warning): ...this.
8390 (format_warning_at_substring_n): Convert to...
8391 (format_string_diagnostic_t::emit_warning_n): ...this.
8392 * substring-locations.h (class format_string_diagnostic_t): New
8394 (format_warning_va): Convert to
8395 format_string_diagnostic_t::emit_warning_va.
8396 (format_warning_n_va): Convert to
8397 format_string_diagnostic_t::emit_warning_n_va.
8398 (format_warning_at_substring): Convert to
8399 format_string_diagnostic_t::emit_warning.
8400 (format_warning_at_substring_n): Convert to
8401 format_string_diagnostic_t::emit_warning_n.
8403 2018-09-17 Cesar Philippidis <cesar@codesourcery.com>
8404 Bernd Schmidt <bernds_cb1@t-online.de>
8406 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with
8409 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
8411 * config/i386/i386.md (truncxf<mode>2_i387_noop_unspec): Change
8412 operand 0 predicate to nonimmediate operand.
8413 (rint<mode>2_frndint): Remove insn pattern.
8414 (rint<mode>2): Change operand 1 predicate to general_operand.
8415 Extend operand 1 to XFmode and generate rintxf2 insn.
8416 (frndintxf2_<rounding>): Rename from frndint<mode>2_<rounding>.
8417 Do not use X87MODEF mode macro.
8418 (frndintxf2_<rounding>_i387): Rename from
8419 frndint<mode>2_<rounding>_i387. Do not use X87MODEF mode macro.
8420 (<rounding_insn><mode>2): For non-SSE modes, extend operand 1
8421 to XFmode and generate significandxf3 insn.
8423 2018-09-17 Richard Biener <rguenther@suse.de>
8425 PR tree-optimization/87328
8426 * tree-ssa-sccvn.c (process_bb): Remove assertion about not
8427 visiting unexecutable backedges when not iterating.
8428 (do_rpo_vn): Mark all edges not executable even when not
8431 2018-09-17 Martin Jambor <mjambor@suse.cz>
8434 * doc/invoke.texi (Warning Options): Likewise.
8436 2018-09-17 Richard Biener <rguenther@suse.de>
8438 PR tree-optimization/87301
8439 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup): Properly
8440 clean EH info from leftover copy assignments.
8442 2018-09-17 Martin Liska <mliska@suse.cz>
8444 PR gcov-profile/85871
8445 * gcov.c (output_intermediate_file): Fix out of bounds
8448 2018-09-17 Vineet Gupta <vgupta@synopsys.com>
8450 * config/arc/arc.c: Object attributes for core4 not reflected
8452 * config/arc/arc.h: Don't restrict DBNZ to core3 (core4 includes
8455 2018-09-17 Alexey Brodkin <abrodkin@synopsys.com>
8457 * config/arc/linux.h (LINK_EH_SPEC): Add missing space.
8459 2018-09-17 Martin Liska <mliska@suse.cz>
8461 * doc/gcov.texi: Document new option --use-hotness-colors.
8462 * gcov.c (struct source_info): Declare new field.
8463 (source_info::source_info): Set default for maximum_count.
8464 (print_usage): Add new -q option.
8465 (process_args): Process it.
8466 (accumulate_line_info): Save src->maximum_count.
8467 (output_line_beginning): Make color line number if
8468 flag_use_hotness_colors is set.
8469 (output_line_details): Pass default argument value.
8470 (output_lines): Pass src->maximum_count.
8472 2018-09-17 Martin Liska <mliska@suse.cz>
8474 * common/config/i386/i386-common.c (ix86_get_valid_option_values):
8475 Use processor_names table.
8476 * config/i386/i386.c (ix86_default_align): Use
8477 processor_cost_table for alignment values.
8478 (ix86_option_override_internal): Use processor_names.
8479 (ix86_function_specific_print): Likewise.
8480 * config/i386/i386.h (struct processor_costs):
8481 Add alignment values.
8482 (struct ptt): Remove and replace with const char *.
8483 * config/i386/x86-tune-costs.h (struct processor_costs):
8484 Declare default alignments for all costs.
8486 2018-09-17 Aldy Hernandez <aldyh@redhat.com>
8488 * tree-vrp.c (extract_range_from_unary_expr): Do not special case
8489 symbolics or VR_VARYING ranges for ABS_EXPR.
8490 * wide-int-range.cc (wide_int_range_abs): Return positive numbers
8491 when range will wrap.
8493 2018-09-15 Eric Botcazou <ebotcazou@adacore.com>
8496 * cfgexpand.c (expand_gimple_basic_block): Be prepared for a BARRIER
8497 before and after a JUMP_TABLE_DATA.
8499 2018-09-14 John David Anglin <danglin@gcc.gnu.org>
8502 * dojump.c (do_compare_and_jump): Canonicalize function pointers
8503 when one operand is a function pointer. Use POINTER_TYPE_P and
8504 FUNC_OR_METHOD_TYPE_P.
8505 * expr.c (do_store_flag): Use POINTER_TYPE_P and FUNC_OR_METHOD_TYPE_P.
8506 * fold-const.c (build_range_check): Likewise.
8507 * match.pd (simple_comparison): Likewise.
8509 2018-09-14 David Malcolm <dmalcolm@redhat.com>
8512 * spellcheck.c (get_edit_distance_cutoff): New function.
8513 (selftest::test_edit_distance_unit_test_oneway): Rename to...
8514 (selftest::test_get_edit_distance_one_way): ...this.
8515 (selftest::test_get_edit_distance_unit): Rename to...
8516 (selftest::test_get_edit_distance_both_ways): ...this.
8517 (selftest::test_edit_distances): Move tests to this new function,
8518 and test some more pairs of strings. Update for above renaming.
8519 (selftest::get_old_cutoff): New function.
8520 (selftest::test_get_edit_distance_cutoff): New function.
8521 (selftest::assert_suggested_for): New function.
8522 (ASSERT_SUGGESTED_FOR): New macro.
8523 (selftest::assert_not_suggested_for): New function.
8524 (ASSERT_NOT_SUGGESTED_FOR): New macro.
8525 (selftest::test_suggestions): New function.
8526 (selftest::spellcheck_c_tests): Move test_get_edit_distance_unit
8527 tests to selftest::test_edit_distances and call it. Add calls to
8528 selftest::test_get_edit_distance_cutoff and
8529 selftest::test_suggestions.
8530 * spellcheck.h (get_edit_distance_cutoff): New function declaration.
8531 (best_match::consider): Replace hard-coded cutoff calculation with
8533 (best_match::get_cutoff): New declaration.
8534 (best_match::get_best_meaningful_candidate): Likewise.
8536 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
8538 * builtins.c (fold_builtin_strlen): Remove TODO comment.
8540 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
8543 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
8545 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
8546 terminated string literal.
8548 2018-09-14 Martin Sebor <msebor@redhat.com>
8550 * builtins.c (unterminated_array): Handle ARRAY_REF.
8551 (expand_builtin_stpcpy_1): Detect unterminated char arrays.
8552 * builtins.h (unterminated_array): Declare extern.
8553 * gimple-fold.c (gimple_fold_builtin_stpcpy): Detect unterminated
8555 (gimple_fold_builtin_sprintf): Propagate NO_WARNING to transformed
8558 2018-09-14 Martin Sebor <msebor@redhat.com>
8559 Jeff Law <law@redhat.com>
8561 * builtins.c (unterminated_array): New.
8562 (expand_builtin_strcpy): Adjust.
8563 (expand_builtin_strcpy_args): Detect unterminated arrays.
8564 * gimple-fold.c (get_maxval_strlen): Add argument. Detect
8565 unterminated arrays.
8566 * gimple-fold.h (get_maxval_strlen): Add argument.
8567 (gimple_fold_builtin_strcpy): Detec unterminated arrays.
8569 * gimple-fold.c (get_range_strlen): Add argument.
8570 (get_maxval_strlen): Adjust.
8571 * gimple-fold.h (get_range_strlen): Add argument.
8573 2018-09-14 Wei Xiao <wei3.xiao@intel.com>
8575 * config/i386/movdirintrin.h: Fix copyright year.
8577 2018-09-14 Uros Bizjak <ubizjak@gmail.com>
8579 * reg-stack.c: Include regs.h.
8580 (replace_reg): Assert that mode is MODE_FLOAT or MODE_COMPLEX_FLOAT.
8581 (emit_pop_insn): Default pop insn mode to the reg_raw_mode of
8582 FIRST_STACK_REG, not DFmode.
8583 (emit_swap_insn): Default swap insn mode to the reg_raw_mode of
8584 FIRST_STACK_REG, not XFmode. Explicitly construct swap RTX.
8585 (change stack): Default register mode to the reg_raw_mode of
8586 FIRST_STACK_REG, not DFmode.
8587 * config/i386/i386.md (*swap<mode>): Remove insn pattern.
8588 (*swapxf): Rename from swapxf.
8590 2018-09-14 Carl Love <cel@us.ibm.com>
8592 * config/rs6000/emmintrin.h: Add _MM_SHUFFLE2.
8593 * config/rs6000/xmmintrin.h: Add _MM_SHUFFLE.
8595 2018-09-14 Segher Boessenkool <segher@kernel.crashing.org>
8598 * config/rs6000/rs6000.md (*mov<mode>_hardfloat64): Add Z to the Y
8601 2018-09-14 Sam Tebbs <sam.tebbs@arm.com>
8604 * config/aarch64/aarch64.md (*aarch64_bfxilsi_uxtw): Define.
8606 2018-09-14 Jason Merrill <jason@redhat.com>
8608 Fix --enable-gather-detailed-mem-stats.
8609 * hash-table.c (hash_table_usage): Change from variable to function.
8610 * hash-table.h: Adjust.
8611 * Makefile.in: Add missing dependencies on hash-table.h.
8613 2018-09-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8615 PR tree-optimization/87259
8617 (pass_cse_reciprocals::execute): Run optimize_recip_sqrt after
8618 execute_cse_reciprocals_1 has tried transforming.
8620 2018-09-14 Aldy Hernandez <aldyh@redhat.com>
8622 * tree-vrp.c (extract_range_from_binary_expr_1): Normalize
8623 VR_VARYING for PLUS/MINUS_EXPR.
8625 2018-09-14 Ilya Leoshkevich <iii@linux.ibm.com>
8627 * config/s390/s390-passes.def (INSERT_PASS_BEFORE): Improve
8630 2018-09-14 Richard Biener <rguenther@suse.de>
8633 * tree-ssa-coalesce.c (ssa_conflicts_merge): Remove conflict
8634 bits for the merged partition.
8636 2018-09-13 Martin Sebor <msebor@redhat.com>
8637 Bernd Edlinger <bernd.edlinger@hotmail.de>
8639 * builtins.h (c_srlen): Add argument.
8640 * builtins.c (warn_string_no_nul): New function.
8641 (c_strlen): Add argument and use it. Update recursive calls.
8642 Pass DECL argument to string_constant to get info on non
8643 terminated strings. Update *NONSTR as needed.
8644 (fold_builtin_strlen): Add argument to calls to c_strlen.
8645 Warn for unterminated arrays.
8646 (warn_string_no_null): Add prototype.
8647 * expr.c (string_constant): Update arguments. Update recursive
8648 calls appropriately. Detect missing NUL terminator and outermost
8649 declaration its missing in.
8650 Improve checks for arrays with nonzero lower bound or elements
8651 that are not a single byte. Simplify offset computation.
8652 Simplify checks for non-NUL terminated strings.
8653 * gimple-fold.c (get_range_strlen): Add argument to c_strlen call.
8654 * gimple-ssa-sprintf.c (get_string_length): Remove unnecessary code.
8656 2018-09-13 Bernd Edlinger <bernd.edlinger@hotmail.de>
8658 * builtins.c (c_strlen): Handle not zero terminated STRING_CSTs
8660 * fold-const.c (c_getstr): Fix function comment. Remove unused third
8661 argument. Fix range checks.
8662 * fold-const.h (c_getstr): Adjust protoype.
8663 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid folding when
8664 string is constant but contains no NUL byte.
8666 * expr.c (string_constant): Adjust function comment.
8667 Remove bogus check for zero termination.
8669 * fold-const.c (c_getstr): Clamp STRING_LENGTH to STRING_SIZE.
8671 * varasm.c (compare_constant): Compare type size of STRING_CSTs.
8672 (get_constant_size): Don't make STRING_CSTs larger than they are.
8673 (check_string_literal): New check function for STRING_CSTs.
8674 (output_constant): Use it.
8676 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
8679 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
8681 2018-09-13 Richard Biener <rguenther@suse.de>
8683 PR tree-optimization/87263
8684 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
8685 (struct unwind_state): Add max_rpo field.
8686 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
8687 Compute max_rpo, the max RPO number a block can be backwards reached
8688 from. Re-write non-iterating mode to a RPO ordered worklist approach,
8689 separating it from the iterating mode.
8691 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
8693 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
8694 (rfs_decision): New scheduling decision.
8696 2018-09-13 Richard Biener <rguenther@suse.de>
8699 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
8700 (vn_nary_op_insert_pieces_predicated): Do not write useless
8701 valid_dominated_by_p entry outside of the allocated storage.
8703 2018-09-13 Omar Sandoval <osandov@osandov.com>
8704 Tom de Vries <tdevries@suse.de>
8707 * dwarf2out.c (is_c): New function.
8708 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
8710 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
8713 * config/aarch64/aarch64.md (*aarch64_bfxil):
8715 * config/aarch64/constraints.md (Ulc): Define.
8716 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
8718 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
8721 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
8723 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
8724 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
8725 aarch64_layout_frame call.
8726 (aarch64_expand_epilogue): Likewise.
8727 (aarch64_initial_elimination_offset): Likewise.
8728 (aarch64_get_separate_components): Likewise.
8729 (aarch64_use_return_insn_p): Likewise.
8730 (aarch64_layout_frame): Remove unneeded check.
8732 2018-09-13 Jakub Jelinek <jakub@redhat.com>
8734 * configure.ac: Only append
8735 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
8736 gcc_config_arguments if it was never reconfigured or last reconfigure
8737 was with different arguments.
8738 * configure: Regenerated.
8740 2018-09-13 Jakub Jelinek <jakub@redhat.com>
8741 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8744 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
8745 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
8747 2018-09-13 Jakub Jelinek <jakub@redhat.com>
8749 PR tree-optimization/87287
8750 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
8751 X % C == 0 to X % (unsigned) C == 0 optimization to ...
8752 * match.pd (X % C == 0): ... here. New optimization.
8754 2018-09-12 Jakub Jelinek <jakub@redhat.com>
8757 * expr.h (maybe_optimize_mod_cmp): Declare.
8758 * expr.c (mod_inv): New function.
8759 (maybe_optimize_mod_cmp): New function.
8760 (do_store_flag): Use it.
8761 * cfgexpand.c (expand_gimple_cond): Likewise.
8763 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
8764 Julian Brown <julian@codesourcery.com>
8767 * gimplify.c (gimplify_scan_omp_clauses): Set
8768 target_firstprivatize_array_bases in OpenACC parallel and kernels
8769 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
8770 OpenACC data regions.
8772 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
8774 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
8775 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
8777 2018-09-12 Richard Biener <rguenther@suse.de>
8779 PR tree-optimization/87280
8780 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
8781 edge but unreachable target.
8782 (do_rpo_vn): For conservatively handling a PHI only mark
8783 the backedge executable but not the block reachable.
8785 2018-09-12 Richard Biener <rguenther@suse.de>
8787 PR tree-optimization/87266
8788 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
8791 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
8793 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
8795 ("trunc<BFP:mode><DFP_ALL:mode>2")
8796 ("trunc<DFP_ALL:mode><BFP:mode>2")
8797 ("extend<BFP:mode><DFP_ALL:mode>2")
8798 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
8799 according to the target operand type.
8801 2018-09-12 Jakub Jelinek <jakub@redhat.com>
8802 Andreas Krebbel <krebbel@linux.ibm.com>
8804 PR tree-optimization/86844
8805 * gimple-ssa-store-merging.c
8806 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
8807 there are any overlapping stores in between them, make sure they are
8808 also coalesced or we give up completely.
8810 2018-09-12 Jakub Jelinek <jakub@redhat.com>
8813 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
8814 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
8816 2018-09-12 Tom de Vries <tdevries@suse.de>
8818 * common.opt (gdescribe-dies): Add option.
8819 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
8820 attribute for artifical and nameless decls.
8821 (dwarf2out_register_external_die): Add description attribute to
8822 external reference die.
8823 (add_desc_attribute): New functions.
8824 (gen_subprogram_die): Add description attribute to
8825 DW_TAG_call_site_parameter.
8826 * tree-pretty-print.c (print_generic_expr_to_str): New function.
8827 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
8828 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
8830 (@item -gdescribe-dies): Add.
8832 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
8834 * tree-vrp.c (vrp_shift_undefined_p): Remove.
8835 (extract_range_from_binary_expr_1: Call
8836 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
8837 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
8840 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
8842 * gimple-ssa-warn-alloca.c
8843 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
8844 field for ALLOCA_BOUND_*_LARGE.
8846 2018-09-11 Nathan Sidwell <nathan@acm.org>
8848 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
8850 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
8852 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
8853 for clobbers. Remove obsolete comment.
8855 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
8857 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
8858 mpxchk, mpxld and mpxst types.
8859 (define_attr length_immediate): Remove all processing of mpx types.
8860 (define_attr prefix_0f): Ditto.
8861 (define_attr memory): Ditto.
8863 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
8865 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
8866 (log<mode>2): Change operand 1 predicate to general_operand.
8867 Extend operand 1 to XFmode and generate logxf3 insn.
8868 (log10<mode>2): Change operand 1 predicate to general_operand.
8869 Extend operand 1 to XFmode and generate log10xf3 insn.
8870 (log2<mode>2): Change operand 1 predicate to general_operand.
8871 Extend operand 1 to XFmode and generate log2xf3 insn.
8872 (fyl2xp1_extend<mode>xf3_i387): Remove.
8873 (log1p<mode>2): Change operand 1 predicate to general_operand.
8874 Extend operand 1 to XFmode and generate log1pxf3 insn.
8875 (fxtract_extend<mode>xf3_i387): Remove.
8876 (logb<mode>2): Change operand 1 predicate to general_operand.
8877 Extend operand 1 to XFmode and generate logbxf3 insn.
8878 (ilogb<mode>2): Change operand 1 predicate to general_operand.
8879 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
8880 (significand<mode>2): Change operand 1 predicate to general_operand.
8881 Extend operand 1 to XFmode and generate significandxf3 insn.
8883 2018-09-11 Nathan Sidwell <nathan@acm.org>
8885 * gcc.c (perror_with_name, pfatal_with_name): Delete.
8886 (load_specs): Use fatal_error.
8887 (DELETE_IF_ORDINARY, process_command): Use error.
8888 (execute, run_attempt): Use fatal_error.
8890 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
8892 * diagnostic-core.h (sorry_at): New prototype.
8893 * diagnostic.c (sorry_at): New function.
8895 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
8897 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
8898 by zero as VR_UNDEFINED.
8900 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
8902 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
8903 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
8904 (<sincos>mode2): New expander.
8905 (sincos_extend<mode>xf3_i387): Remove insn pattern.
8906 (sincos -> sin, cos splitters): Remove splitter patterns.
8907 (sincos<mode>3): Change operand 2 predicate to general_operand.
8908 Extend operand 2 to XFmode and generate sincosxf3 insn.
8909 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
8910 Change operand 3 predicate to const1_operand.
8911 (fptan_extend<mode>xf4_i387): Remove insn pattern.
8912 (tanxf2): Update operands in the call to fptanxf4_i387.
8913 (tan<mode>2): Change operand 1 predicate to general_operand.
8914 Extend operand 1 to XFmode and generate tanxf3 insn.
8915 (atan2xf3): Rename from *fpatanxf3_i387.
8916 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
8917 (atan2xf3): Remove expander.
8918 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
8919 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
8920 (atan<mode>2): Change operand 1 predicate to general_operand.
8921 Extend operand 1 to XFmode and generate atanxf3 insn.
8923 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
8925 * config/i386/i386.md (x87/SSE constant load splitter): Use
8926 memory_operand instead of nonimmediate_operand for input operand
8929 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
8931 * config/i386/i386.md (float partial SSE register stall splitter): Move
8932 splitter near its instruction pattern.
8933 (float_extend partial SSE register stall splitter): Ditto.
8934 (float_truncate partial SSE register stall splitter): Ditto.
8936 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
8939 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
8940 to speculation_safe_value_not_needed.
8943 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
8944 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
8945 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
8946 leaf_function_p, instead use has_hard_reg_initial_val.
8948 2018-09-09 Nathan Sidwell <nathan@acm.org>
8950 * gcc.h (pfatal_with_name): Don't declare here.
8951 * gcc.c (pfatal_with_name): Make static.
8953 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
8955 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
8958 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
8960 PR rtl-optimization/85458
8961 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
8962 priority hook to reduce the priority of EXPR.
8964 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
8966 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
8967 DImode for x87 on 32bit targets. Conditionally disable x87 modes
8968 with X87_ENABLE_FLOAT. Remove preparation code.
8969 (*float<SWI48:mode><MODEF:mode>2): Rename from
8970 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
8971 math using "enabled" attribute.
8972 (*floatdi<MODEF:mode>2_i387): Rename from
8973 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
8974 enable for 32bit targets only.
8975 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
8977 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
8978 as operand 1 predicate. Rewrite as define_insn_and_split.
8979 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
8981 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
8983 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
8984 to fallthru to FLOAT case.
8986 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
8989 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
8990 around folding of vec_sl to handle out of range shift values.
8992 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
8994 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
8995 Update callers to gen_fix_trunc<mode>_i387_fisttp
8996 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
8997 nonimmediate_operand.
8998 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
8999 and corresponding splitters.
9000 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
9001 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
9002 (fix_truncdi_i387_with_temp): Remove insn pattern
9003 and corresponding splitters.
9004 (fix_trunc<mode>_i387): Change operand 0 predicate to
9005 nonimmediate_operand.
9006 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
9007 and corresponding splitters.
9008 (*fistdi2_1): Remove.
9010 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
9011 (lrintxfdi2): Remove expander. Reimplement as define_insn.
9012 (*fist<mode>2_1): Remove.
9013 (fist<mode>2): Ditto.
9014 (fist<mode>2_with_temp): Remove insn pattern and corresponding
9016 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
9017 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
9018 (fistdi2_<rounding>): Change operand 0 predicate to
9019 nonimmediate_operand.
9020 (fistdi2_<rounding>_with_temp): Remove insn pattern
9021 and corresponding splitters.
9022 (fist<mode>2_<rounding>): Change operand 0 predicate to
9023 nonimmediate_operand.
9024 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
9025 and corresponding splitters.
9027 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
9029 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
9031 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
9034 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
9036 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
9037 early gimple folding of vec_splat().
9038 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
9039 * gimple-fold.h: Add an extern define for tree_vec_extract().
9041 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
9043 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
9044 wrappers around TREE_TYPE comparisons.
9046 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
9049 * config/s390/predicates.md: Add nonsym_memory_operand.
9050 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
9051 contains a SYMBOL_REF, load it into an intermediate pseudo.
9052 (s390_emit_compare_and_swap): Legitimize operand.
9053 * config/s390/s390.md: Use the new nonsym_memory_operand
9054 with UNSPECV_CAS patterns.
9056 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
9059 * config/s390/s390-passes.def: New file.
9060 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
9062 (make_pass_s390_early_mach): Add declaration.
9063 * config/s390/s390.c (make_pass_s390_early_mach):
9064 (s390_option_override): Remove dynamic registration.
9065 * config/s390/t-s390: Add s390-passes.def.
9067 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
9069 * config/s390/s390.c (s390_decompose_constant_pool_ref):
9070 Remove UNSPEC_LTREL_BASE check.
9071 (annotate_constant_pool_refs): Likewise.
9072 (find_constant_pool_ref): Likewise.
9073 (find_ltrel_base): Removed.
9074 (replace_ltrel_base): Removed.
9075 (s390_mainpool_finish): Remove replace_ltrel_base call.
9076 (s390_chunkify_start): Remove pending LTREL_BASE logic.
9077 (s390_chunkify_finish): Remove replace_ltrel_base call.
9078 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
9080 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
9083 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
9084 to speculation_safe_value_not_needed.
9086 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
9087 Bernd Schmidt <bernds_cb1@t-online.de>
9089 * config/nvptx/nvptx-opts.h: New file.
9090 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
9091 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
9093 (TARGET_SM35): New macro.
9094 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
9096 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
9098 (misa=): New option.
9099 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
9101 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
9103 * config/i386/i386.md (truncdfsf2): Remove expander.
9104 (truncdfsf2_with_temp): Ditto.
9105 (truncxf<mode>2): Ditto.
9106 (*truncdfsf_fast_mixed): Remove insn pattern.
9107 (*truncdfsf_fast_i387): Ditto.
9108 (*truncdfsf_mixed): Ditto.
9109 (*truncdfsf_i387): Ditto.
9110 (*truncdfsf2_i387_1): Ditto.
9111 (*truncxfsf2_mixed): Ditto.
9112 (*truncxfdf2_mixed): Ditto.
9113 (*truncxf<mode>2_i387_noop): Ditto. Update callers
9114 to call gen_truncxf<mode>2 instead.
9115 (*truncxf<mode>2_i387): Remove.
9116 (reg->reg splitters): Remove splitter pattern.
9117 (reg->mem splitters): Ditto.
9119 (truncdfsf2): New insn pattern.
9120 (truncxf<mode>2): Ditto.
9122 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9124 * tree-ssa-math-opts.c (is_mult_by): New function.
9125 (is_square_of): Use the above.
9126 (optimize_recip_sqrt): New function.
9127 (pass_cse_reciprocals::execute): Use the above.
9129 2018-09-05 Richard Biener <rguenther@suse.de>
9132 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
9133 to zero-init the emplaced vec.
9135 2018-09-05 Martin Liska <mliska@suse.cz>
9137 PR tree-optimization/87205
9138 * tree-switch-conversion.c (pass_lower_switch::execute):
9139 Group cases for switch statements.
9141 2018-09-05 Richard Biener <rguenther@suse.de>
9143 PR tree-optimization/87217
9144 * tree-ssa-sccvn.c (vuse_valueize): New.
9145 (vn_reference_lookup_pieces): Use it.
9146 (vn_reference_lookup): Likewise.
9148 2018-09-05 Nathan Sidwell <nathan@acm.org>
9151 * stor-layout.c (place_field): Scan forwards to check last
9152 bitfield when ms_bitfield_placement is in effect.
9154 2018-09-05 Richard Biener <rguenther@suse.de>
9157 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
9160 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
9161 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
9163 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
9164 * config.gcc (extra_objs): Build it.
9165 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
9167 * config/aarch64/aarch64-passes.def
9168 (pass_tag_collision_avoidance): New pass.
9169 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
9170 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
9171 (aarch64_classify_address): Remove static qualifier.
9172 (aarch64_address_info, aarch64_address_type): Move to...
9173 * config/aarch64/aarch64-protos.h: ... here.
9174 (make_pass_tag_collision_avoidance): New function.
9175 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
9178 2018-09-05 Martin Liska <mliska@suse.cz>
9180 * doc/gcov.texi: Update documentation of humar
9182 * gcov.c (format_count): Print one decimal place, it provides
9183 more fine number of situations like '1G' vs. '1.4G'.
9185 2018-09-05 Martin Liska <mliska@suse.cz>
9188 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
9189 * optc-gen.awk: Allow 'Var' for Deprecated options in order
9190 to generate a MASK value.
9192 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
9195 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
9196 if frame pointer isn't used.
9197 (compute_frame_pointer_to_fb_displacement): Likewise.
9199 2018-09-04 Jakub Jelinek <jakub@redhat.com>
9202 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
9203 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
9204 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
9205 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
9206 and OPTION_MASK_ISA_XSAVEC_UNSET.
9208 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
9210 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
9211 XOR operations in NAND case.
9213 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
9215 * wide-int-range.cc (wide_int_range_convert): New.
9216 * wide-int-range.h (wide_int_range_convert): New.
9217 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
9218 code into wide_int_range_convert.
9219 (extract_range_into_wide_ints): Do not munge anti range constants
9220 into the entire domain. Just return the range back.
9222 2018-09-04 Martin Liska <mliska@suse.cz>
9224 * genmatch.c (output_line_directive): Add new argument
9226 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
9228 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
9230 * doc/invoke.texi (Option Summary): Add whitespace.
9232 * doc/invoke.texi (Option Summary): Add -Waligned-new.
9234 2018-09-04 Richard Biener <rguenther@suse.de>
9236 PR tree-optimization/87211
9237 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
9238 backedge value we're supposed to treat as VARYING also number
9239 the PHI to VARYING in case it got a different value-number already.
9241 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
9243 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
9244 (extract_range_from_binary_expr_1): Do not call
9245 vrp_can_optimize_bit_op.
9246 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
9248 (wide_int_range_get_mask_and_bounds): New.
9249 (wide_int_range_optimize_bit_op): New.
9250 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
9251 (wide_int_range_bit_and): Same.
9252 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
9253 (wide_int_range_optimize_bit_op): New.
9254 (wide_int_range_get_mask_and_bounds): New.
9256 2018-09-04 Richard Biener <rguenther@suse.de>
9258 PR tree-optimization/87176
9259 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
9260 variable. When value-numbering a virtual PHI node make sure
9261 to not value-number to the backedge value.
9263 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
9265 * doc/extend.texi (Long Long, Hex Floats): Document support for
9266 long long and hex floats in more recent versions of ISO C++.
9268 2018-09-03 Richard Biener <rguenther@suse.de>
9270 PR tree-optimization/87177
9271 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
9274 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
9276 * bb-reorder.c (edge_order): Convert to C-qsort-style
9277 tri-state comparator.
9278 (reorder_basic_blocks_simple): Change std::stable_sort to
9281 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
9283 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
9284 tri-state comparator.
9285 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
9287 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
9289 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
9290 (mergesort): ... here as maximum count for using netsort.
9291 (gcc_qsort): Set nlim to 3 if stable sort is requested.
9292 (gcc_stablesort): New.
9293 * system.h (gcc_stablesort): Declare.
9295 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
9297 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
9298 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
9299 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
9301 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
9303 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
9304 lxsdx and stxsdx alternatives.
9305 (*mov<mode>_hardfloat64): Ditto.
9306 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
9308 2018-09-03 Richard Biener <rguenther@suse.de>
9310 PR tree-optimization/87200
9311 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
9314 2018-09-03 Martin Liska <mliska@suse.cz>
9316 PR tree-optimization/87201
9317 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
9318 Fix parenthesis in an expression.
9320 2018-09-03 Richard Biener <rguenther@suse.de>
9322 PR tree-optimization/87197
9323 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
9324 visited. CSE the VN_INFO hashtable lookup.
9326 PR tree-optimization/87169
9327 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
9328 iterating make sure there's no extra backedges from irreducible
9329 regions feeding the header. Mark the destination block
9332 2018-09-03 Martin Liska <mliska@suse.cz>
9335 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
9336 * common/common-targhooks.c (default_get_valid_option_values):
9338 * common/common-targhooks.h (default_get_valid_option_values):
9340 * common/config/i386/i386-common.c: Move processor_target_table
9342 (ix86_get_valid_option_values): New function.
9343 (TARGET_GET_VALID_OPTION_VALUES): New macro.
9344 * config/i386/i386.c (struct ptt): Move to i386-common.c.
9345 (PTA_*): Move all defined masks into i386-common.c.
9346 (ix86_function_specific_restore): Use new processor_cost_table.
9347 * config/i386/i386.h (struct ptt): Moved from i386.c.
9348 (struct pta): Likewise.
9349 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
9350 * doc/tm.texi.in: Likewise.
9351 * opt-suggestions.c (option_proposer::suggest_option):
9352 Pass prefix to build_option_suggestions.
9353 (option_proposer::get_completions): Likewise.
9354 (option_proposer::build_option_suggestions): Use the new target
9356 * opts.c (struct option_help_tuple): New struct.
9357 (print_filtered_help): Use the new target hook.
9359 2018-09-03 Martin Liska <mliska@suse.cz>
9362 * predict.c (set_even_probabilities): Add likely_edges
9363 argument and handle cases where we have precisely one
9365 (combine_predictions_for_bb): Catch also likely_edges.
9366 (tree_predict_by_opcode): Handle gswitch statements.
9367 * tree-cfg.h (find_case_label_for_value): New declaration.
9368 (find_taken_edge_switch_expr): Likewise.
9369 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
9370 Find pivot in decision tree based on probabily, not by number of
9373 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
9375 * doc/standards.texi (Standards): Update Objective-C reference.
9377 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
9379 * doc/install.texi (Prerequisites): Update link for MPC.
9381 2018-09-01 Michael Matz <matz@suse.de>
9383 PR tree-optimization/87074
9384 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
9385 PHIs for outer-loop uses.
9387 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
9389 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
9390 * doc/invoke.texi (C Dialect Options): Ditto.
9392 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
9394 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
9396 2018-08-31 Richard Biener <rguenther@suse.de>
9398 PR tree-optimization/87168
9399 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
9400 (rpo_elim::eliminate_avail): When OP was not visited it must
9403 2018-08-31 David Malcolm <dmalcolm@redhat.com>
9405 * tree-vrp.c (copy_value_range): Convert param "from" from
9406 "value_range *" to "const value_range *".
9407 (range_is_null): Likewise for param "vr".
9408 (range_int_cst_p): Likewise.
9409 (range_int_cst_singleton_p): Likewise.
9410 (symbolic_range_p): Likewise.
9411 (value_ranges_intersect_p): Likewise for both params.
9412 (value_range_nonnegative_p): Likewise for param "vr".
9413 (value_range_constant_singleton): Likewise.
9414 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
9415 (extract_range_into_wide_ints): Likewise for param "vr".
9416 (extract_range_from_multiplicative_op): Likewise for params "vr0"
9418 (vrp_can_optimize_bit_op): Likewise.
9419 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
9421 (extract_range_from_unary_expr): Likewise.
9422 (debug_value_range): Likewise for param "vr".
9423 (value_range::dump): Add "const" qualifier.
9424 (vrp_prop::check_array_ref): Convert local "vr" from
9425 "value_range *" to "const value_range *".
9426 (vrp_prop::check_mem_ref): Likewise.
9427 (vrp_prop::visit_stmt): Likewise for local "old_vr".
9428 (vrp_intersect_ranges_1): Likewise for param "vr_1".
9429 (vrp_intersect_ranges): Likewise.
9430 (simplify_stmt_for_jump_threading): Likewise for local "vr".
9431 (vrp_prop::vrp_finalize): Likewise.
9432 * tree-vrp.h (value_range::dump): Add "const" qualifier.
9433 (vrp_intersect_ranges): Add "const" qualifier to params as above.
9434 (extract_range_from_unary_expr): Likewise.
9435 (value_range_constant_singleton): Likewise.
9436 (symbolic_range_p): Likewise.
9437 (copy_value_range): Likewise.
9438 (extract_range_from_binary_expr_1): Likewise.
9439 (range_int_cst_p): Likewise.
9440 (vrp_set_zero_nonzero_bits): Likewise.
9441 (range_int_cst_singleton_p): Likewise.
9443 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
9445 * config/aarch64/arm_neon.h (vabsd_s64): New.
9446 (vnegd_s64): Likewise.
9448 2018-08-31 Martin Jambor <mjambor@suse.cz>
9450 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
9452 2018-08-31 Martin Liska <mliska@suse.cz>
9454 * ipa-icf.c (sem_item::add_type): Use
9455 sem_item::m_type_hash_cache.
9456 * ipa-icf.h: Move the cache from sem_item_optimizer
9459 2018-08-31 Nathan Sidwell <nathan@acm.org>
9461 * doc/extend.texi (Backwards Compatibility): Remove implicit
9462 extern C leeway of () being (...).
9464 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9466 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
9468 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
9472 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
9474 2018-08-31 Jakub Jelinek <jakub@redhat.com>
9477 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
9478 gen_int_mode. Formatting fixes.
9480 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
9482 * target.def (custom_function_descriptors): Improve documentation.
9483 * doc/tm.texi.in (Trampolines): Expand discussion of function
9484 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
9485 beginning of the section.
9486 * doc/tm.texi: Regenerated.
9488 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
9490 * cfg.h (class auto_edge_flag): Spell out the template-id of the
9491 base class in the initializer list. This is a workaround for
9492 building with older GCC.
9493 (class auto_bb_flag): Likewise.
9495 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
9497 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
9498 (altivec_vcmpequ<VI_char>_p): Remove star.
9499 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
9501 (expand_strncmp_vec_sequence): New function.
9502 (emit_final_str_compare_vec): New function.
9503 (expand_strn_compare): Add support for vector strncmp.
9504 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
9505 length specification to bytes.
9506 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
9507 (vcmpnezb_p): New pattern.
9508 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
9509 for option -mstring-compare-inline-limit.
9511 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
9513 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
9514 (PTA_SKYLAKE): Add PTA_AES.
9515 (PTA_GOLDMONT): Likewise.
9517 2018-08-29 Jan Hubicka <jh@suse.cz>
9520 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
9521 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
9523 2018-08-29 Jan Hubicka <jh@suse.cz>
9525 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
9527 (hash_tree): Do not visit TYPE_STUB_DECL.
9528 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
9529 stream TYPE_STUB_DECL.
9530 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
9531 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
9532 after free_lang_data.
9533 (type_in_anonymous_namespace_p): Likewise.
9535 2018-08-29 Jan Hubicka <jh@suse.cz>
9537 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
9538 comment that it has to be even number.
9539 (class sreal): Change m_sig type to int32_t.
9540 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
9541 int64_t for temporary calculations.
9542 (sreal_verify_basics): Drop one bit from minimum and maximum.
9544 2018-08-30 Richard Biener <rguenther@suse.de>
9546 PR tree-optimization/87147
9547 * tree-ssa-sccvn.c (SSA_VISITED): New function.
9548 (visit_phi): When the degenerate result is from the backedge and
9549 we didn't visit its definition yet drop to VARYING.
9550 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
9552 2018-08-29 Jan Hubicka <jh@suse.cz>
9554 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
9556 (hash_tree): Likewise.
9558 2018-08-29 Jan Hubicka <jh@suse.cz>
9560 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
9561 and TYPE_NEXT_REF_TO.
9563 2018-08-29 Jan Hubicka <jh@suse.cz>
9565 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
9566 comment that it has to be even number.
9567 (class sreal): Change m_sig type to int32_t.
9568 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
9569 int64_t for temporary calculations.
9570 (sreal_verify_basics): Drop one bit from minimum and maximum.
9572 2018-08-30 Tamar Christina <tamar.christina@arm.com>
9574 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
9576 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
9579 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
9580 if to_add is negative.
9582 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
9585 * builtins.c (c_strlen): Improve range checks.
9587 2018-08-29 Martin Sebor <msebor@redhat.com>
9588 Jeff Law <law@redhat.com>
9590 PR tree-optimization/86714
9591 PR tree-optimization/86711
9592 * builtins.c (c_strlen): Add arguments to call to string_constant.
9593 * expr.c (string_constant): Add argument. Detect missing nul
9594 terminator and outermost declaration it's missing in.
9595 * expr.h (string_constant): Add argument.
9596 * fold-const.c (read_from_constant_string): Add arguments to call to
9598 (c_getstr): Likewise.
9599 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
9601 * tree-ssa-strlen.c (get_stridx): Likewise.
9603 2018-08-29 Jan Hubicka <jh@suse.cz>
9605 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
9606 Do not stream DECL_VINDEX.
9607 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
9608 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
9609 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
9611 2018-08-29 Richard Biener <rguenther@suse.de>
9613 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
9614 virtual operands that are not default defs to honor region
9616 (rpo_vn_valueize): Remove ineffective code here.
9618 2018-08-29 Richard Biener <rguenther@suse.de>
9620 PR tree-optimization/87132
9621 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
9622 when skipping defs reachable over backedges.
9624 2018-08-29 Richard Biener <rguenther@suse.de>
9626 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
9627 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
9628 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
9629 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
9630 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
9631 (vn_reference_lookup_3): Remove use of const_parms.
9632 (free_rpo_vn): Do not free const_parms.
9633 (do_rpo_vn): Do not call init_const_parms.
9634 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
9635 SSA_NAME_POINTS_TO_READONLY_MEMORY.
9636 (call_may_clobber_ref_p_1): Likewise.
9638 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
9641 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
9643 (-ftree-scev-cprop): Document.
9645 2018-08-29 Jan Hubicka <jh@suse.cz>
9647 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
9649 (sreal constructor): Update.
9650 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
9651 sreal:operator/): Update.
9653 2018-08-29 Martin Liska <mliska@suse.cz>
9655 * tree-switch-conversion.c (switch_conversion::expand):
9656 Strenghten assumption about gswitch statements.
9658 2018-08-29 Richard Biener <rguenther@suse.de>
9660 PR tree-optimization/87117
9661 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
9662 re-value-number released SSA VDEFs.
9664 2018-08-29 Richard Biener <rguenther@suse.de>
9666 PR tree-optimization/87126
9667 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
9669 2018-08-28 Jim Wilson <jimw@sifive.com>
9671 * config/riscv/pic.md: Rewrite.
9672 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
9674 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
9675 (SOFTF, default_load, softload, softstore): New.
9677 2018-08-28 Jeff Law <law@redhat.com>
9679 * fold-const.c (fold_binary_loc): Remove recently added assert.
9681 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
9683 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
9684 to OP parmeter of generated function.
9686 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
9688 PR tree-optimization/87009
9689 * match.pd: Add boolean optimizations.
9691 2018-08-28 Martin Sebor <msebor@redhat.com>
9694 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
9695 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
9696 (pass_walloca::gate): Use it.
9697 (alloca_call_type): Same.
9698 (pass_walloca::execute): Same.
9699 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
9701 2018-08-28 David Malcolm <dmalcolm@redhat.com>
9703 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
9704 GCC_VERSION for usage of "__gcc_dump_printf__" format from
9707 2018-08-28 Richard Biener <rguenther@suse.de>
9709 PR tree-optimization/87124
9710 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
9711 constants before looking up avail.
9713 2018-08-28 Jakub Jelinek <jakub@redhat.com>
9716 * calls.c (maybe_warn_nonstring_arg): Punt early if
9717 warn_stringop_overflow is zero. Don't call get_range_strlen
9718 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
9719 Swap comparison operands to have constants on rhs. Only use
9720 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
9721 increment lenrng[0].
9723 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
9725 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
9726 use of tree_to_shwi. Remove duplicated test for the size being
9727 a whole number of bytes.
9729 2018-08-28 Richard Biener <rguenther@suse.de>
9731 PR tree-optimization/87117
9732 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
9733 Handle removed stmt without LHS (GIMPLE_NOP).
9735 2018-08-28 Richard Biener <rguenther@suse.de>
9737 PR tree-optimization/87117
9738 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
9739 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
9741 2018-08-28 Richard Biener <rguenther@suse.de>
9743 PR tree-optimization/87117
9744 * tree-ssa-pre.c (compute_avail): Do not make expressions
9745 with predicated values available.
9746 (get_expr_value_id): Assert we do not run into predicated value
9749 2018-08-28 Richard Biener <rguenther@suse.de>
9751 PR tree-optimization/87117
9752 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
9753 get virtual operands.
9754 (get_expr_operands): Handle STRING_CST like other decls.
9756 2018-08-28 Martin Liska <mliska@suse.cz>
9758 * tree.h: Update documentation of fndecl_built_in_p
9762 2018-08-27 Jeff Law <law@redhat.com>
9763 PR tree-optimization/87110
9764 * tree-ssa-dse.c (compute_trims): Handle non-constant
9767 2018-08-27 Martin Sebor <msebor@redhat.com>
9769 PR tree-optimization/86914
9770 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
9772 2018-08-27 Martin Sebor <msebor@redhat.com>
9774 PR tree-optimization/87112
9775 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
9776 the type of the bound argument.
9778 2018-08-27 Jeff Law <law@redhat.com>
9780 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
9781 type does not have a TYPE_SIZE_UNIT.
9783 2018-08-27 Steve Ellcey <sellcey@cavium.com>
9785 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
9786 with include of backend.h.
9788 2018-08-27 Richard Biener <rguenther@suse.de>
9790 PR tree-optimization/86927
9791 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
9792 use const cond reduction code.
9794 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
9796 PR tree-optimization/85758
9797 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
9799 2018-08-27 David Malcolm <dmalcolm@redhat.com>
9802 * diagnostic-show-locus.c (class layout_range): Update for
9803 conversion of show_caret_p to a tri-state.
9804 (layout_range::layout_range): Likewise.
9805 (make_range): Likewise.
9806 (layout::maybe_add_location_range): Likewise.
9807 (layout::should_print_annotation_line_p): Don't show annotation
9808 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
9809 (layout::get_state_at_point): Update for conversion of
9810 show_caret_p to a tri-state. Bail out early for
9811 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
9812 underlining or source colorization.
9813 (gcc_rich_location::add_location_if_nearby): Update for conversion
9814 of show_caret_p to a tri-state.
9815 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
9816 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
9818 (selftest::test_one_liner_labels): Likewise.
9819 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
9820 conversion of show_caret_p to a tri-state.
9821 * pretty-print.c (text_info::set_location): Likewise.
9822 * pretty-print.h (text_info::set_location): Likewise.
9823 * substring-locations.c (format_warning_n_va): Likewise.
9824 * tree-diagnostic.c (default_tree_printer): Likewise.
9825 * tree-pretty-print.c (newline_and_indent): Likewise.
9827 2018-08-27 David Malcolm <dmalcolm@redhat.com>
9830 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
9831 line above for line-insertion fix-it hints.
9832 (selftest::test_fixit_insert_containing_newline): Update the
9833 expected results, and add a test with line-numbering enabled.
9835 2018-08-27 Martin Liska <mliska@suse.cz>
9838 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
9839 params with DECL_HAS_VALUE_EXPR_P.
9841 2018-08-27 Martin Liska <mliska@suse.cz>
9843 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
9844 selected expansion strategy.
9846 2018-08-27 Martin Liska <mliska@suse.cz>
9848 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
9849 * builtins.c (is_builtin_fn): Likewise.
9850 * attribs.c (diag_attr_exclusions): Use new function
9851 fndecl_built_in_p and remove check for FUNCTION_DECL if
9853 (builtin_mathfn_code): Likewise.
9854 (fold_builtin_expect): Likewise.
9855 (fold_call_expr): Likewise.
9856 (fold_builtin_call_array): Likewise.
9857 (fold_call_stmt): Likewise.
9858 (set_builtin_user_assembler_name): Likewise.
9859 (is_simple_builtin): Likewise.
9860 * calls.c (gimple_alloca_call_p): Likewise.
9861 (maybe_warn_nonstring_arg): Likewise.
9862 * cfgexpand.c (expand_call_stmt): Likewise.
9863 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
9864 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
9865 (cgraph_node::verify_node): Likewise.
9866 * cgraphclones.c (build_function_decl_skip_args): Likewise.
9867 (cgraph_node::create_clone): Likewise.
9868 * config/arm/arm.c (arm_insert_attributes): Likewise.
9869 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
9870 * dse.c (scan_insn): Likewise.
9871 * expr.c (expand_expr_real_1): Likewise.
9872 * fold-const.c (operand_equal_p): Likewise.
9873 (fold_binary_loc): Likewise.
9874 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
9875 * gimple-low.c (lower_stmt): Likewise.
9876 * gimple-pretty-print.c (dump_gimple_call): Likewise.
9877 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
9879 * gimple.c (gimple_build_call_from_tree): Likewise.
9880 (gimple_call_builtin_p): Likewise.
9881 (gimple_call_combined_fn): Likewise.
9882 * gimplify.c (gimplify_call_expr): Likewise.
9883 (gimple_boolify): Likewise.
9884 (gimplify_modify_expr): Likewise.
9885 (gimplify_addr_expr): Likewise.
9886 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
9887 * ipa-cp.c (determine_versionability): Likewise.
9888 * ipa-fnsummary.c (compute_fn_summary): Likewise.
9889 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
9890 * ipa-split.c (visit_bb): Likewise.
9891 (split_function): Likewise.
9892 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
9893 * lto-cgraph.c (input_node): Likewise.
9894 * lto-streamer-out.c (write_symbol): Likewise.
9895 * omp-low.c (setjmp_or_longjmp_p): Likewise.
9896 (lower_omp_1): Likewise.
9897 * predict.c (strip_predict_hints): Likewise.
9898 * print-tree.c (print_node): Likewise.
9899 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
9900 * trans-mem.c (is_tm_irrevocable): Likewise.
9901 (is_tm_load): Likewise.
9902 (is_tm_simple_load): Likewise.
9903 (is_tm_store): Likewise.
9904 (is_tm_simple_store): Likewise.
9905 (is_tm_abort): Likewise.
9906 (tm_region_init_1): Likewise.
9907 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
9908 * tree-cfg.c (verify_gimple_call): Likewise.
9909 (move_stmt_r): Likewise.
9910 (stmt_can_terminate_bb_p): Likewise.
9911 * tree-eh.c (lower_eh_constructs_2): Likewise.
9912 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
9913 * tree-inline.c (remap_gimple_stmt): Likewise.
9914 (copy_bb): Likewise.
9915 (estimate_num_insns): Likewise.
9916 (fold_marked_statements): Likewise.
9917 * tree-sra.c (scan_function): Likewise.
9918 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
9919 (optimize_stack_restore): Likewise.
9920 (pass_fold_builtins::execute): Likewise.
9921 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
9922 (mark_all_reaching_defs_necessary_1): Likewise.
9923 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
9924 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
9925 (pass_forwprop::execute): Likewise.
9926 * tree-ssa-loop-im.c (stmt_cost): Likewise.
9927 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
9928 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
9929 * tree-ssa-strlen.c (get_string_length): Likewise.
9930 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
9931 (find_func_aliases_for_call): Likewise.
9932 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
9933 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
9934 * tree-tailcall.c (find_tail_calls): Likewise.
9935 * tree.c (need_assembler_name_p): Likewise.
9936 (free_lang_data_in_decl): Likewise.
9937 (get_call_combined_fn): Likewise.
9938 * ubsan.c (is_ubsan_builtin_p): Likewise.
9939 * varasm.c (incorporeal_function_p): Likewise.
9940 * tree.h (DECL_BUILT_IN): Remove and replace with
9942 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
9943 (fndecl_built_in_p): New.
9945 2018-08-27 Martin Liska <mliska@suse.cz>
9947 PR tree-optimization/86847
9948 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
9949 Dump also subtree probability.
9950 (switch_decision_tree::do_jump_if_equal): New function.
9951 (switch_decision_tree::emit_case_nodes): Handle special
9952 situations in balanced tree that can be emitted much simpler.
9953 Fix calculation of probabilities that happen in tree expansion.
9954 * tree-switch-conversion.h (struct cluster): Add
9956 (struct simple_cluster): Likewise.
9957 (struct case_tree_node): Add new function has_child.
9958 (do_jump_if_equal): New.
9960 2018-08-27 Martin Liska <mliska@suse.cz>
9962 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
9963 Add new argument to bit_test_cluster constructor.
9964 (bit_test_cluster::emit): Set bits really number of values
9966 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
9967 probability argument.
9968 * tree-switch-conversion.h (struct bit_test_cluster):
9969 Add m_handles_entire_switch.
9971 2018-08-27 Martin Liska <mliska@suse.cz>
9973 PR tree-optimization/86702
9974 * tree-switch-conversion.c (jump_table_cluster::emit):
9975 Make probabilities even for values in jump table
9976 according to number of cases handled.
9977 (switch_decision_tree::compute_cases_per_edge): Pass
9978 argument to reset_out_edges_aux function.
9979 (switch_decision_tree::analyze_switch_statement): Likewise.
9980 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
9983 2018-08-27 Martin Liska <mliska@suse.cz>
9985 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
9986 cfun argument explicitly.
9987 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
9988 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
9989 function gimple_switch_default_bb.
9990 (convert_switch_statements):
9992 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
9993 * stmt.c (label_to_block_fn): Use label_to_block and pass
9994 cfun argument explicitly and use gimple_switch_label_bb.
9995 (expand_case): Likewise.
9996 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
9997 cfun argument explicitly. Likewise.
9998 (make_edges_bb): Likewise.
9999 (make_cond_expr_edges): Likewise.
10000 (get_cases_for_edge): Likewise.
10001 (make_gimple_switch_edges): Likewise.
10002 (label_to_block_fn): Likewise.
10003 (label_to_block): Likewise.
10004 (make_goto_expr_edges): Likewise.
10005 (make_gimple_asm_edges): Likewise.
10006 (main_block_label): Likewise.
10007 (group_case_labels_stmt): Likewise.
10008 (find_taken_edge_computed_goto): Likewise.
10009 (find_taken_edge_switch_expr): Likewise.
10010 (gimple_verify_flow_info): Likewise.
10011 (gimple_redirect_edge_and_branch): Likewise.
10012 (gimple_switch_label_bb): New function.
10013 (gimple_switch_default_bb): Likewise.
10014 (gimple_switch_edge): Likewise.
10015 (gimple_switch_default_edge): Likewise.
10016 * tree-cfg.h (label_to_block_fn): Remove and replace ...
10017 (label_to_block): ... with this.
10018 (gimple_switch_label_bb): New.
10019 (gimple_switch_default_bb): Likewise.
10020 (gimple_switch_edge): Likewise.
10021 (gimple_switch_default_edge): Likewise.
10022 * tree-cfgcleanup.c (convert_single_case_switch): Use
10023 new gimple functions and pass new argument to label_to_block.
10024 (cleanup_control_flow_bb):
10025 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
10026 cfun argument explicitly.
10027 (make_eh_edges): Likewise.
10028 (redirect_eh_dispatch_edge): Likewise.
10029 (lower_resx): Likewise.
10030 (lower_eh_dispatch): Likewise.
10031 (maybe_remove_unreachable_handlers): Likewise.
10032 (unsplit_eh): Likewise.
10033 (cleanup_empty_eh): Likewise.
10034 (verify_eh_edges): Likewise.
10035 (verify_eh_dispatch_edge): Likewise.
10036 * tree-ssa-dom.c (record_edge_info): Likewise.
10037 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
10038 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
10039 (thread_through_normal_block): Likewise.
10040 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
10041 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
10042 * tree-switch-conversion.c (switch_conversion::collect): Use new
10044 (switch_conversion::check_final_bb): Likewise.
10045 (switch_conversion::gather_default_values): Pass new argument
10047 (switch_conversion::build_constructors): Likewise.
10048 (switch_decision_tree::compute_cases_per_edge): Use new
10049 gimple_switch_edge function.
10050 (switch_decision_tree::analyze_switch_statement): Pass new argument
10052 (switch_decision_tree::try_switch_expansion): Use
10053 gimple_switch_default_edge.
10054 * tree-vrp.c (find_switch_asserts): Pass new argument
10056 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
10057 (vr_values::simplify_switch_using_ranges): Likewise.
10059 2018-08-27 Richard Biener <rguenther@suse.de>
10061 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
10062 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
10064 * tree-ssa-sccvn.h (struct vn_pval): New structure.
10065 (struct vn_nary_op_s): Add unwind_to member. Add
10066 predicated_values flag and put result into a union together
10067 with a linked list of vn_pval.
10068 (struct vn_ssa_aux): Add name member to make maintaining
10069 a map of SSA name to vn_ssa_aux possible. Remove no longer
10070 needed info, dfsnum, low, visited, on_sccstack, use_processed
10071 and range_info_anti_range_p members.
10072 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
10073 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
10075 (vn_valueize): New global.
10076 (vn_context_bb): Likewise.
10077 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
10078 VN_INFO_PTR_INFO): Remove.
10079 * tree-ssa-sccvn.c: ... (rewrite)
10080 (pass_fre::execute): For -O2+ initialize loops and run
10081 RPO VN in optimistic mode (iterating). For -O1 and -Og
10082 run RPO VN in non-optimistic mode.
10083 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
10084 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
10085 * doc/invoke.texi (sccvn-max-scc-size): Remove.
10086 (rpo-vn-max-loop-depth): Document.
10087 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
10088 when valuezing the VUSE signals we walked out of the region.
10089 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
10090 (phi_translate): Set VN context block to use for availability
10092 (compute_avail): Likewise.
10093 (pre_valueize): New function.
10094 (pass_pre::execute): Adjust to the RPO VN API.
10096 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
10097 (propagate_constants_for_unrolling): Remove.
10098 (tree_unroll_loops_completely): Perform value-numbering
10099 on the unrolled bodies loop parent.
10101 2018-08-27 Richard Biener <rguenther@suse.de>
10103 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
10104 for partial antic compute.
10106 2018-08-27 Jakub Jelinek <jakub@redhat.com>
10108 PR rtl-optimization/87065
10109 * combine.c (simplify_if_then_else): Formatting fix.
10110 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
10112 (known_cond): Don't return const_true_rtx for vector modes. Use
10113 CONST0_RTX instead of const0_rtx. Formatting fixes.
10115 2018-08-27 Martin Liska <mliska@suse.cz>
10117 PR gcov-profile/87069
10118 * gcov.c (process_file): Record files already processed
10119 and warn about a file being processed multiple times.
10121 2018-08-27 Martin Liska <mliska@suse.cz>
10124 * config/aarch64/aarch64.c (aarch64_override_options_internal):
10125 Set default values for x_aarch64_*_string strings.
10126 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
10127 prefix. For -mabi do not print '=ABI' in help and use
10128 <option_value> format for -msve-vector-bits and -moverride
10131 2018-08-26 Jeff Law <law@redhat.com>
10133 * config/mips/frame-header-opt.c: Include "backend.h" rather than
10136 2018-08-26 Marek Polacek <polacek@redhat.com>
10138 PR c++/87029, Implement -Wredundant-move.
10139 * doc/invoke.texi: Document -Wredundant-move.
10141 2018-08-25 Martin Sebor <msebor@redhat.com>
10143 PR tree-optimization/87059
10144 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
10145 to the same type as the other.
10146 * fold-const.c (fold_binary_loc): Assert expectation.
10148 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
10150 * config/darwin.c (machopic_legitimize_pic_address): Clean up
10151 extraneous parentheses, dead code section and formatting.
10153 2018-08-24 David Malcolm <dmalcolm@redhat.com>
10156 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
10157 wide enough for jumps in the line-numbering to be visible.
10158 (layout::print_gap_in_line_numbering): New member function.
10159 (layout::calculate_line_spans): When using line numbering, merge
10160 line spans that are only 1 line apart.
10161 (diagnostic_show_locus): When printing line numbers, show gaps in
10162 line numbering directly, rather than printing headers.
10163 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
10164 line-numbering with multiple line spans.
10165 (selftest::test_fixit_insert_containing_newline_2): Add test of
10166 line-numbering, in which the spans are close enough to be merged.
10168 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
10170 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
10171 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
10172 * tree-vrp.c (range_is_nonnull): Remove.
10173 (range_includes_zero_p): Accept value_range instead of min/max.
10174 (extract_range_from_binary_expr_1): Do not early bail on
10176 Use range_includes_zero_p instead of range_is_nonnull.
10177 (extract_range_from_unary_expr): Use range_includes_zero_p instead
10178 of range_is_nonnull.
10179 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
10180 special case VR_ANTI_RANGE.
10181 (vrp_finalize): Same.
10182 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
10183 instead of min/max.
10184 (range_is_nonnull): Remove.
10185 * vr-values.c (vrp_stmt_computes_nonzero): Use
10186 range_includes_zero_p instead of range_is_nonnull.
10187 (extract_range_basic): Pass value_range to range_includes_zero_p
10188 instead of range_is_nonnull.
10190 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
10192 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
10193 * emit-rtl.h (rtl_data): Remove return_bnd.
10194 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
10195 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
10196 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
10197 (POINTER_BOUNDS_MODE): Remove definition.
10198 (make_pointer_bounds_mode): Remove.
10199 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
10200 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
10201 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
10202 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
10203 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
10204 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
10205 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
10207 * config/i386/i386-modes.def (BND32, BND64): Remove.
10208 * config/i386/i386.c (dbx_register_map): Remove bound registers.
10209 (dbx64_register_map): Ditto.
10210 (svr4_dbx_register_map): Ditto.
10211 (indirect_thunk_bnd_needed): Remove.
10212 (indirect_thunks_bnd_used): Ditto.
10213 (indirect_return_bnd_needed): Ditto.
10214 (indirect_return_via_cx_bnd): Ditto.
10215 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
10216 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
10217 (output_indirect_thunk): Ditto. Remove need_prefix argument.
10218 (output_indirect_thunk_function): Remove handling of
10219 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
10220 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
10221 (ix86_save_reg): Remove handling of crtl->return_bnd.
10222 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
10223 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
10224 and UNSPEC_BNDLX_ADDR.
10225 (ix86_output_indirect_branch_via_reg): Remove handling of
10226 indirect_thunk_prefix_bnd.
10227 (ix86_output_indirect_branch_via_push): Ditto.
10228 (ix86_output_function_return): Ditto.
10229 (ix86_output_indirect_function_return): Ditto.
10230 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
10231 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
10232 (CALL_USED_REGISTERS): Ditto.
10233 (REG_ALLOC_ORDER): Update for removal of bound registers.
10234 (HI_REGISTER_NAMES): Ditto.
10235 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
10236 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
10237 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
10238 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
10239 (FIRST_PSEUDO_REG): Update.
10240 (BND): Remove mode iterator.
10241 * config/i386/predicates.md (bnd_mem_operator): Remove.
10243 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
10245 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
10248 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
10250 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
10251 the case in which the permute needs only a single element and
10252 repeats for every vector of the result. Extend that case to
10253 handle variable-length vectors.
10254 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
10256 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
10259 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
10260 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
10262 2018-08-24 Richard Biener <rguenther@suse.de>
10264 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
10265 bb_flags_allocated members.
10266 (auto_flag): New RAII class for allocating flags.
10267 (auto_edge_flag): New RAII class for allocating edge flags.
10268 (auto_bb_flag): New RAII class for allocating bb flags.
10269 * cfgloop.c (verify_loop_structure): Allocate temporary edge
10271 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
10272 in favor of temporarily allocated BB flag.
10273 * hsa-brig.c: Re-order includes.
10274 * hsa-dump.c: Likewise.
10275 * hsa-regalloc.c: Likewise.
10276 * print-rtl.c: Likewise.
10277 * profile-count.c: Likewise.
10279 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
10282 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
10285 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
10288 * wide-int-range.cc (wide_int_range_div): Do not ignore result
10289 from wide_int_range_multiplicative_op.
10291 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10293 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
10296 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
10298 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
10301 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
10303 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
10304 full GENERIC expression used for replacement.
10306 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
10308 * tree-vrp.c (abs_extent_range): Remove.
10309 (extract_range_into_wide_ints): Pass wide ints by reference.
10310 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
10311 Pass wide ints by reference in all calls to
10312 extract_range_into_wide_ints.
10313 * wide-int-range.cc (wide_int_range_div): New.
10314 * wide-int-range.h (wide_int_range_div): New.
10315 (wide_int_range_includes_zero_p): New.
10316 (wide_int_range_zero_p): New.
10318 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
10320 * config/aarch64/aarch64.md (arches): New enum.
10321 (arch): New enum attr.
10322 (arch_enabled): New attr.
10323 (enabled): Now uses arch_enabled only.
10324 (simd, sve, fp16): Removed attribute.
10325 (fp): Attr now defined in terms of 'arch'.
10326 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
10327 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
10328 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
10329 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
10330 attributes into 'arch'.
10331 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
10332 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
10333 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
10334 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
10335 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
10336 'simd' attribute into 'arch'.
10337 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
10338 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
10339 Convert use of 'fp' attribute to 'arch'.
10340 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
10341 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
10343 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
10344 (different modes) Merge 'fp' and 'simd' into 'arch'.
10345 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
10346 'simd' into 'arch'.
10348 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
10350 PR rtl-optimization/87026
10351 * expmed.c (canonicalize_comparison): If we can no longer create
10352 pseudoregisters, don't.
10354 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
10357 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
10359 * config/arm/arm.c (speculation_barrier_libfunc): New static
10361 (arm_init_libfuncs): Initialize it.
10362 (arm_emit_speculation_barrier): New function.
10363 * config/arm/arm.md (speculation_barrier): Call
10364 arm_emit_speculation_barrier for architectures that do not have
10366 (speculation_barrier_insn): Only match on Armv7 or later.
10368 2018-08-23 Richard Biener <rguenther@suse.de>
10370 PR middle-end/87024
10371 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
10374 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
10376 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
10377 of single-vector TBLs.
10378 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
10379 one input is given.
10381 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
10384 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
10385 aarch64_evpc_tbl guard.
10387 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
10389 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
10392 2018-08-22 Martin Sebor <msebor@redhat.com>
10394 PR middle-end/87052
10395 * tree-pretty-print.c (pretty_print_string): Add argument.
10396 (dump_generic_node): Call to pretty_print_string with string size.
10398 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
10400 PR rtl-optimization/86771
10401 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
10402 of two SETs into those two SETs, one to be placed at i2, if that SETs
10403 destination is modified between i2 and i3.
10405 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
10407 PR tree-optimization/86725
10408 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
10410 (vect_analyze_scalar_cycles_1): Check it.
10412 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
10414 PR tree-optimization/86725
10415 * tree-vect-loop.c (vect_is_simple_reduction): When treating
10416 an outer loop phi as a double reduction, make sure that the
10417 single user of the phi result is an inner loop phi.
10419 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
10421 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
10422 grouped stores with gaps to a strided group.
10424 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
10426 * tree-vect-stmts.c (get_group_load_store_type)
10427 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
10428 first statement in a group.
10430 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
10432 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
10433 the sequence used in gcc/gcc.c.
10435 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
10438 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
10441 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
10443 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
10444 Darwin10-specific unwinder-shim.
10445 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
10446 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
10447 New to cater for Darwin10 Rosetta.
10449 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
10451 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
10452 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
10454 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
10459 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
10460 (dwarf2out_switch_text_section): Generate a local label for the
10461 second function sub-section and apply it as the second FDE start
10463 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
10464 second sub-section start.
10466 2018-08-22 Richard Biener <rguenther@suse.de>
10468 PR tree-optimization/86988
10469 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
10471 2018-08-22 Richard Biener <rguenther@suse.de>
10473 PR tree-optimization/86945
10474 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
10476 2018-08-22 Alexandre Oliva <oliva@adacore.com>
10478 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
10479 a comment about how uses of r2 for .sdata2 come about.
10481 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
10483 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
10485 2018-08-21 Marek Polacek <polacek@redhat.com>
10487 PR c++/86981, Implement -Wpessimizing-move.
10488 * doc/invoke.texi: Document -Wpessimizing-move.
10490 2018-08-21 Jan Hubicka <jh@suse.cz>
10492 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
10493 * tree.h (is_redundant_typedef): Remove.
10494 * dwarf2out.c (is_redundant_typedef): Turn into static function.
10496 2018-08-21 Jan Hubicka <jh@suse.cz>
10498 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
10501 2018-08-21 Tamar Christina <tamar.christina@arm.com>
10503 * expmed.c (extract_low_bits): Reject invalid subregs early.
10505 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
10507 PR middle-end/86121
10508 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
10511 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
10513 * config/vxworks.h: Guard vxworks_asm_out_constructor and
10514 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
10515 * config/vxworks.c: Likewise.
10517 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
10519 * config/vxworks.c: Set targetm.have_ctors_dtors
10520 if HAVE_INITFINI_ARRAY_SUPPORT.
10521 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
10522 if HAVE_INITFINI_ARRAY_SUPPORT.
10524 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
10526 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
10527 default search path for VxWorks < 7.
10529 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
10531 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
10532 (get_format_string): Refer to c_getstr.
10534 2018-08-21 Tom de Vries <tdevries@suse.de>
10536 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
10537 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
10538 (debuginfo_early_stop): Declare.
10539 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
10540 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
10541 (debuginfo_early_stop): New function.
10542 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
10543 and debuginfo_early_stop.
10544 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
10545 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
10546 (general_init): Call debuginfo_early_init.
10547 (finalize): Call debuginfo_fini.
10548 (do_compile): Call debuginfo_init.
10549 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
10550 -fdump-early-debug.
10551 (@item -fdump-debug, @item -fdump-earlydebug): Add.
10553 2018-08-21 Tom de Vries <tdevries@suse.de>
10555 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
10556 flag_dump_noaddr and flag_dump_unnumbered.
10558 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
10560 * wide-int-range.cc (wide_int_range_abs): New.
10561 (wide_int_range_order_set): Rename from wide_int_range_min_max.
10562 * wide-int-range.h (wide_int_range_abs): New.
10563 (wide_int_range_min_max): New.
10564 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
10565 case to call wide_int_range_abs.
10566 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
10567 (extract_range_from_abs_expr): Delete.
10569 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
10572 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
10573 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
10576 2018-08-20 Nathan Sidwell <nathan@acm.org>
10577 Jeff Law <law@redhat.com>
10579 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
10580 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
10582 2018-08-20 David Malcolm <dmalcolm@redhat.com>
10585 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
10586 (decl_attributes): Likewise.
10587 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
10589 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
10590 * diagnostic-core.h (class auto_diagnostic_group): New class.
10591 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
10592 (diagnostic_report_diagnostic): Handle the first diagnostics within
10594 (emit_diagnostic): Add auto_diagnostic_group instance.
10595 (inform): Likewise.
10596 (inform_n): Likewise.
10597 (warning): Likewise.
10598 (warning_at): Likewise.
10599 (warning_n): Likewise.
10600 (pedwarn): Likewise.
10601 (permerror): Likewise.
10603 (error_n): Likewise.
10604 (error_at): Likewise.
10606 (fatal_error): Likewise.
10607 (internal_error): Likewise.
10608 (internal_error_no_backtrace): Likewise.
10609 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
10610 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
10611 * diagnostic.h (struct diagnostic_context): Add fields
10612 "diagnostic_group_nesting_depth",
10613 "diagnostic_group_emission_count", "begin_group_cb",
10615 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
10616 Add auto_diagnostic_group instance(s).
10617 (find_explicit_erroneous_behavior): Likewise.
10618 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
10619 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
10620 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
10621 (gimplify_va_arg_expr): Likewise.
10622 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
10623 (HSA_SORRY_AT): Likewise.
10624 * ipa-devirt.c (compare_virtual_tables): Likewise.
10625 (warn_odr): Likewise.
10626 * multiple_target.c (expand_target_clones): Likewise.
10627 * opts-common.c (cmdline_handle_error): Likewise.
10628 * reginfo.c (globalize_reg): Likewise.
10629 * substring-locations.c (format_warning_n_va): Likewise.
10630 * tree-inline.c (expand_call_inline): Likewise.
10631 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
10632 * tree-ssa-loop-niter.c
10633 (do_warn_aggressive_loop_optimizations): Likewise.
10634 * tree-ssa-uninit.c (warn_uninit): Likewise.
10635 * tree.c (warn_deprecated_use): Likewise.
10637 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
10640 * config/i386/i386.md (eh_return): Always update EH return
10641 address in word_mode.
10643 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
10645 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
10646 TARGET_SPLIT_COMPLEX_ARG is defined.
10648 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
10650 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
10652 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
10655 * expr.c (expand_assignment): Assert that bitpos is positive.
10656 (store_field): Likewise
10657 (expand_expr_real_1): Make sure that bitpos is positive.
10658 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
10661 2018-08-20 Nathan Sidwell <nathan@acm.org>
10663 * Makefile.in (CPP_ID_DATA_H): Delete.
10664 (CPP_INTERNAL_H): Don't add it.
10665 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
10666 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
10668 2018-08-20 Richard Biener <rguenther@suse.de>
10670 PR tree-optimization/78655
10671 * tree-vrp.c (extract_range_from_binary_expr_1): Make
10672 pointer + offset nonnull if either operand is nonnull work.
10674 2018-08-20 Tom de Vries <tdevries@suse.de>
10676 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
10677 unless the referenced die describes the added property using
10678 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
10679 Otherwise, add a DW_AT_location to the referenced die.
10681 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
10684 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
10685 register_operand when calling ix86_set_reg_reg_cost.
10686 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
10687 Set *total to 0 for operands that satisfy x86_64_immediate_operand
10688 predicate and to 1 otherwise.
10690 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
10692 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
10693 emit a diagnostic that it is not supported and reset the option.
10694 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
10695 supported and consume the option. (ASM_FINAL_SPEC): New.
10697 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
10699 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
10702 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
10704 C-SKY port: Documentation
10706 * doc/extend.texi (C-SKY Function Attributes): New section.
10707 * doc/invoke.texi (Option Summary): Add C-SKY options.
10708 (C-SKY Options): New section.
10709 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
10711 2018-08-17 Jojo <jijie_rong@c-sky.com>
10712 Huibin Wang <huibin_wang@c-sky.com>
10713 Sandra Loosemore <sandra@codesourcery.com>
10714 Chung-Lin Tang <cltang@codesourcery.com>
10716 C-SKY port: Backend implementation
10718 * config/csky/*: New.
10719 * common/config/csky/*: New.
10721 2018-08-17 Jojo <jijie_rong@c-sky.com>
10722 Huibin Wang <huibin_wang@c-sky.com>
10723 Sandra Loosemore <sandra@codesourcery.com>
10724 Chung-Lin Tang <cltang@codesourcery.com>
10725 Andrew Jenner <andrew@codesourcery.com>
10727 C-SKY port: Configury
10729 * config.gcc (csky-*-*): New.
10730 * configure.ac: Add csky to targets for dwarf2 debug_line support.
10731 * configure: Regenerated.
10733 2018-08-17 David Malcolm <dmalcolm@redhat.com>
10735 * dump-context.h: Include "dumpfile.h".
10736 (dump_context::dump_printf_va): Convert final param from va_list
10737 to va_list *. Convert from ATTRIBUTE_PRINTF to
10738 ATTRIBUTE_GCC_DUMP_PRINTF.
10739 (dump_context::dump_printf_loc_va): Likewise.
10740 * dumpfile.c: Include "stringpool.h".
10741 (make_item_for_dump_printf_va): Delete.
10742 (make_item_for_dump_printf): Delete.
10743 (class dump_pretty_printer): New class.
10744 (dump_pretty_printer::dump_pretty_printer): New ctor.
10745 (dump_pretty_printer::emit_items): New member function.
10746 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
10748 (dump_pretty_printer::emit_item): New member function.
10749 (dump_pretty_printer::stash_item): New member function.
10750 (dump_pretty_printer::format_decoder_cb): New member function.
10751 (dump_pretty_printer::decode_format): New member function.
10752 (dump_context::dump_printf_va): Reimplement in terms of
10753 dump_pretty_printer.
10754 (dump_context::dump_printf_loc_va): Convert final param from va_list
10756 (dump_context::begin_scope): Reimplement call to
10757 make_item_for_dump_printf.
10758 (dump_printf): Update for change to dump_printf_va.
10759 (dump_printf_loc): Likewise.
10760 (selftest::test_capture_of_dump_calls): Convert "stmt" from
10761 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
10762 with %T, %E, and %G.
10763 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
10764 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
10765 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
10766 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
10767 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
10768 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
10769 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
10770 within a dump_printf_loc call to "%wu".
10771 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
10772 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
10773 missing space after "=".
10774 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
10775 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
10776 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
10777 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
10778 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
10779 duplicate "vectorized" from message.
10781 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
10783 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
10784 polyNxK_t element's TYPE_STRING_FLAG.
10786 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
10788 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
10789 (they were unnamed before). Fix comments.
10791 2018-08-17 Nathan Sidwell <nathan@acm.org>
10793 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
10795 2018-08-17 Richard Biener <rguenther@suse.de>
10797 PR tree-optimization/86841
10798 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
10800 2018-08-17 Martin Liska <mliska@suse.cz>
10802 * common.opt: Remove Warn, Init and Report for options with
10803 Ignore/Deprecated flag. Warning is done automatically for
10805 * config/i386/i386.opt: Likewise.
10806 * config/ia64/ia64.opt: Likewise.
10807 * config/rs6000/rs6000.opt: Likewise.
10808 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
10809 Remove usage of flag_check_pointer_bounds.
10810 * lto-wrapper.c (merge_and_complain): Do not handle
10811 OPT_fcheck_pointer_bounds.
10812 (append_compiler_options): Likewise.
10813 * opt-functions.awk: Do not handle Deprecated.
10814 * optc-gen.awk: Check that Var, Report and Init are not
10815 used for an option with Ignore/Deprecated flag.
10816 * opts-common.c (decode_cmdline_option): Do not report
10818 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
10820 * opts.h (struct cl_option): Remove cl_deprecated flag.
10821 (CL_ERR_DEPRECATED): Remove error enum value.
10823 2018-08-17 Richard Biener <rguenther@suse.de>
10825 PR middle-end/86505
10826 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
10827 across a va-arg-pack using call adjust its return value accordingly.
10829 2018-08-16 Martin Sebor <msebor@redhat.com>
10831 PR tree-optimization/86853
10832 * gimple-ssa-sprintf.c (struct format_result): Rename member.
10833 (struct fmtresult): Add member and initialize it in ctors.
10834 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
10835 (format_string): Handle %S the same as %ls. Set MAYFAIL.
10836 (format_directive): Set POSUNDER4K when MAYFAIL is set.
10837 (parse_directive): Handle %C same as %c.
10838 (sprintf_dom_walker::compute_format_length): Adjust.
10839 (is_call_safe): Adjust.
10841 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
10843 * builtins.c (c_strlen): Add new parameter eltsize. Use it
10844 for determining how to count the elements.
10845 * builtins.h (c_strlen): Adjust prototype.
10846 * expr.c (string_constant): Add new parameter mem_size.
10847 Set *mem_size appropriately.
10848 * expr.h (string_constant): Adjust protoype.
10849 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
10850 * gimple-fold.h (get_range_strlen): Adjust prototype.
10851 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
10852 (format_string): Call get_string_length with eltsize.
10854 2018-08-16 David Malcolm <dmalcolm@redhat.com>
10856 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
10857 to emit the span, rather than setting it as the prefix.
10859 2018-08-16 David Malcolm <dmalcolm@redhat.com>
10861 * diagnostic-show-locus.c (layout::start_annotation_line): Add
10862 "margin_char" parameter, defaulting to space. Use it in place
10863 of pp_space for the initial part of the margin.
10864 (layout::print_leading_fixits): Use '+' when filling the margin
10865 of line-insertion fix-it hints.
10867 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
10869 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
10872 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
10874 * config/rs6000/altivec.md: Don't set length attribute to the default
10876 * config/rs6000/darwin.md: Ditto.
10877 * config/rs6000/dfp.md: Ditto.
10878 * config/rs6000/htm.md: Ditto.
10879 * config/rs6000/rs6000.md: Ditto.
10880 * config/rs6000/sync.md: Ditto.
10881 * config/rs6000/vsx.md: Ditto.
10883 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
10885 * config/rs6000/altivec.md: Don't set length attribute to the default
10886 value, for branch instructions.
10887 * config/rs6000/darwin.md: Ditto.
10888 * config/rs6000/rs6000.md: Ditto.
10890 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
10892 * config/rs6000/rs6000.md (length): Always define as const_int 4.
10893 (unnamed conditional branch define_insn): Set length to 4 or 8
10894 depending on offset.
10895 (<bd>_<mode>): Similar, for alternative 0.
10896 (<bd>tf_<mode>): Ditto.
10898 2018-08-16 Tamar Christina <tamar.christina@arm.com>
10900 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
10902 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
10904 * doc/rtl.texi: Replace old RTX class names with new names.
10907 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
10909 * expmed.h (canonicalize_comparison): New declaration.
10910 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
10911 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
10912 * optabs.c (prepare_cmp_insn): Likewise.
10913 * rtl.h (unsigned_condition_p): New function which checks if a
10914 comparison operator is unsigned.
10916 2018-08-16 Nathan Sidwell <nathan@acm.org>
10918 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
10919 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
10921 2018-08-16 Tamar Christina <tamar.christina@arm.com>
10924 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
10925 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
10926 (mov<mov>): ..this and enable unconditionally.
10928 2018-08-16 Tamar Christina <tamar.christina@arm.com>
10930 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
10932 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
10934 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
10935 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
10936 "Common" with "Target".
10938 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
10940 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
10941 * doc/invoke.texi (mmitigate-rop): Remove.
10942 * config/i386/i386.c: Do not include "regrename.h".
10943 (ix86_rop_should_change_byte_p, reg_encoded_number)
10944 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
10946 (ix86_reorg): Remove call to ix86_mitigate_rop.
10947 * config/i386/i386.md (attr "modrm_class"): Remove.
10948 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
10949 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
10950 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
10952 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
10954 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
10955 allow folding of mergeh() and mergel() for the float and double types.
10956 (fold_mergehl_helper): Rework to handle building a permute tree
10959 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
10961 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
10964 2018-08-15 David Malcolm <dmalcolm@redhat.com>
10966 * common.opt (fdiagnostics-show-labels): New option.
10967 * diagnostic-show-locus.c (class layout_range): Add field
10969 (class layout): Add field "m_show_labels_p".
10970 (layout_range::layout_range): Add param "label" and use it to
10971 initialize m_label.
10972 (make_range): Pass in NULL for new "label" param of layout_range's
10974 (layout::layout): Initialize m_show_labels_p.
10975 (layout::maybe_add_location_range): Pass in loc_range->m_label
10976 when constructing layout_range instances.
10977 (struct line_label): New struct.
10978 (layout::print_any_labels): New member function.
10979 (layout::print_line): Call it if label-printing is enabled.
10980 (selftest::test_one_liner_labels): New test.
10981 (selftest::test_diagnostic_show_locus_one_liner): Call it.
10982 * diagnostic.c (diagnostic_initialize): Initialize
10983 context->show_labels_p.
10984 * diagnostic.h (struct diagnostic_context): Add field
10986 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
10987 -fno-diagnostics-show-labels.
10988 * dwarf2out.c (gen_producer_string): Add
10989 OPT_fdiagnostics_show_labels to the ignored options.
10990 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
10992 (gcc_rich_location::maybe_add_expr): Likewise.
10993 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
10994 label" param, defaulting to NULL.
10995 (gcc_rich_location::add_expr): Add "label" param.
10996 (gcc_rich_location::maybe_add_expr): Likewise.
10997 (class text_range_label): New class.
10998 (class range_label_for_type_mismatch): New class.
10999 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
11000 of format_warning_va.
11001 (fmtwarn_n): Likewise for new params of format_warning_n_va.
11002 * lto-wrapper.c (merge_and_complain): Add
11003 OPT_fdiagnostics_show_labels to the "pick one setting" options.
11004 (append_compiler_options): Likewise to the dropped options.
11005 (append_diag_options): Likewise to the passed-on options.
11006 * opts.c (common_handle_option): Handle the new option.
11007 * selftest-diagnostic.c
11008 (test_diagnostic_context::test_diagnostic_context): Enable
11010 * substring-locations.c: Include "gcc-rich-location.h".
11011 (format_warning_n_va): Add "fmt_label" and "param_label" params
11012 and use them as appropriate.
11013 (format_warning_va): Add "fmt_label" and "param_label" params,
11014 passing them on to format_warning_n_va.
11015 (format_warning_at_substring): Likewise.
11016 (format_warning_at_substring_n): Likewise.
11017 * substring-locations.h (format_warning_va): Add "fmt_label" and
11018 "param_label" params.
11019 (format_warning_n_va): Likewise.
11020 (format_warning_at_substring): Likewise.
11021 (format_warning_at_substring_n): Likewise.
11022 * toplev.c (general_init): Initialize global_dc->show_labels_p.
11024 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
11027 * builtins.c (expand_builtin_memcmp): Do not expand the call
11028 when overflow is detected.
11030 2018-08-15 Martin Sebor <msebor@redhat.com>
11032 PR tree-optimization/71625
11033 * config/aarch64/aarch64-builtins.c
11034 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
11036 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
11038 * config/s390/s390.c (s390_reorg): Remove loop.
11040 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
11043 (darwin_function_switched_text_sections): Delete.
11044 * gcc/config/darwin.h
11045 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
11047 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
11050 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
11051 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
11052 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
11054 2018-08-15 Martin Liska <mliska@suse.cz>
11056 PR tree-optimization/86925
11057 * predict.c (expr_expected_value_1): When taking
11058 later predictor, assign also probability.
11059 Use fold_build2_initializer_loc in order to fold
11060 the expression in -frounding-math.
11062 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
11064 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
11066 (expand_vec_perm_1): Try the new method.
11068 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
11071 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
11072 Check whether lra_live_max_point is 0 before dividing.
11074 2018-08-14 Martin Sebor <msebor@redhat.com>
11076 PR tree-optimization/86650
11077 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
11078 (vrp_prop::check_mem_ref): Same.
11080 2018-08-13 Liu Hao <lh_mouse@126.com>
11082 * pretty-print.c (eat_esc_sequence): Swap the foreground and
11083 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
11084 and clear it thereafter, as it only works for DBCS.
11086 2018-08-13 Liu Hao <lh_mouse@126.com>
11088 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
11089 handle returned by _get_osf_handle().
11091 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
11093 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
11094 for folding vec_perm.
11096 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
11098 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
11099 Add support for gimple-folding of vec_pack() and vec_unpack()
11102 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
11104 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
11105 vec_xst variants to the list.
11106 (rs6000_gimple_fold_builtin): Add support for folding unaligned
11107 vector loads and stores.
11109 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
11111 * config.gcc (rs6000-ibm-aix4.x): Delete.
11112 (rs6000-ibm-aix5.1): Delete.
11113 (rs6000-ibm-aix5.2): Delete.
11114 (rs6000-ibm-aix5.3): Delete.
11115 * config/rs6000/aix43.h: Delete.
11116 * config/rs6000/aix51.h: Delete.
11117 * config/rs6000/aix52.h: Delete.
11118 * config/rs6000/t-aix43: Delete.
11120 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
11122 * config/s390/s390.c (s390_decompose_constant_pool_ref):
11124 (s390_decompose_address): Factor out constant pool ref
11127 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
11129 * config/nds32/nds32-predicates.c
11130 (nds32_can_use_bclr_p): Change return type as bool.
11131 (nds32_can_use_bset_p): Ditto.
11132 (nds32_can_use_btgl_p): Ditto.
11133 (nds32_can_use_bitci_p): Ditto.
11134 * config/nds32/nds32-protos.h
11135 (nds32_can_use_bclr_p): Change declaration.
11136 (nds32_can_use_bset_p): Ditto.
11137 (nds32_can_use_btgl_p): Ditto.
11138 (nds32_can_use_bitci_p): Ditto.
11140 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
11142 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
11143 Support -msched-prolog-epilog option.
11144 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
11146 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
11148 * common/config/nds32/nds32-common.c
11149 (nds32_option_optimization_table): Enalbe -malways-align.
11151 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
11153 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
11155 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
11156 OPT_misr_secure_ case.
11157 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
11158 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
11159 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
11161 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
11162 (nds32_isr_info): New field security_level.
11163 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
11164 * config/nds32/nds32.md (return_internal): Consider critical attribute.
11165 * config/nds32/nds32.opt (misr-secure): New option.
11166 * config/nds32/nds32_init.inc: New file.
11167 * config/nds32/nds32_isr.h: New file.
11169 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
11171 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
11172 Update comment for atomic instructions.
11173 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
11174 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
11176 (atomic_loaddi): Revise fence expansion to only emit fence prior to
11177 load for __ATOMIC_SEQ_CST model.
11178 (atomic_loaddi_1): Remove float register target.
11179 (atomic_storedi): Handle CONST_INT values.
11180 (atomic_storedi_1): Remove float register source. Add special case
11182 (memory_barrier): New expander and insn.
11184 2018-08-11 Jakub Jelinek <jakub@redhat.com>
11186 PR tree-optimization/86835
11187 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
11188 new_stmt after def_gsi, make sure to insert new_square_stmt after
11189 that stmt, not 2 stmts before it.
11191 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
11194 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
11197 2018-08-10 Martin Liska <mliska@suse.cz>
11200 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
11202 * builtins.c (expand_builtin_expect_with_probability):
11204 (expand_builtin_expect_with_probability): New function.
11205 (build_builtin_expect_predicate): Add new argumnet probability
11206 for BUILT_IN_EXPECT_WITH_PROBABILITY.
11207 (fold_builtin_expect):
11210 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
11211 * builtins.h (fold_builtin_expect): Set new argument.
11212 * doc/extend.texi: Document __builtin_expect_with_probability.
11213 * doc/invoke.texi: Likewise.
11214 * gimple-fold.c (gimple_fold_call): Pass new argument.
11215 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
11216 also BUILT_IN_EXPECT_WITH_PROBABILITY.
11217 * predict.c (get_predictor_value): New function.
11218 (expr_expected_value): Add new argument probability. Assume
11219 that predictor and probability are always non-null.
11220 (expr_expected_value_1): Likewise. For __builtin_expect and
11221 __builtin_expect_with_probability set probability. Handle
11222 combination in binary expressions.
11223 (tree_predict_by_opcode): Simplify code by simply calling
11224 get_predictor_value.
11225 (pass_strip_predict_hints::execute): Add handling of
11226 BUILT_IN_EXPECT_WITH_PROBABILITY.
11227 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
11229 * tree.h (DECL_BUILT_IN_P): New function.
11231 2018-08-10 Martin Liska <mliska@suse.cz>
11233 PR tree-optimization/85799
11234 * passes.def: Add argument for pass_strip_predict_hints.
11235 * predict.c (class pass_strip_predict_hints): Add new argument
11237 (strip_predictor_early): New function.
11238 (pass_strip_predict_hints::execute): Call the function to
11240 (strip_predict_hints): New function.
11241 * predict.def: Fix comment.
11243 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
11245 * Makefile.in: Clarify which tm.texi to copy over to assert the
11246 right to grant a GFDL license for all.
11248 2018-08-09 Jeff Law <law@redhat.com>
11250 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
11253 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
11255 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
11258 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
11260 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
11261 reductions for variable-length vectors.
11263 2018-08-09 David Malcolm <dmalcolm@redhat.com>
11266 * common.opt (fdiagnostics-show-line-numbers): New option.
11267 * diagnostic-show-locus.c (class layout): Add fields
11268 "m_show_line_numbers_p" and "m_linenum_width";
11269 (num_digits): New function.
11270 (test_num_digits): New function.
11271 (layout::layout): Initialize new fields. Update m_x_offset
11272 logic to handle any left margin.
11273 (layout::print_source_line): Print line number when requested.
11274 (layout::start_annotation_line): New member function.
11275 (layout::print_annotation_line): Call it.
11276 (layout::print_leading_fixits): Likewise.
11277 (layout::print_trailing_fixits): Likewise. Update calls to
11278 move_to_column for new parameter.
11279 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
11280 it to potentially call start_annotation_line.
11281 (layout::show_ruler): Call start_annotation_line.
11282 (selftest::test_line_numbers_multiline_range): New selftest.
11283 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
11284 and selftest::test_line_numbers_multiline_range.
11285 * diagnostic.c (diagnostic_initialize): Initialize
11286 show_line_numbers_p.
11287 * diagnostic.h (struct diagnostic_context): Add field
11288 "show_line_numbers_p".
11289 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
11290 -fno-diagnostics-show-line-numbers.
11291 * dwarf2out.c (gen_producer_string): Add
11292 OPT_fdiagnostics_show_line_numbers to the ignored options.
11293 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
11294 one setting" options.
11295 (append_compiler_options): Likewise to the dropped options.
11296 (append_diag_options): Likewise to the passed-on options.
11297 * opts.c (common_handle_option): Handle the new option.
11298 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
11300 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
11302 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
11303 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
11304 third argument of type "const signed char" to descriptions of
11305 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
11306 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
11307 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
11308 __builtin_bcdsub_ov functions.
11310 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
11312 PR tree-optimization/86858
11313 * tree-vect-loop.c (vect_is_simple_reduction): Restore
11314 flow_bb_inside_loop_p calls.
11316 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
11318 PR tree-optimization/86871
11319 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
11320 instead of gimple_assign_lhs.
11322 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
11325 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
11326 register constraint to operand 0.
11327 (add<mode>3_carryinC): Likewise.
11328 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
11330 2018-08-09 Martin Liska <mliska@suse.cz>
11333 * common.opt: Remove extra line.
11335 2018-08-09 Martin Liska <mliska@suse.cz>
11337 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
11338 at the end of a line, make first letter capital and end up
11339 a sentence with a dot.
11340 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
11341 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
11342 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
11343 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
11344 (PARAM_MAX_ISL_OPERATIONS): Likewise.
11345 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
11346 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
11347 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
11348 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
11349 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
11350 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
11351 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
11352 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
11353 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
11354 (PARAM_TREE_REASSOC_WIDTH): Likewise.
11355 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
11356 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
11357 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
11359 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
11362 * config/s390/s390.c (s390_option_override_internal): Reduce the
11363 stack-clash-protection-probe-interval param if it would be too big
11366 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
11369 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
11370 * config/m68k/m68k.c (handle_move_double): Don't call
11371 m68k_final_prescan_insn.
11372 (m68k_adjust_decorated_operand): Renamed from
11373 m68k_final_prescan_insn, remove first and third operand and
11375 (print_operand): Call it.
11376 (print_operand_address): Call it.
11378 2018-08-08 Nathan Sidwell <nathan@acm.org>
11380 * diagnostic.c (diagnostic_report_current_module): Use
11381 linemap_included_from & linemap_included_from_linemap.
11383 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
11385 * config/aarch64/aarch64-cores.def: Add phecda core.
11386 * config/aarch64/aarch64-tune.md: Regenerate.
11387 * doc/invoke.texi: Add phecda core.
11389 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
11392 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
11394 * config/s390/s390.md ("movti"): Add more alternatives for
11395 constant to GPR copies.
11397 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
11399 * config/s390/s390.c: Fix whitespace damage throughout the file.
11400 * config/s390/s390.h: Likewise.
11401 * config/s390/tpf.h: Likewise.
11403 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
11405 * config/s390/s390.c (s390_loadrelative_operand_p):
11406 Remove TARGET_CPU_ZARCH usages.
11407 (s390_rtx_costs): Likewise.
11408 (s390_legitimate_constant_p): Likewise.
11409 (s390_cannot_force_const_mem): Likewise.
11410 (legitimate_reload_constant_p): Likewise.
11411 (s390_preferred_reload_class): Likewise.
11412 (legitimize_pic_address): Likewise.
11413 (legitimize_tls_address): Likewise.
11414 (s390_split_branches): Removed.
11415 (s390_add_execute): Removed.
11416 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
11417 (s390_mainpool_start): Likewise.
11418 (s390_mainpool_finish): Likewise.
11419 (s390_mainpool_cancel): Removed.
11420 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
11421 (s390_chunkify_cancel): Likewise.
11422 (s390_return_addr_rtx): Likewise.
11423 (s390_register_info): Remove split_branches_pending_p uages.
11424 (s390_optimize_register_info): Likewise.
11425 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
11426 split_branches_pending_p usages.
11427 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
11428 (s390_load_got): Likewise.
11429 (s390_expand_split_stack_prologue): Likewise.
11430 (output_asm_nops): Likewise.
11431 (s390_function_profiler): Likewise.
11432 (s390_emit_call): Likewise.
11433 (s390_conditional_register_usage): Likewise.
11434 (s390_optimize_prologue): Likewise.
11435 (s390_reorg): Remove TARGET_CPU_ZARCH and
11436 split_branches_pending_p usages.
11437 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
11439 (s390_output_indirect_thunk_function): Likewise.
11440 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
11441 (TARGET_CPU_ZARCH_P): Removed.
11442 (struct machine_function): Remove split_branches_pending_p.
11443 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
11445 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
11447 * common/config/s390/s390-common.c (processor_flags_table):
11449 * config.gcc: Remove with_arch/with_tune support.
11450 * config/s390/2064.md: Remove cpu attribute comparisons.
11451 * config/s390/driver-native.c (s390_host_detect_local_cpu):
11453 * config/s390/linux.h (ASM_SPEC):
11454 Remove -march support.
11455 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
11456 Use a table to get an arch level.
11457 * config/s390/s390-opts.h (enum processor_type):
11458 Remove enum values.
11459 * config/s390/s390.c
11460 (processor_table): Remove entries, add arch_level values.
11461 (s390_issue_rate): Remove cases.
11462 (s390_option_override): Adjust
11463 s390_option_override_internal() call.
11464 (s390_option_override_internal): Remove deprecation warning.
11465 (s390_valid_target_attribute_tree): Adjust
11466 s390_option_override_internal() call.
11467 * config/s390/s390.h (struct s390_processor):
11468 Share with s390-c.c, add arch_level field.
11469 * config/s390/s390.md:
11470 Remove occurrences in cpu attribute.
11471 * config/s390/s390.opt: Remove -march/-mtune support.
11472 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
11473 * doc/invoke.texi: Remove deprecation warning.
11475 2018-08-08 Luis Machado <luis.machado@linaro.org>
11477 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
11479 (qdf24xx_tunings): Set vector cost structure to
11480 qdf24xx_vector_cost.
11482 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
11483 <register_sextend>: Set to 3.
11485 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
11488 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
11489 * config/aarch64/aarch64-simd.md
11490 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
11491 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
11492 (aarch64_frecpx<mode>): ...this new pattern.
11493 * config/aarch64/aarch64-simd-builtins.def: Remove comment
11494 about aarch64_frecp<FRECP:frecp_suffix><mode>.
11496 2018-08-07 Martin Liska <mliska@suse.cz>
11498 PR middle-end/83023
11499 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
11500 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
11501 * predict.def (PRED_MALLOC_NONNULL): New predictor.
11502 * doc/extend.texi: Document that malloc attribute adds
11505 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
11508 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11509 Define to speculation_safe_value_not_needed.
11511 2018-08-06 Jeff Law <law@redhat.com>
11513 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
11514 the vr_values instance to cprop_into_stmt.
11515 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
11516 (cprop_operand): Also query EVRP to determine if OP is a constant.
11518 2018-08-06 Nathan Sidwell <nathan@acm.org>
11520 * diagnostic.c (diagnostic_report_current_module): Reroll
11521 included-at loop. Translate text.
11523 2018-08-06 David Malcolm <dmalcolm@redhat.com>
11525 * function-tests.c (selftest::test_expansion_to_rtl): Call
11526 free_after_compilation.
11528 2018-08-06 Alan Hayward <alan.hayward@arm.com>
11530 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
11532 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
11534 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
11535 loops with memory block operations from getting unrolled.
11537 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
11540 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11541 Define to speculation_safe_value_not_needed.
11543 2018-08-06 Jeff Law <law@redhat.com>
11545 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
11548 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
11551 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
11552 with all enabled __intN types.
11554 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
11556 2018-08-06 Alan Hayward <alan.hayward@arm.com>
11558 * alias.c (record_set): Check for clobber high.
11559 * cfgexpand.c (expand_gimple_stmt): Likewise.
11560 * combine-stack-adj.c (single_set_for_csa): Likewise.
11561 * combine.c (find_single_use_1): Likewise.
11562 (set_nonzero_bits_and_sign_copies): Likewise.
11563 (get_combine_src_dest): Likewise.
11564 (is_parallel_of_n_reg_sets): Likewise.
11565 (try_combine): Likewise.
11566 (record_dead_and_set_regs_1): Likewise.
11567 (reg_dead_at_p_1): Likewise.
11568 (reg_dead_at_p): Likewise.
11569 * dce.c (deletable_insn_p): Likewise.
11570 (mark_nonreg_stores_1): Likewise.
11571 (mark_nonreg_stores_2): Likewise.
11572 * df-scan.c (df_find_hard_reg_defs): Likewise.
11573 (df_uses_record): Likewise.
11574 (df_get_call_refs): Likewise.
11575 * dwarf2out.c (mem_loc_descriptor): Likewise.
11576 * haifa-sched.c (haifa_classify_rtx): Likewise.
11577 * ira-build.c (create_insn_allocnos): Likewise.
11578 * ira-costs.c (scan_one_insn): Likewise.
11579 * ira.c (equiv_init_movable_p): Likewise.
11580 (rtx_moveable_p): Likewise.
11581 (interesting_dest_for_shprep): Likewise.
11582 * jump.c (mark_jump_label_1): Likewise.
11583 * postreload-gcse.c (record_opr_changes): Likewise.
11584 * postreload.c (reload_cse_simplify): Likewise.
11585 (struct reg_use): Add source expr.
11586 (reload_combine): Check for clobber high.
11587 (reload_combine_note_use): Likewise.
11588 (reload_cse_move2add): Likewise.
11589 (move2add_note_store): Likewise.
11590 * print-rtl.c (print_pattern): Likewise.
11591 * recog.c (decode_asm_operands): Likewise.
11592 (store_data_bypass_p): Likewise.
11593 (if_test_bypass_p): Likewise.
11594 * regcprop.c (kill_clobbered_value): Likewise.
11595 (kill_set_value): Likewise.
11596 * reginfo.c (reg_scan_mark_refs): Likewise.
11597 * reload1.c (maybe_fix_stack_asms): Likewise.
11598 (eliminate_regs_1): Likewise.
11599 (elimination_effects): Likewise.
11600 (mark_not_eliminable): Likewise.
11601 (scan_paradoxical_subregs): Likewise.
11602 (forget_old_reloads_1): Likewise.
11603 * reorg.c (find_end_label): Likewise.
11604 (try_merge_delay_insns): Likewise.
11605 (redundant_insn): Likewise.
11606 (own_thread_p): Likewise.
11607 (fill_simple_delay_slots): Likewise.
11608 (fill_slots_from_thread): Likewise.
11609 (dbr_schedule): Likewise.
11610 * resource.c (update_live_status): Likewise.
11611 (mark_referenced_resources): Likewise.
11612 (mark_set_resources): Likewise.
11613 * rtl.c (copy_rtx): Likewise.
11614 * rtlanal.c (reg_referenced_p): Likewise.
11615 (single_set_2): Likewise.
11616 (noop_move_p): Likewise.
11617 (note_stores): Likewise.
11618 * sched-deps.c (sched_analyze_reg): Likewise.
11619 (sched_analyze_insn): Likewise.
11621 2018-08-06 Alan Hayward <alan.hayward@arm.com>
11623 * cse.c (invalidate_reg): New function extracted from...
11624 (invalidate): ...here.
11625 (canonicalize_insn): Check for clobber high.
11626 (invalidate_from_clobbers): invalidate clobber highs.
11627 (invalidate_from_sets_and_clobbers): Likewise.
11628 (count_reg_usage): Check for clobber high.
11629 (insn_live_p): Likewise.
11630 * cselib.c (cselib_expand_value_rtx_1):Likewise.
11631 (cselib_invalidate_regno): Check for clobber in setter.
11632 (cselib_invalidate_rtx): Pass through setter.
11633 (cselib_invalidate_rtx_note_stores):
11634 (cselib_process_insn): Check for clobber high.
11635 * cselib.h (cselib_invalidate_rtx): Add operand.
11637 2018-08-06 Alan Hayward <alan.hayward@arm.com>
11639 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
11640 (mark_not_eliminable): Likewise.
11641 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
11642 * lra-lives.c (process_bb_lives): Check for clobber high.
11643 * lra.c (new_insn_reg): Remember clobber highs.
11644 (collect_non_operand_hard_regs): Check for clobber high.
11645 (lra_set_insn_recog_data): Likewise.
11646 (add_regs_to_insn_regno_info): Likewise.
11647 (lra_update_insn_regno_info): Likewise.
11649 2018-08-06 Alan Hayward <alan.hayward@arm.com>
11651 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
11652 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
11654 2018-08-06 Alan Hayward <alan.hayward@arm.com>
11656 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
11657 (copy_insn_1): Likewise.
11658 (gen_hard_reg_clobber_high): New gen function.
11659 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
11660 * genemit.c (gen_exp): Likewise.
11661 (gen_emit_seq): Pass through info.
11662 (gen_insn): Check for CLOBBER_HIGH.
11663 (gen_expand): Pass through info.
11664 (gen_split): Likewise.
11665 (output_add_clobbers): Likewise.
11666 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
11667 (remove_clobbers): Likewise.
11668 * rtl.h (gen_hard_reg_clobber_high): New declaration.
11670 2018-08-06 Alan Hayward <alan.hayward@arm.com>
11672 * doc/rtl.texi (clobber_high): Add.
11673 (parallel): Add in clobber high
11674 * rtl.c (rtl_check_failed_code3): Add function.
11675 * rtl.def (CLOBBER_HIGH): Add expression.
11676 * rtl.h (RTL_CHECKC3): Add macro.
11677 (rtl_check_failed_code3): Add declaration.
11678 (XC3EXP): Add macro.
11680 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
11683 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
11684 cfun->machine->max_used_stack_alignment if needed.
11686 2018-08-04 Martin Sebor <msebor@redhat.com>
11688 PR tree-optimization/86571
11689 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
11692 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
11695 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11698 2018-08-03 Jeff Law <law@redhat.com>
11701 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11702 Define to speculation_safe_value_not_needed.
11704 2018-08-03 David Malcolm <dmalcolm@redhat.com>
11706 * doc/gcov.texi (-x): Remove duplicate "to".
11707 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
11708 (-Wif-not-aligned): Remove duplicate "is".
11709 (-flto): Remove duplicate "the".
11710 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
11711 duplicate "v5.00.b".
11712 (MSP430 Options): Remove duplicate "and" from the description
11713 of "-mgprel-sec=regexp".
11714 (x86 Options): Remove duplicate copies of "vmldLog102" and
11715 vmlsLog104 from description of "-mveclibabi=type".
11717 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
11719 * internal-fn.h (first_commutative_argument): Declare.
11720 * internal-fn.c (first_commutative_argument): New function.
11721 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
11722 restrictions for pattern statements. Use first_commutative_argument
11723 to look for commutative operands in calls to internal functions.
11725 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
11727 * Makefile.in (wide-int-range.o): New.
11728 * tree-vrp.c: Move all the wide_int_* functions to...
11729 * wide-int-range.cc: ...here.
11730 * tree-vrp.h: Move all the wide_int_* prototypes to...
11731 * wide-int-range.h: ...here.
11733 2018-08-03 Tom de Vries <tdevries@suse.de>
11735 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
11737 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
11738 * except.c (output_function_exception_table): Do early exit if
11739 targetm_common.except_unwind_info (&global_options) == UI_NONE.
11741 2018-08-03 Martin Liska <mliska@suse.cz>
11743 * predict.c (dump_prediction): Change to 2 digits
11746 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
11748 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
11750 (falkor_am_1_gtov_gtov): ... a new insn reservation.
11752 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
11754 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
11755 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
11756 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
11757 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
11758 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
11760 2018-08-02 David Malcolm <dmalcolm@redhat.com>
11762 * diagnostic-show-locus.c (diagnostic_show_locus): Use
11763 pp_take_prefix when saving the existing prefix.
11764 * diagnostic.c (diagnostic_append_note): Likewise.
11765 * langhooks.c (lhd_print_error_function): Likewise.
11766 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
11767 param's type. Free the existing prefix.
11768 (pp_take_prefix): New function.
11769 (pretty_printer::pretty_printer): Drop the prefix parameter.
11770 Rename the length parameter to match the comment.
11771 (pretty_printer::~pretty_printer): Free the prefix.
11772 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
11774 (struct pretty_printer): Drop the "const" from "prefix" field's
11775 type and clarify memory management.
11776 (pp_set_prefix): Drop the "const" from the 2nd param.
11777 (pp_take_prefix): New decl.
11779 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
11781 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
11782 for word_mode_ok here instead of passing as argument.
11783 (expand_block_compare): Change select_block_compare_mode() call.
11784 (expand_strncmp_gpr_sequence): New function.
11785 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
11787 2018-08-02 Jeff Law <law@redhat.com>
11790 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11791 Define to speculation_safe_value_not_needed.
11794 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11795 Define to speculation_safe_value_not_needed.
11797 2018-08-02 Tom de Vries <tdevries@suse.de>
11800 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
11801 function. Return UI_TARGET unconditionally.
11802 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
11803 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
11805 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
11807 * genemit.c (print_overload_test): Fix typo.
11809 2018-08-02 Richard Biener <rguenther@suse.de>
11811 PR tree-optimization/86816
11812 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
11813 which checks for value availability before querying it.
11814 (gvn_uses_equal): Use it.
11815 (same_succ_hash): Likewise.
11816 (gimple_equal_p): Likewise.
11818 2018-08-02 Nick Clifton <nickc@redhat.com>
11821 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11822 Define to speculation_safe_value_not_needed.
11825 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11826 Define to speculation_safe_value_not_needed.
11829 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11830 Define to speculation_safe_value_not_needed.
11833 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11834 Define to speculation_safe_value_not_needed.
11837 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11838 Define to speculation_safe_value_not_needed.
11841 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11842 Define to speculation_safe_value_not_needed.
11845 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11846 Define to speculation_safe_value_not_needed.
11849 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11850 Define to speculation_safe_value_not_needed.
11853 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
11854 speculation_safe_value_not_needed.
11857 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
11858 to speculation_safe_value_not_needed.
11860 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
11862 * doc/md.texi: Expand the documentation of instruction names
11863 to mention port-local uses. Document '@' in pattern names.
11864 * read-md.h (overloaded_instance, overloaded_name): New structs.
11865 (mapping): Declare.
11866 (md_reader::handle_overloaded_name): New member function.
11867 (md_reader::get_overloads): Likewise.
11868 (md_reader::m_first_overload): New member variable.
11869 (md_reader::m_next_overload_ptr): Likewise.
11870 (md_reader::m_overloads_htab): Likewise.
11871 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
11872 m_next_overload_ptr and m_overloads_htab.
11873 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
11874 (get_mode_token, get_code_token, get_int_token): New functions.
11875 (map_attr_string): Add an optional argument that passes back
11876 the associated iterator.
11877 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
11878 (md_reader::handle_overloaded_name, add_overload_instance): New
11880 (apply_iterators): Handle '@' names. Report an error if '@'
11881 is used without iterators.
11882 (initialize_iterators): Initialize the new iterator_group fields.
11883 * genopinit.c (handle_overloaded_code_for)
11884 (handle_overloaded_gen): New functions.
11885 (main): Use them to print declarations of maybe_code_for_* and
11886 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
11887 * genemit.c (print_overload_arguments, print_overload_test)
11888 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
11889 (main): Use it to print definitions of maybe_code_for_* and
11890 maybe_gen_* functions.
11891 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
11892 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
11893 instead of explicit mode checks.
11894 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
11895 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
11896 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
11897 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
11898 (aarch64_expand_compare_and_swap): Likewise
11899 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
11900 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
11901 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
11902 (aarch64_constant_pool_reload_icode): Delete.
11903 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
11904 instead of aarch64_constant_pool_reload_icode. Use
11905 code_for_aarch64_reload_mov instead of explicit mode checks.
11906 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
11907 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
11908 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
11909 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
11910 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
11911 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
11912 (aarch64_atomic_load_op_code): Delete.
11913 (aarch64_emit_atomic_load_op): Likewise.
11914 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
11915 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
11916 instead of aarch64_emit_atomic_load_op.
11917 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
11918 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
11919 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
11920 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
11921 character before the pattern name.
11922 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
11923 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
11924 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
11925 (aarch64_frecps<mode>): Likewise.
11926 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
11927 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
11928 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
11929 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
11930 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
11932 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
11934 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
11935 Allow HFmode constants if TARGET_FP_F16INST.
11937 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
11940 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
11941 No longer check last store for clobber of address register.
11943 2018-08-02 Martin Liska <mliska@suse.cz>
11945 PR gcov-profile/86817
11946 * gcov.c (process_all_functions): New function.
11948 (process_file): Move functions processing to
11949 process_all_functions.
11951 2018-08-02 David Malcolm <dmalcolm@redhat.com>
11953 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
11954 "const" to the "gimple *" and "rtx_insn *" parameters.
11955 * dumpfile.h (dump_user_location_t::dump_user_location_t):
11957 (dump_location_t::dump_location_t): Likewise.
11959 2018-08-01 Martin Sebor <msebor@redhat.com>
11961 PR tree-optimization/86650
11962 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
11963 rather than a "gcall *". Directly pass the data of interest
11964 to percent_K_format, rather than building a temporary CALL_EXPR
11966 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
11967 (gimple_fold_builtin_strncat): Adjust.
11968 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
11969 gcall* argument with gimple*.
11970 * gimple-ssa-warn-restrict.c (check_call): Same.
11971 (wrestrict_dom_walker::before_dom_children): Same.
11972 (builtin_access::builtin_access): Same.
11973 (check_bounds_or_overlap): Same
11974 (maybe_diag_overlap): Same.
11975 (maybe_diag_offset_bounds): Same.
11976 * tree-diagnostic.c (default_tree_printer): Move usage of
11977 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
11979 * tree-pretty-print.c (percent_K_format): Add argument.
11980 * tree-pretty-print.h: Add argument.
11981 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
11982 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
11983 (maybe_diag_stxncpy_trunc): Same.
11984 (handle_builtin_stxncpy): Same.
11985 (handle_builtin_strcat): Same.
11987 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
11989 * match.pd: Optimise pointer range checks.
11991 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
11993 PR tree-optimization/86758
11994 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
11995 to remove pattern statements.
11997 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
11999 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
12000 result of dfs_enumerate_from when constructing stmt_vec_infos,
12001 instead of additionally calling get_loop_body.
12003 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12005 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
12007 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
12008 When creating an iv, assert that the step is not known to be zero.
12009 (vect_setup_realignment): Update call accordingly.
12010 * tree-vect-stmts.c (vectorizable_store): Likewise.
12011 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
12013 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12015 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
12016 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
12017 (vectorizable_reduction): Likewise.
12018 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
12019 (vect_detect_hybrid_slp_stmts): Likewise.
12020 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
12022 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
12024 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
12025 (wide_int_set_zero_nonzero_bits): ...this.
12026 (zero_nonzero_bits_from_vr): Rename to...
12027 (vrp_set_zero_nonzero_bits): ...this.
12028 (extract_range_from_multiplicative_op_1): Abstract wide int
12030 (wide_int_range_multiplicative_op): ...here.
12031 (extract_range_from_binary_expr_1): Extract wide int binary
12032 operations into their own functions.
12033 (wide_int_range_lshift): New.
12034 (wide_int_range_can_optimize_bit_op): New.
12035 (wide_int_range_shift_undefined_p): New.
12036 (wide_int_range_bit_xor): New.
12037 (wide_int_range_bit_ior): New.
12038 (wide_int_range_bit_and): New.
12039 (wide_int_range_trunc_mod): New.
12040 (extract_range_into_wide_ints): New.
12041 (vrp_shift_undefined_p): New.
12042 (extract_range_from_multiplicative_op): New.
12043 (vrp_can_optimize_bit_op): New.
12044 * tree-vrp.h (value_range::dump): New.
12045 (wide_int_range_multiplicative_op): New.
12046 (wide_int_range_lshift):New.
12047 (wide_int_range_shift_undefined_p): New.
12048 (wide_int_range_bit_xor): New.
12049 (wide_int_range_bit_ior): New.
12050 (wide_int_range_bit_and): New.
12051 (wide_int_range_trunc_mod): New.
12052 (zero_nonzero_bits_from_bounds): Rename to...
12053 (wide_int_set_zero_nonzero_bits): ...this.
12054 (zero_nonzero_bits_from_vr): Rename to...
12055 (vrp_set_zero_nonzero_bits): ...this.
12056 (range_easy_mask_min_max): Rename to...
12057 (wide_int_range_can_optimize_bit_op): this.
12058 * vr-values.c (simplify_bit_ops_using_ranges): Rename
12059 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
12061 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12063 * tree-vectorizer.h (vect_orig_stmt): New function.
12064 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
12065 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
12066 (vect_create_epilog_for_reduction): Likewise.
12067 (vectorizable_live_operation): Likewise.
12068 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
12069 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
12070 * tree-vect-stmts.c (vectorizable_call): Likewise.
12071 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
12073 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12075 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
12077 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
12078 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
12079 (vect_transform_loop): Likewise.
12080 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
12082 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12084 * tree-vectorizer.h (vect_schedule_slp): Return void.
12085 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
12086 (vect_schedule_slp): Likewise.
12088 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12090 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
12092 (vect_transform_loop): Update calls accordingly. Schedule SLP
12093 instances before the main loop, if any exist.
12095 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12097 PR tree-optimization/86749
12098 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
12099 If the lhs is used in a COND_EXPR, check that it is being used
12100 as the "then" or "else" value.
12102 2018-08-01 Tom de Vries <tdevries@suse.de>
12105 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
12106 speculation_safe_value_not_needed.
12108 2018-08-01 Richard Biener <rguenther@suse.de>
12110 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
12111 as base and offset.
12113 2018-08-01 Martin Liska <mliska@suse.cz>
12115 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
12116 format how successful transformation is dumped.
12117 (gimple_mod_pow2_value_transform): Likewise.
12118 (gimple_mod_subtract_transform): Likewise.
12119 (gimple_stringops_transform): Likewise.
12121 2018-08-01 Martin Liska <mliska@suse.cz>
12123 PR value-prof/35543
12124 * value-prof.c (interesting_stringop_to_profile_p):
12125 Simplify the code and add BUILT_IN_MEMMOVE.
12126 (gimple_stringops_transform): Likewise.
12128 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
12130 * config/aarch64/aarch64-simd.md
12131 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
12132 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
12133 use GPI iterator instead of SI mode.
12135 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
12137 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
12138 rs6000_speculation_barrier.
12139 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
12140 new barrier pattern name.
12142 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
12144 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
12145 (speculation_barrier): New insn.
12147 2018-08-01 Richard Biener <rguenther@suse.de>
12150 * graphite.h: Include isl/id.h and isl/space.h to allow build
12153 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
12156 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
12157 mode for COFF targets.
12158 * defaults.h (TARGET_COFF): Define.
12159 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
12160 TARGET_COFF): Define.
12161 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
12162 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
12163 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
12165 2018-07-31 Alexandre Oliva <oliva@adacore.com>
12167 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
12168 * gimple-streamer-out.c (output_bb): Save it.
12169 * lto-streamer-in.c (input_struct_function_base): Restore
12170 instance discriminator if available. Create map on demand.
12171 * lto-streamer-out.c (output_struct_function_base): Save it if
12173 * final.c (decl_to_instance_map): Document LTO strategy.
12175 2018-07-31 Alexandre Oliva <oliva@adacore.com>
12176 Olivier Hainque <hainque@adacore.com>
12178 * debug.h (decl_to_instance_map_t): New type.
12179 (decl_to_instance_map): Declare.
12180 (maybe_create_decl_to_instance_map): New inline function.
12181 * final.c (bb_discriminator, last_bb_discriminator): New statics,
12182 to track basic block discriminators.
12183 (final_start_function_1): Initialize them.
12184 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
12186 (decl_to_instance_map): New variable.
12187 (map_decl_to_instance, maybe_set_discriminator): New functions.
12188 (notice_source_line): Set discriminator.
12190 2018-07-31 Ian Lance Taylor <iant@golang.org>
12192 * targhooks.c (default_have_speculation_safe_value): Add
12195 2018-07-31 David Malcolm <dmalcolm@redhat.com>
12197 * dump-context.h: Include "pretty-print.h".
12198 (dump_context::refresh_dumps_are_enabled): New decl.
12199 (dump_context::emit_item): New decl.
12200 (class dump_context): Add fields "m_test_pp" and
12202 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
12203 (temp_dump_context::get_dumped_text): New decl.
12204 (class temp_dump_context): Add field "m_pp".
12205 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
12206 (dump_context::refresh_dumps_are_enabled): ...and add a test for
12208 (set_dump_file): Update for above change.
12209 (set_alt_dump_file): Likewise.
12210 (dump_loc): New overload, taking a pretty_printer *.
12211 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
12212 to any test pretty-printer.
12213 (make_item_for_dump_gimple_stmt): New function, adapted from
12214 optinfo::add_gimple_stmt.
12215 (dump_context::dump_gimple_stmt): Call it, and use the result,
12216 eliminating the direct usage of dump_file and alt_dump_file in
12217 favor of indirectly using them via emit_item.
12218 (make_item_for_dump_gimple_expr): New function, adapted from
12219 optinfo::add_gimple_expr.
12220 (dump_context::dump_gimple_expr): Call it, and use the result,
12221 eliminating the direct usage of dump_file and alt_dump_file in
12222 favor of indirectly using them via emit_item.
12223 (make_item_for_dump_generic_expr): New function, adapted from
12225 (dump_context::dump_generic_expr): Call it, and use the result,
12226 eliminating the direct usage of dump_file and alt_dump_file in
12227 favor of indirectly using them via emit_item.
12228 (make_item_for_dump_printf_va): New function, adapted from
12229 optinfo::add_printf_va.
12230 (make_item_for_dump_printf): New function.
12231 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
12232 and use the result, eliminating the direct usage of dump_file and
12233 alt_dump_file in favor of indirectly using them via emit_item.
12234 (make_item_for_dump_dec): New function.
12235 (dump_context::dump_dec): Call it, and use the result,
12236 eliminating the direct usage of dump_file and alt_dump_file in
12237 favor of indirectly using them via emit_item.
12238 (make_item_for_dump_symtab_node): New function, adapted from
12239 optinfo::add_symtab_node.
12240 (dump_context::dump_symtab_node): Call it, and use the result,
12241 eliminating the direct usage of dump_file and alt_dump_file in
12242 favor of indirectly using them via emit_item.
12243 (dump_context::begin_scope): Reimplement, avoiding direct usage
12244 of dump_file and alt_dump_file in favor of indirectly using them
12246 (dump_context::emit_item): New member function.
12247 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
12248 Set up test pretty-printer on the underlying context. Call
12249 refresh_dumps_are_enabled.
12250 (temp_dump_context::~temp_dump_context): Call
12251 refresh_dumps_are_enabled.
12252 (temp_dump_context::get_dumped_text): New member function.
12253 (selftest::verify_dumped_text): New function.
12254 (ASSERT_DUMPED_TEXT_EQ): New macro.
12255 (selftest::test_capture_of_dump_calls): Run all tests twice, with
12256 and then without optinfo enabled. Add uses of
12257 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
12258 * dumpfile.h: Update comment for the dump_* API.
12259 * optinfo-emit-json.cc
12260 (selftest::test_building_json_from_dump_calls): Update for new
12261 param for temp_dump_context ctor.
12262 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
12263 and "m_owned" field.
12264 (optinfo_item::~optinfo_item): Likewise.
12265 (optinfo::add_item): New member function.
12266 (optinfo::emit): Update comment.
12267 (optinfo::add_string): Delete.
12268 (optinfo::add_printf): Delete.
12269 (optinfo::add_printf_va): Delete.
12270 (optinfo::add_gimple_stmt): Delete.
12271 (optinfo::add_gimple_expr): Delete.
12272 (optinfo::add_tree): Delete.
12273 (optinfo::add_symtab_node): Delete.
12274 (optinfo::add_dec): Delete.
12275 * optinfo.h (class dump_context): New forward decl.
12276 (optinfo::add_item): New decl.
12277 (optinfo::add_string): Delete.
12278 (optinfo::add_printf): Delete.
12279 (optinfo::add_printf_va): Delete.
12280 (optinfo::add_gimple_stmt): Delete.
12281 (optinfo::add_gimple_expr): Delete.
12282 (optinfo::add_tree): Delete.
12283 (optinfo::add_symtab_node): Delete.
12284 (optinfo::add_dec): Delete.
12285 (optinfo::add_poly_int): Delete.
12286 (optinfo_item::optinfo_item): Remove "owned" param.
12287 (class optinfo_item): Remove field "m_owned".
12289 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
12291 PR middle-end/86705
12292 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
12293 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
12294 requested variable alignment.
12295 (expand_one_ssa_partition): Likewise.
12296 (expand_one_var): Likewise.
12298 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12300 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
12301 to speculation_safe_value_not_needed.
12303 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12305 * targhooks.h (speculation_safe_value_not_needed): New prototype.
12306 * targhooks.c (speculation_safe_value_not_needed): New function.
12307 * target.def (have_speculation_safe_value): Update documentation.
12308 * doc/tm.texi: Regenerated.
12310 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12312 * config/aarch64/iterators.md (ALLI_TI): New iterator.
12313 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
12315 (despeculate_copy<ALLI:mode>_insn): New insn.
12316 (despeculate_copyti_insn): New insn.
12317 (despeculate_simple<ALLI:mode>): New insn
12318 (despeculate_simpleti): New insn.
12319 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
12321 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
12322 aarch64_speculation_safe_value.
12323 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
12325 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12327 * config/aarch64/aarch64-speculation.cc: New file.
12328 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
12329 before pass_reorder_blocks.
12330 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
12332 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
12333 X14 and X15 when tracking speculation.
12334 * config/aarch64/aarch64.md (register name constants): Add
12335 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
12336 (unspec): Add UNSPEC_SPECULATION_TRACKER.
12337 (speculation_barrier): New insn attribute.
12338 (cmp<mode>): Allow SP in comparisons.
12339 (speculation_tracker): New insn.
12340 (speculation_barrier): Add speculation_barrier attribute.
12341 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
12342 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
12343 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
12345 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12347 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
12348 aarch64_track_speculation is true.
12349 (tb<optab><mode>1): Likewise.
12350 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
12351 generate CB[N]Z when tracking speculation.
12352 (aarch64_split_compare_and_swap): Likewise.
12353 (aarch64_split_atomic_op): Likewise.
12355 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12357 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
12359 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12361 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
12362 (speculation_barrier): New insn.
12364 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12366 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
12367 * config/arm/arm.md (speculation_barrier): New expand.
12368 (speculation_barrier_insn): New pattern.
12370 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12372 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
12373 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
12374 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
12375 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
12377 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
12378 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
12379 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
12380 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
12381 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
12382 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
12383 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
12384 * builtins.c (expand_speculation_safe_value): New function.
12385 (expand_builtin): Call it.
12386 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
12387 * doc/extend.texi: Document __builtin_speculation_safe_value.
12388 * doc/md.texi: Document "speculation_barrier" pattern.
12389 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
12390 TARGET_HAVE_SPECULATION_SAFE_VALUE.
12391 * doc/tm.texi: Regenerated.
12392 * target.def (have_speculation_safe_value, speculation_safe_value): New
12394 * targhooks.c (default_have_speculation_safe_value): New function.
12395 (default_speculation_safe_value): New function.
12396 * targhooks.h (default_have_speculation_safe_value): Add prototype.
12397 (default_speculation_safe_value): Add prototype.
12399 2018-07-31 David Malcolm <dmalcolm@redhat.com>
12401 * dump-context.h (dump_context::dump_loc): New decl.
12402 * dumpfile.c (dump_context::dump_loc): New member function.
12403 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
12404 and dump_gimple_stmt.
12405 (dump_context::dump_gimple_expr_loc): Likewise, using
12407 (dump_context::dump_generic_expr_loc): Likewise, using
12409 (dump_context::dump_printf_loc_va): Likewise, using
12411 (dump_context::begin_scope): Explicitly using the global function
12412 "dump_loc", rather than the member function.
12414 2018-07-31 Martin Sebor <msebor@redhat.com>
12416 PR tree-optimization/86741
12417 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
12419 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
12421 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
12422 depend on whether prefetch instructions will be emitted or not.
12423 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
12424 will be emitted or not.
12425 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
12426 (TARGET_SETMEM_PFD): New macros.
12428 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12430 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
12431 (NULL_STMT_VEC_INFO): Delete.
12432 (stmt_vec_info::operator*): Likewise.
12433 (stmt_vec_info::operator gimple *): Likewise.
12434 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
12435 of NULL_STMT_VEC_INFO.
12436 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
12437 (vect_reassociating_reduction_p): Likewise.
12438 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
12439 (vectorizable_store): Likewise.
12440 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
12441 (vec_info::free_stmt_vec_infos): Likewise.
12443 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12445 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
12446 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
12448 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12450 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
12451 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
12452 (vec_info::free_stmt_vec_info): New private member functions.
12453 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
12454 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
12455 * tree-parloops.c (gather_scalar_reductions): Remove calls to
12456 set_stmt_vec_info_vec and free_stmt_vec_infos.
12457 * tree-vect-loop.c (_loop_vec_info): Remove call to
12458 set_stmt_vec_info_vec.
12459 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
12460 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
12461 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
12462 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
12463 (vec_info::free_stmt_vec_info): ...these new functions. Remove
12464 assignments in {vec_info::,}new_stmt_vec_info that are redundant
12465 with the clearing in the xcalloc.
12466 (stmt_vec_info_vec): Delete.
12467 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
12468 (vectorize_loops): Likewise.
12469 (vec_info::~vec_info): Remove argument from call to
12470 free_stmt_vec_infos.
12471 (vec_info::add_stmt): Remove vinfo argument from call to
12474 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12476 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
12477 rather than a gimple stmt.
12478 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
12479 information for pattern statements when passed the original
12480 statement; instead wait to be passed the pattern statement itself.
12481 Don't call set_vinfo_for_stmt here.
12482 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
12483 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
12484 stmt_vec_infos here.
12485 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
12486 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
12487 stmt_vec_infos entry.
12489 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12491 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
12492 * tree-vectorizer.c (vec_info::replace_stmt): New function.
12493 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
12494 * tree-vect-stmts.c (vectorizable_call): Likewise.
12495 (vectorizable_simd_clone_call): Likewise.
12497 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12499 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
12500 * tree-vectorizer.c (vec_info::remove_stmt): New function.
12501 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
12502 * tree-vect-loop.c (vect_transform_loop): Likewise.
12503 * tree-vect-slp.c (vect_schedule_slp): Likewise.
12504 * tree-vect-stmts.c (vect_remove_stores): Likewise.
12506 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12508 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
12509 (vect_dr_stmt): Delete.
12510 * tree-vectorizer.c (vec_info::lookup_dr): New function.
12511 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
12513 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
12514 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
12515 (vect_verify_datarefs_alignment, vect_peeling_supportable)
12516 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
12517 (vect_analyze_data_refs): Likewise.
12518 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
12520 (vect_find_same_alignment_drs): Likewise.
12521 (vect_slp_analyze_node_dependences): Update calls accordingly.
12522 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
12523 instead of DR_VECT_AUX.
12524 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
12525 of a vector data references. Use vec_info::lookup_dr instead of
12527 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
12528 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
12529 instead of DR_VECT_AUX.
12531 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12533 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
12535 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
12537 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
12538 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
12539 (vect_gen_prolog_loop_niters): Likewise.
12541 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12543 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
12544 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
12545 (vect_known_alignment_in_bytes, vect_dr_behavior)
12546 (vect_get_scalar_dr_size): Take references as dr_vec_infos
12547 instead of data_references. Update calls to other routines for
12548 which the same change has been made.
12549 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
12550 dr_vec_infos instead of stmt_vec_infos.
12551 (vect_analyze_data_ref_dependence): Update call accordingly.
12552 (vect_slp_analyze_data_ref_dependence)
12553 (vect_record_base_alignments): Use DR_VECT_AUX.
12554 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
12555 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
12556 (vector_alignment_reachable_p, vect_get_data_access_cost)
12557 (vect_peeling_supportable, vect_analyze_group_access_1)
12558 (vect_analyze_group_access, vect_analyze_data_ref_access)
12559 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
12560 (vect_compile_time_alias, vect_small_gap_p)
12561 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
12562 (vect_supportable_dr_alignment): Take references as dr_vec_infos
12563 instead of data_references. Update calls to other routines for
12564 which the same change has been made.
12565 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
12566 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
12567 (vect_slp_analyze_and_verify_node_alignment)
12568 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
12569 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
12570 (vect_setup_realignment): Use dr_vec_infos. Update calls after
12572 (_vect_peel_info::dr): Replace with...
12573 (_vect_peel_info::dr_info): ...this new field.
12574 (vect_peeling_hash_get_most_frequent)
12575 (vect_peeling_hash_choose_best_peeling): Update accordingly.
12576 (vect_peeling_hash_get_lowest_cost):
12577 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
12578 routines for which the same change has been made.
12579 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
12581 * tree-vect-loop-manip.c (get_misalign_in_elems)
12582 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
12584 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
12585 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
12586 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
12587 (get_group_load_store_type, get_negative_load_store_type)
12588 (vect_get_data_ptr_increment, vectorizable_store)
12589 (vectorizable_load): Likewise.
12590 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
12591 Update calls to other routines for which the same change has been made.
12593 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12595 * tree-vectorizer.h (vec_info::move_dr): New member function.
12596 (dataref_aux): Rename to...
12597 (dr_vec_info): ...this and add "dr" and "stmt" fields.
12598 (_stmt_vec_info::dr_aux): Update accordingly.
12599 (_stmt_vec_info::data_ref_info): Delete.
12600 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
12601 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
12602 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
12603 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
12605 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
12606 (STMT_VINFO_DR_INFO): New macro.
12607 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
12608 (set_dr_misalignment): Update after rename of dataref_aux.
12609 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
12610 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
12611 initialization of STMT_VINFO_DATA_REF.
12612 * tree-vectorizer.c (vec_info::move_dr): New function.
12613 * tree-vect-patterns.c (vect_recog_bool_pattern)
12614 (vect_recog_mask_conversion_pattern)
12615 (vect_recog_gather_scatter_pattern): Use it.
12616 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
12617 the "dr" and "stmt" fields of dr_vec_info instead of
12618 STMT_VINFO_DATA_REF.
12620 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12622 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
12623 (is_pattern_stmt_p): Use it.
12624 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
12625 on pattern statements.
12627 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12629 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
12630 original stmt as a stmt_vec_info rather than a gimple stmt.
12631 (vect_pattern_recog_1): Take the statement directly as a
12632 stmt_vec_info, rather than via a gimple_stmt_iterator.
12633 Update call to vect_mark_pattern_stmts.
12634 (vect_pattern_recog): Update calls accordingly.
12636 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12638 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
12639 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
12640 a vect_def_type for the first argument.
12641 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
12642 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
12643 operand if it isn't defined by a vectorized statement.
12644 (vect_build_gather_load_calls): Remove the mask_dt argument and
12645 update calls to vect_get_vec_def_for_stmt_copy.
12646 (vectorizable_bswap): Likewise the dt argument.
12647 (vectorizable_call): Update calls to vectorizable_bswap and
12648 vect_get_vec_def_for_stmt_copy.
12649 (vectorizable_simd_clone_call, vectorizable_assignment)
12650 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
12651 (vectorizable_comparison): Update calls to
12652 vect_get_vec_def_for_stmt_copy.
12653 (vectorizable_store): Likewise. Remove now-unnecessary calls to
12654 vect_is_simple_use.
12655 (vect_get_loop_based_defs): Remove dt argument and update call
12656 to vect_get_vec_def_for_stmt_copy.
12657 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
12658 and vect_get_vec_def_for_stmt_copy.
12659 (vectorizable_load): Update calls to vect_build_gather_load_calls
12660 and vect_get_vec_def_for_stmt_copy.
12661 * tree-vect-loop.c (vect_create_epilog_for_reduction)
12662 (vectorizable_reduction, vectorizable_live_operation): Update calls
12663 to vect_get_vec_def_for_stmt_copy.
12665 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12667 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
12668 and gimple stmt arguments with a stmt_vec_info.
12669 (vect_record_base_alignments): Update calls accordingly.
12670 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
12671 and gimple stmt arguments with a stmt_vec_info.
12672 (vect_build_slp_tree_1): Remove vinfo argument and update call
12673 to vect_record_max_nunits.
12674 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
12675 and vect_record_max_nunits.
12677 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12679 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
12680 file and take a stmt_vec_info instead of a gimple stmt.
12681 (supportable_widening_operation, vect_finish_replace_stmt)
12682 (vect_finish_stmt_generation, vect_get_store_rhs)
12683 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
12684 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
12685 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
12686 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
12687 (vect_create_data_ref_ptr, bump_vector_ptr)
12688 (vect_permute_store_chain, vect_setup_realignment)
12689 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
12690 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
12691 (vectorizable_reduction, vectorizable_induction)
12692 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
12693 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
12694 than gimple stmts as arguments.
12695 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
12696 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
12697 (can_group_stmts_p, vect_check_gather_scatter)
12698 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
12699 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
12700 (vect_permute_load_chain, vect_shift_permute_load_chain)
12701 (vect_transform_grouped_load)
12702 (vect_record_grouped_load_vectors): Likewise.
12703 * tree-vect-loop.c (vect_fixup_reduc_chain)
12704 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
12705 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
12706 (vectorizable_reduction, vectorizable_induction)
12707 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
12708 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
12709 (vect_get_load_store_mask): Likewise.
12710 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
12711 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
12712 * tree-vect-stmts.c (vect_mark_relevant)
12713 (is_simple_and_all_uses_invariant)
12714 (exist_non_indexing_operands_for_use_p, process_use)
12715 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
12716 (vect_get_vec_def_for_operand, vect_get_vec_defs)
12717 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
12718 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
12719 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
12720 (get_negative_load_store_type, get_load_store_type)
12721 (vect_check_load_store_mask, vect_check_store_rhs)
12722 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
12723 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
12724 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
12725 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
12726 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
12727 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
12728 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
12729 (supportable_widening_operation): Likewise.
12731 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12733 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
12734 a stmt_vec_info instead of a gcall.
12735 (vect_check_gather_scatter): Update call accordingly.
12736 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
12738 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
12739 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
12740 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
12741 instead of a gimple stmt.
12742 (vect_transform_loop): Update calls accordingly.
12743 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
12744 stmt_vec_infos instead of gimple stmts.
12745 (vect_analyze_slp_instance): Update use accordingly.
12746 * tree-vect-stmts.c (read_vector_array, write_vector_array)
12747 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
12748 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
12749 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
12750 (vect_gen_widened_results_half, vect_get_loop_based_defs)
12751 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
12752 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
12753 down to subroutines.
12755 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12757 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
12758 of the worklist from a vector of gimple stmts to a vector of
12760 * tree-vect-stmts.c (vect_mark_relevant, process_use)
12761 (vect_mark_stmts_to_be_vectorized): Likewise
12763 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12765 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
12766 statement before passing it to vect_analyze_stmt.
12767 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
12768 the chain of phi vector definitions. Track the exit phi via its
12770 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
12771 STMT_VINFO_REDUC_DEF.
12772 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
12773 stmt_vec_infos to handle the statement chains.
12774 (vect_get_slp_defs): Record the first statement in the node
12775 using a stmt_vec_info.
12776 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
12777 statements here and pass their stmt_vec_info down to subroutines.
12778 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
12779 down to vect_finish_stmt_generation.
12780 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
12781 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
12782 stmt_vec_infos to subroutines.
12783 (vect_remove_stores): Use stmt_vec_infos to handle the statement
12786 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12788 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
12789 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
12790 (vect_permute_store_chain, vect_setup_realignment)
12791 (vect_permute_load_chain, vect_shift_permute_load_chain)
12792 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
12793 stmts internally, and when passing values to other vectorizer routines.
12794 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
12795 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
12796 (vect_analyze_loop_operations, get_initial_def_for_reduction)
12797 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
12798 (vectorizable_reduction, vectorizable_induction)
12799 (vectorizable_live_operation, vect_transform_loop_stmt)
12800 (vect_transform_loop): Likewise.
12801 * tree-vect-patterns.c (vect_reassociating_reduction_p)
12802 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
12803 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
12804 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
12805 (vect_slp_analyze_node_operations_1): Likewise.
12806 * tree-vect-stmts.c (vect_mark_relevant, process_use)
12807 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
12808 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
12809 (vect_finish_stmt_generation_1, get_group_load_store_type)
12810 (get_load_store_type, vect_build_gather_load_calls)
12811 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
12812 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
12813 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
12814 (vectorizable_store, vectorizable_load, vectorizable_condition)
12815 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
12816 (supportable_widening_operation): Likewise.
12817 (vect_get_vector_types_for_stmt): Likewise.
12818 * tree-vectorizer.h (vect_dr_behavior): Likewise.
12820 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12822 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
12823 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
12824 (vect_permute_store_chain, vect_permute_load_chain)
12825 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
12826 repeated stmt_vec_info lookups.
12827 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
12828 (vect_update_ivs_after_vectorizer): Likewise.
12829 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
12830 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
12831 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
12832 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
12833 (vect_bb_slp_scalar_cost): Likewise.
12834 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
12836 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12838 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
12839 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
12840 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
12841 to get gassigns and gcalls, rather than operating on generc gimple
12843 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
12844 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
12845 (vectorizable_load, vect_analyze_stmt): Likewise.
12846 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
12848 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12850 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
12851 return stmt_vec_infos rather than gimple stmts. Do not accept
12853 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
12855 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
12856 Update use of get_later_stmt.
12857 (vect_get_constant_vectors): Update call accordingly.
12858 (vect_schedule_slp_instance): Likewise
12859 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
12860 (vect_slp_analyze_instance_dependence): Likewise.
12861 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
12863 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12865 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
12866 (stmt_info_for_cost::stmt_info): ...this new field.
12867 (add_stmt_costs): Update accordingly.
12868 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
12869 (vect_get_known_peeling_cost): Likewise.
12870 (vect_estimate_min_profitable_iters): Likewise.
12871 * tree-vect-stmts.c (record_stmt_cost): Likewise.
12873 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12875 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
12876 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
12877 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
12879 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
12881 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12883 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
12884 a gimple stmt to a stmt_vec_info.
12885 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
12887 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12889 * tree-vectorizer.h (vec_info::grouped_stores): Change from
12890 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
12891 (_loop_vec_info::reduction_chains): Likewise.
12892 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
12894 * tree-vect-slp.c (vect_analyze_slp): Likewise.
12896 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12898 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
12899 a gimple stmt to a stmt_vec_info.
12900 (_stmt_vec_info::next_element): Likewise.
12901 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
12902 (vect_slp_analyze_and_verify_node_alignment)
12903 (vect_analyze_group_access_1, vect_analyze_group_access)
12904 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
12905 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
12906 (vect_supportable_dr_alignment): Update accordingly.
12907 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
12908 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
12909 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
12910 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
12911 * tree-vect-slp.c (vect_build_slp_tree_1)
12912 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
12913 (vect_split_slp_store_group, vect_analyze_slp_instance)
12914 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
12915 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
12916 (get_group_load_store_type, get_load_store_type)
12917 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
12918 (vect_transform_stmt, vect_remove_stores): Likewise.
12920 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12922 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
12923 than a gimple stmt.
12924 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
12925 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
12926 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
12927 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
12928 (vector_alignment_reachable_p, vect_get_data_access_cost)
12929 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
12930 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
12931 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
12932 (vect_analyze_group_access_1, vect_analyze_group_access)
12933 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
12934 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
12935 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
12936 result of vect_dr_stmt and use the stmt_vec_info instead of
12937 the associated gimple stmt.
12938 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
12939 (vect_gen_prolog_loop_niters): Likewise.
12940 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
12942 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12944 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
12945 to a vec<stmt_vec_info>.
12946 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
12947 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
12948 vec<stmt_vec_info>.
12949 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
12950 to a vec<stmt_vec_info>.
12951 (bst_traits::value_type, bst_traits::value_type): Likewise.
12952 (bst_traits::hash): Update accordingly.
12953 (vect_get_and_check_slp_defs): Change the stmts parameter from
12954 a vec<gimple *> to a vec<stmt_vec_info>.
12955 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
12956 (vect_build_slp_tree): Likewise.
12957 (vect_build_slp_tree_2): Likewise. Update uses of
12958 SLP_TREE_SCALAR_STMTS.
12959 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
12960 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
12961 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
12962 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
12963 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
12964 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
12965 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
12966 (vect_get_constant_vectors, vect_get_slp_defs)
12967 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
12968 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
12969 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
12970 instead of gimple stmts.
12971 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
12972 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
12973 (vect_slp_analyze_instance_dependence): Update uses of
12974 SLP_TREE_SCALAR_STMTS.
12975 (vect_slp_analyze_and_verify_node_alignment): Likewise.
12976 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
12977 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
12978 (get_initial_defs_for_reduction): Likewise.
12979 (vect_create_epilog_for_reduction): Likewise.
12980 (vectorize_fold_left_reduction): Likewise.
12981 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
12982 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
12983 (can_vectorize_live_stmts): Likewise.
12985 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12987 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
12988 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
12989 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
12991 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
12993 (gather_scalar_reductions): Update after above interface changes.
12994 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
12995 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
12997 (vect_force_simple_reduction): Likewise.
12998 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
12999 LOOP_VINFO_REDUCTIONS.
13000 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
13002 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13004 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
13005 a gimple stmt to a stmt_vec_info.
13006 * tree-vect-loop.c (vect_active_double_reduction_p)
13007 (vect_force_simple_reduction, vectorizable_reduction): Update
13010 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13012 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
13013 vec<gimple *> to a vec<stmt_vec_info>.
13014 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
13015 the reduction_phis argument from a vec<gimple *> to a
13016 vec<stmt_vec_info>.
13017 (vectorizable_reduction): Likewise the phis local variable that
13018 is passed to vect_create_epilog_for_reduction. Update for new type
13019 of SLP_TREE_VEC_STMTS.
13020 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
13021 (vectorizable_live_operation): Likewise.
13022 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
13023 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
13025 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13027 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
13028 a gimple stmt to a stmt_vec_info.
13029 (vectorizable_condition, vectorizable_live_operation)
13030 (vectorizable_reduction, vectorizable_induction): Pass back the
13031 vectorized statement as a stmt_vec_info.
13032 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
13033 use of STMT_VINFO_VEC_STMT.
13034 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
13035 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
13036 as stmt_vec_infos rather than gimple stmts.
13037 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
13038 to a stmt_vec_info.
13039 (vectorizable_live_operation): Likewise.
13040 (vectorizable_reduction, vectorizable_induction): Likewise,
13041 updating use of STMT_VINFO_VEC_STMT.
13042 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
13043 of STMT_VINFO_VEC_STMT.
13044 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
13045 (vectorizable_simd_clone_call, vectorizable_conversion)
13046 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
13047 (vectorizable_store, vectorizable_load, vectorizable_condition)
13048 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
13049 from a gimple stmt to a stmt_vec_info.
13050 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
13051 pointer to a stmt_vec_info to the vectorizable_* routines.
13053 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13055 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
13056 a gimple stmt to a stmt_vec_info.
13057 (is_pattern_stmt_p): Update accordingly.
13058 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
13059 (vect_record_grouped_load_vectors): Likewise.
13060 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
13061 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
13062 (vect_model_reduction_cost): Likewise.
13063 (vect_create_epilog_for_reduction): Likewise.
13064 (vectorizable_reduction, vectorizable_induction): Likewise.
13065 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
13066 Return the stmt_vec_info for the pattern statement.
13067 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
13068 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
13069 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
13070 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
13071 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
13072 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
13073 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
13074 (free_stmt_vec_info, vect_is_simple_use): Likewise.
13076 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13078 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
13079 (vect_finish_stmt_generation): Likewise.
13080 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
13081 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
13082 (vect_build_gather_load_calls): Use the return value of the above
13083 functions instead of a separate call to vinfo_for_stmt. Use narrow
13084 scopes for the input gimple stmt and wider scopes for the associated
13085 stmt_vec_info. Use vec_info::lookup_def when setting these
13086 stmt_vec_infos from an SSA_NAME definition.
13087 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
13088 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
13089 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
13090 (vectorizable_store, vectorizable_load, vectorizable_condition)
13091 (vectorizable_comparison): Likewise.
13092 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
13093 (vectorizable_reduction): Likewise.
13095 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13097 * tree-vectorizer.h (vect_is_simple_use): Add an optional
13098 stmt_vec_info * parameter before the optional gimple **.
13099 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
13100 (process_use, vect_get_vec_def_for_operand_1): Update callers.
13101 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
13102 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13103 (vectorizable_live_operation): Likewise.
13104 * tree-vect-patterns.c (type_conversion_p): Likewise.
13105 (vect_look_through_possible_promotion): Likewise.
13106 (vect_recog_rotate_pattern): Likewise.
13107 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
13109 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13111 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
13112 a typedef to a wrapper class.
13113 (NULL_STMT_VEC_INFO): New macro.
13114 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
13115 (stmt_vec_info::operator*): New function.
13116 (stmt_vec_info::operator gimple *): Likewise.
13117 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
13118 (add_stmt_costs): Likewise.
13119 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
13120 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
13121 (vect_get_known_peeling_cost): Likewise.
13122 (vect_estimate_min_profitable_iters): Likewise.
13123 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
13124 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
13125 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
13126 (vectorizable_store, free_stmt_vec_infos): Likewise.
13127 (new_stmt_vec_info): Change return type of xcalloc to
13130 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13132 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
13133 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
13134 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
13135 a single_imm_use-based sequence.
13136 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
13138 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13140 * tree-vectorizer.h (vec_info::lookup_def): Declare.
13141 * tree-vectorizer.c (vec_info::lookup_def): New function.
13142 * tree-vect-patterns.c (vect_get_internal_def): Use it.
13143 (vect_widened_op_tree): Likewise.
13144 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
13145 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
13146 (vectorizable_reduction): Likewise.
13147 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
13149 (vect_is_slp_reduction): Update calls accordingly. Use
13150 vec_info::lookup_def.
13151 (vect_is_simple_reduction): Likewise
13152 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
13154 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13156 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
13157 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
13158 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
13160 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
13161 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
13162 (vect_update_vf_for_slp, vect_analyze_loop_operations)
13163 (vect_is_slp_reduction, vectorizable_induction)
13164 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
13165 * tree-vect-patterns.c (vect_init_pattern_stmt):
13166 (vect_determine_min_output_precision_1, vect_determine_precisions)
13167 (vect_pattern_recog): Likewise.
13168 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
13169 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
13170 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
13171 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
13172 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
13173 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
13174 info field from a loop to a loop_vec_info.
13176 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13178 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
13179 (vec_info::add_stmt): Declare.
13180 * tree-vectorizer.c (vec_info::add_stmt): New function.
13181 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
13182 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
13183 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
13184 (vectorizable_induction): Likewise.
13185 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
13186 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
13187 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
13188 (vectorizable_load): Likewise.
13189 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
13190 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
13191 (vect_recog_gather_scatter_pattern): Likewise.
13192 (append_pattern_def_seq): Likewise. Remove a check that is
13193 performed by add_stmt itself.
13195 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13197 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
13198 which make_ssa_name was called with new_stmt before new_stmt
13201 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13203 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
13205 (vect_is_slp_reduction): ...here...
13206 (vect_is_simple_reduction): ...and here. Remove repetition of tests
13207 that are already known to be false.
13209 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13211 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
13212 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
13213 STMT_VINFO_NUM_SLP_USES when it's true.
13214 (vect_free_slp_instance): Add a final_p parameter and pass it to
13215 vect_free_slp_tree.
13216 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
13217 (vect_analyze_slp_instance): Likewise.
13218 (vect_slp_analyze_operations): Likewise.
13219 (vect_slp_analyze_bb_1): Likewise.
13220 * tree-vectorizer.c (vec_info): Likewise.
13221 * tree-vect-loop.c (vect_transform_loop): Likewise.
13223 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13225 * tree-vect-loop.c (vectorizable_reduction): Assert that the
13226 function is not called for second and subsequent members of
13229 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13231 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
13232 cases for nested loops from here to ...
13233 (vect_create_epilog_for_reduction): ...here. Only call
13234 vect_is_simple_use for inner-loop reductions.
13236 2018-07-31 Martin Liska <mliska@suse.cz>
13238 PR gcov-profile/85338
13239 PR gcov-profile/85350
13240 PR gcov-profile/85372
13241 * profile.c (struct location_triplet): New.
13242 (struct location_triplet_hash): Likewise.
13243 (output_location): Do not output a BB that
13244 is already recorded for a line.
13245 (branch_prob): Use streamed_locations.
13247 2018-07-31 Martin Liska <mliska@suse.cz>
13249 PR gcov-profile/85370
13250 * coverage.c (coverage_begin_function): Do not mark target
13251 clones as artificial functions.
13253 2018-07-31 Martin Liska <mliska@suse.cz>
13255 PR gcov-profile/83813
13256 PR gcov-profile/84758
13257 PR gcov-profile/85217
13258 PR gcov-profile/85332
13259 * profile.c (branch_prob): Do not record GOTO expressions
13260 for GIMPLE statements which locations are already streamed.
13262 2018-07-31 Olivier Hainque <hainque@adacore.com>
13264 * gcc.c (handle_spec_function): Accept a soft_matched_part
13265 argument, as do_spec_1. Pass it down to ...
13266 (eval_spec_function): Accept a soft_matched_part argument,
13267 and pass it down to ...
13268 (do_spec_2): Accept a soft_matched_part argument, and pass
13269 it down to do_spec_1.
13270 (do_spec_1): Pass soft_matched_part to handle_spec_function.
13271 (handle_braces): Update call to handle_spec_function.
13272 (driver::set_up_specs): Update calls to do_spec_2.
13273 (compare_debug_dump_opt_spec_function): Likewise.
13274 (compare_debug_self_opt_spec_function): Likewise.
13276 2018-07-31 Olivier Hainque <hainque@adacore.com>
13278 * common.opt (nolibc): New option.
13279 * doc/invoke.texi (Link Options): Document it.
13280 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
13281 * config/alpha/linux.h: Likewise.
13282 * config/arc/elf.h: Likewise.
13283 * config/arm/uclinux-elf.h: Likewise.
13284 * config/arm/unknown-elf.h: Likewise.
13285 * config/avr/avrlibc.h: Likewise.
13286 * config/bfin/bfin.h: Likewise.
13287 * config/bfin/linux.h: Likewise.
13288 * config/bfin/uclinux.h: Likewise.
13289 * config/darwin.h: Likewise.
13290 * config/darwin10.h: Likewise.
13291 * config/darwin12.h: Likewise.
13292 * config/gnu-user.h: Likewise.
13293 * config/lm32/uclinux-elf.h: Likewise.
13294 * config/pa/pa-hpux11.h: Likewise.
13295 * config/pa/pa64-hpux.h: Likewise.
13296 * config/sparc/sparc.h: Likewise.
13298 2018-07-31 Olivier Hainque <hainque@adacore.com>
13300 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
13301 undefined variables.
13303 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
13306 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
13307 instead of GEN_INT.
13309 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
13311 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
13312 terminated string literal.
13314 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
13316 PR rtl-optimization/85160
13317 * combine.c (is_just_move): New function.
13318 (try_combine): Allow combining two instructions into two if neither of
13319 the original instructions was a move.
13321 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
13324 * doc/extend.texi (Global Register Variables): Discourage use of type
13326 (Local Register Variables): Likewise.
13328 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
13330 PR tree-optimization/86506
13331 * hwint.h (ceil_log2): Resync with hwint.c implementation.
13333 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
13336 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
13337 hard_regno, make sure no insn between `from` and `to` clobbers it.
13339 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
13340 Tom de Vries <tdevries@suse.de>
13342 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
13343 (PTX_DEFAULT_RUNTIME_DIM): ... this.
13344 (nvptx_goacc_validate_dims): Set default worker and gang dims to
13345 PTX_DEFAULT_RUNTIME_DIM.
13346 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
13348 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
13350 * config/pa/pa.c (pa_output_addr_vec): Align address table.
13351 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
13352 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
13354 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
13356 * config/rs6000/constraints.md (wG constraint): Delete, no longer
13358 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
13359 predicate to reflect toc fusion has been deleted.
13360 (toc_fusion_mem_raw): Delete, no longer used.
13361 (toc_fusion_mem_wrapped): Likewise.
13362 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
13364 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
13365 Delete, no longer used.
13366 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
13367 meant to be used for toc fusion.
13368 (rs6000_debug_print_mode): Delete toc fusion debugging.
13369 (rs6000_debug_reg_global): Likewise.
13370 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
13371 fusion and secondary reload support that were never used.
13372 (rs6000_option_override_internal): Delete TOC fusion, that was only
13373 partially defined, and it did not work unless you also used the
13375 (rs6000_legitimate_address_p): Delete TOC fusion support.
13376 (rs6000_opt_masks): Likewise.
13377 (fusion_wrap_memory_address): Delete function, no longer used.
13378 (fusion_split_address); Delete TOC fusion support.
13379 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
13380 longer used with toc fusion being deleted.
13381 (TARGET_TOC_FUSION_FP): Likewise.
13382 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
13384 (toc fusion spliter): Delete TOC fusion support.
13385 (toc_fusionload_<mode>): Likewise.
13386 (toc_fusionload_di): Likewise.
13387 (fusion_gpr_load_<mode>): Delete generator function, this insn no
13388 longer needs to be named. Rename predicate to delete TOC fusion.
13389 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
13390 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
13391 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
13392 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
13393 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
13395 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
13397 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
13398 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
13399 __int128 in built-in function prototypes.
13400 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
13401 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
13403 2018-07-27 Martin Sebor <msebor@redhat.com>
13405 PR tree-optimization/86696
13406 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
13407 types, including enums.
13408 (handle_char_store): Be prepared for the above function to fail.
13410 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
13412 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
13413 when optimization level is lower than 2 or optimize for size.
13415 2018-07-26 Martin Sebor <msebor@redhat.com>
13417 PR tree-optimization/86043
13418 PR tree-optimization/86042
13419 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
13420 (get_string_cst_length): Rename...
13421 (get_min_string_length): ...to this. Add argument.
13422 (handle_char_store): Extend to handle multi-character stores by
13424 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
13425 * tree.h (initializer_zerop): Add argument.
13427 2018-07-26 Jakub Jelinek <jakub@redhat.com>
13429 PR middle-end/86660
13430 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
13431 declare target to variables if they have always,{to,from,tofrom} map
13434 2018-07-26 Martin Liska <mliska@suse.cz>
13437 * lto-wrapper.c: Add linker_output as prefix
13438 for ltrans_output_file.
13440 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
13442 PR rtl-optimization/85805
13443 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
13444 value for hard registers if that was written in the same mode.
13446 2018-07-26 Martin Liska <mliska@suse.cz>
13448 PR gcov-profile/86536
13449 * gcov.c (format_gcov): Use printf format %.*f directly
13450 and do not handle special values.
13452 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
13454 * common/config/arc/arc-common.c (arc_option_optimization_table):
13455 Update default optimizations for size.
13457 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
13459 * config/arc/arc.md (movsf_insn): Add short instruction selection.
13460 * config/arc/constraints.md (CfZ): New constraint.
13461 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
13462 (subsf3_fpu): Likewise.
13463 (cmpsf_fpu): Likewise.
13464 (cmpsf_fpu_uneq): Likewise.
13466 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
13468 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
13470 (arc_is_uncached_mem_p): uncached applies to both the variable and
13473 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
13475 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
13478 2018-07-25 David Malcolm <dmalcolm@redhat.com>
13480 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
13481 field "m_scopes" from vec to auto_vec.
13483 2018-07-25 Martin Liska <mliska@suse.cz>
13485 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
13488 2018-07-25 Richard Biener <rguenther@suse.de>
13491 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
13492 special wrt context_die late.
13493 (gen_subprogram_die): Re-use DIEs in local scope.
13495 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
13497 PR tree-optimization/86644
13498 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
13500 2018-07-25 Martin Liska <mliska@suse.cz>
13502 PR middle-end/86645
13503 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
13504 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
13506 2018-07-25 Martin Liska <mliska@suse.cz>
13509 * params.def: Explain ASan abbreviation and provide
13510 a documentation link.
13512 2018-07-24 Martin Sebor <msebor@redhat.com>
13514 PR tree-optimization/86622
13515 PR tree-optimization/86532
13516 * builtins.h (string_length): Declare.
13517 * builtins.c (c_strlen): Correct handling of non-constant offsets.
13518 (check_access): Be prepared for non-constant length ranges.
13519 (string_length): Make extern.
13520 * expr.c (string_constant): Only handle the minor non-constant
13521 array index. Use string_constant to compute the length of
13522 a generic string constant.
13524 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
13526 PR tree-optimization/86618
13527 * tree-vect-stmts.c (vectorizable_call): Don't take the address
13528 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
13530 2018-07-24 David Malcolm <dmalcolm@redhat.com>
13532 PR tree-optimization/86636
13533 * json.cc (json::object::set): Fix comment. Add assertions.
13534 (json::array::append): Move here from json.h. Add comment and an
13536 (json::string::string): Likewise.
13537 * json.h (json::array::append): Move to json.cc.
13538 (json::string::string): Likewise.
13539 * optinfo-emit-json.cc
13540 (optrecord_json_writer::impl_location_to_json): Assert that we
13541 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
13542 wrapper around it. Expand the location once, rather than three
13544 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
13545 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
13547 (optrecord_json_writer::optinfo_to_json): Likewise, in four
13548 places. Fix some overlong lines.
13550 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
13552 * config/aarch64/aarch64-simd.md
13553 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
13554 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
13555 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
13556 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
13557 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
13558 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
13559 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
13560 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
13561 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
13563 2018-07-24 Jakub Jelinek <jakub@redhat.com>
13565 PR middle-end/86627
13566 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
13567 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
13568 and abs_d == d, do the power of two handling if profitable.
13570 2018-07-24 Richard Biener <rguenther@suse.de>
13572 * match.pd: Add BIT_FIELD_REF canonicalizations.
13574 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
13577 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
13579 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
13581 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
13582 terminated STRING_CST object.
13584 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
13586 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
13588 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
13590 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
13591 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
13592 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
13593 the elements into a register.
13594 (rs6000_split_v4si_init_di_reg): Delete.
13595 (rs6000_split_v4si_init): Delete.
13596 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
13597 (vsx_init_v4si): Rewrite as a define_expand.
13599 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
13601 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
13602 zero_extend argument from memory): New.
13604 2018-07-22 Martin Sebor <msebor@redhat.com>
13607 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
13608 diagnosing calls with unknown arguments unless -Walloca-larger-than
13609 is restricted to less than PTRDIFF_MAX bytes.
13611 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
13613 * doc/gcov.texi (Invoking Gcov): Editorial changes.
13615 2018-07-20 David Malcolm <dmalcolm@redhat.com>
13617 * pretty-print.c (text_info::set_location): Remove redundant
13618 "line_table" parameter from call to rich_location::set_range.
13620 2018-07-20 Martin Sebor <msebor@redhat.com>
13622 PR middle-end/82063
13623 * builtins.c (expand_builtin_alloca): Adjust.
13624 * calls.c (alloc_max_size): Simplify.
13625 * cgraphunit.c (cgraph_node::expand): Adjust.
13626 * common.opt (larger_than_size, warn_frame_larger_than): Remove
13628 (frame_larger_than_size): Same.
13629 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
13630 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
13632 * doc/invoke.texi (GCC Command Options): Document option arguments.
13633 Explain byte-size arguments and suffixes.
13634 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
13635 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
13636 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
13637 * doc/options.texi (UInteger): Expand.
13638 (Host_Wide_Int, ByteSize): Document new properties.
13639 * final.c (final_start_function_1): Include sizes in an error message.
13640 * function.c (frame_offset_overflow): Same.
13641 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
13642 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
13643 Diagnose unbounded alloca calls only for limits of less than
13645 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
13646 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
13648 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
13649 only for limits of less than PTRDIFF_MAX.
13650 * langhooks-def.h (lhd_handle_option): Change function argument
13652 * langhooks.c (lhd_handle_option): Same.
13653 * langhooks.h (handle_option): Same.
13654 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
13656 (var_type, var_type_struct): Same.
13657 (var_set): Handle ByteSize flag.
13658 * optc-gen.awk: Add comments to output to ease debugging. Make
13659 use of HOST_WIDE_INT where appropriate.
13660 * opts-gen-save.awk: Use %lx to format unsigned long.
13661 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
13662 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
13663 arguments. Parse bytes-size suffixes.
13664 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
13665 (enum_value_to_arg): Same.
13666 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
13667 (handle_option): Adjust.
13668 (generate_option): Change function argument to HOST_WIDE_INT.
13669 (cmdline_handle_error): Adjust.
13670 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
13671 (set_option): Change function argument to HOST_WIDE_INT.
13672 (option_enabled): Handle cl_host_wide_int.
13673 (get_option_state): Handle CLVC_SIZE.
13674 (control_warning_option): Same.
13675 * opts.c (common_handle_option): Change function argument to
13676 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
13677 OPT_Wvla_larger_than_.
13678 * opts.h (enum cl_var_type): Add an enumerator.
13679 * stor-layout.c (layout_decl): Print a more meaningful warning.
13680 * toplev.c (output_stack_usage): Adjust.
13682 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
13684 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
13685 call to inline_expand_builtin_string_cmp.
13686 (expand_builtin_strcmp): Likewise.
13687 (expand_builtin_strncmp): Likewise.
13688 (inline_string_cmp): Delete the last parameter, change char_type_node
13689 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
13691 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
13692 the inlining expansion on target where the type of the call has same or
13693 narrower precision than unsigned char.
13695 2018-07-20 David Malcolm <dmalcolm@redhat.com>
13697 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
13698 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
13699 * common.opt (fsave-optimization-record): New option.
13700 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
13701 * doc/invoke.texi (-fsave-optimization-record): New option.
13702 * dumpfile.c: Include "optinfo-emit-json.h".
13703 (struct kv_pair): Move to coretypes.h.
13704 (optgroup_options): Make non-static.
13705 (dump_context::end_scope): Call
13706 optimization_records_maybe_pop_dump_scope.
13707 * dumpfile.h (optgroup_options): New decl.
13708 * json.cc: New file.
13709 * json.h: New file.
13710 * optinfo-emit-json.cc: New file.
13711 * optinfo-emit-json.h: New file.
13712 * optinfo.cc: Include "optinfo-emit-json.h".
13713 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
13714 (optinfo_enabled_p): Check optimization_records_enabled_p.
13715 (optinfo_wants_inlining_info_p): Likewise.
13716 * optinfo.h: Update comment.
13717 * profile-count.c (profile_quality_as_string): New function.
13718 * profile-count.h (profile_quality_as_string): New decl.
13719 (profile_count::quality): New accessor.
13720 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
13721 and optinfo_emit_json_cc_tests.
13722 * selftest.h (selftest::json_cc_tests): New decl.
13723 (selftest::optinfo_emit_json_cc_tests): New decl.
13724 * toplev.c: Include "optinfo-emit-json.h".
13725 (compile_file): Call optimization_records_finish.
13726 (do_compile): Call optimization_records_start.
13727 * tree-ssa-live.c: Include optinfo.h.
13728 (remove_unused_scope_block_p): Retain inlining information if
13729 optinfo_wants_inlining_info_p returns true.
13731 2018-07-20 Richard Biener <rguenther@suse.de>
13734 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
13735 to cover -flto-partition=none.
13737 2018-07-20 Martin Liska <mliska@suse.cz>
13739 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
13740 (get_decl_source_range): Remove unused function.
13742 2018-07-20 Richard Biener <rguenther@suse.de>
13744 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
13745 (struct vn_phi_s): Likewise.
13746 (struct vn_reference_s): Likewise.
13747 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
13748 for searching the slot of an entry known to be in the hash itself.
13749 (vn_phi_hasher::equal): Likewise.
13750 (vn_reference_hasher::equal): Likewise.
13751 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
13753 (optimistic_info, current_info): Remove, keeping only valid_info.
13754 (vn_reference_lookup_1): Remove fallback lookup.
13755 (vn_reference_lookup_2): Likewise.
13756 (vn_nary_op_lookup_1): Likewise.
13757 (vn_phi_lookup): Likewise.
13758 (vn_nary_build_or_lookup_1): Make sure to not chain the built
13760 (vn_reference_insert): Adjust, chain the inserted hash element
13761 at last_inserted_ref.
13762 (vn_reference_insert_pieces): Likewise.
13763 (visit_reference_op_call): Likewise.
13764 (vn_nary_op_insert_into): Chain the inserted hash element at
13765 last_inserted_nary.
13766 (vn_nary_op_insert_pieces): Adjust.
13767 (vn_nary_op_insert): Likewise.
13768 (vn_nary_op_insert_stmt): Likewise.
13769 (vn_phi_insert): Adjust, chain the inserted hash element at
13771 (process_scc): Remove clearing and copying the optimistic
13772 table. Instead remove elements inserted during an optimistic
13773 iteration from the single table we maintain.
13774 (init_scc_vn): Adjust.
13775 (free_scc_vn): Likewise.
13776 (sccvn_dom_walker::record_cond): Likewise.
13777 (sccvn_dom_walker::after_dom_children): Likewise.
13779 2018-07-19 Martin Sebor <msebor@redhat.com>
13781 PR tree-optimization/84047
13782 PR tree-optimization/83776
13783 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
13784 (check_array_bounds): Call it.
13786 2018-07-19 Martin Sebor <msebor@redhat.com>
13788 * align.h (align_flags): Use member initialization.
13790 2018-07-19 David Malcolm <dmalcolm@redhat.com>
13792 * Makefile.in (OBJS): Add optinfo.o.
13793 * coretypes.h (class symtab_node): New forward decl.
13794 (struct cgraph_node): New forward decl.
13795 (class varpool_node): New forward decl.
13796 * dump-context.h: New file.
13797 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
13799 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
13800 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
13801 (set_alt_dump_file): Likewise.
13802 (dump_context::~dump_context): New dtor.
13803 (dump_gimple_stmt): Move implementation to...
13804 (dump_context::dump_gimple_stmt): ...this new member function.
13805 Add the stmt to any pending optinfo, creating one if need be.
13806 (dump_gimple_stmt_loc): Move implementation to...
13807 (dump_context::dump_gimple_stmt_loc): ...this new member function.
13808 Start a new optinfo and add the stmt to it.
13809 (dump_gimple_expr): Move implementation to...
13810 (dump_context::dump_gimple_expr): ...this new member function.
13811 Add the stmt to any pending optinfo, creating one if need be.
13812 (dump_gimple_expr_loc): Move implementation to...
13813 (dump_context::dump_gimple_expr_loc): ...this new member function.
13814 Start a new optinfo and add the stmt to it.
13815 (dump_generic_expr): Move implementation to...
13816 (dump_context::dump_generic_expr): ...this new member function.
13817 Add the tree to any pending optinfo, creating one if need be.
13818 (dump_generic_expr_loc): Move implementation to...
13819 (dump_context::dump_generic_expr_loc): ...this new member
13820 function. Add the tree to any pending optinfo, creating one if
13822 (dump_printf): Move implementation to...
13823 (dump_context::dump_printf_va): ...this new member function. Add
13824 the text to any pending optinfo, creating one if need be.
13825 (dump_printf_loc): Move implementation to...
13826 (dump_context::dump_printf_loc_va): ...this new member function.
13827 Start a new optinfo and add the stmt to it.
13828 (dump_dec): Move implementation to...
13829 (dump_context::dump_dec): ...this new member function. Add the
13830 value to any pending optinfo, creating one if need be.
13831 (dump_context::dump_symtab_node): New member function.
13832 (dump_context::get_scope_depth): New member function.
13833 (dump_context::begin_scope): New member function.
13834 (dump_context::end_scope): New member function.
13835 (dump_context::ensure_pending_optinfo): New member function.
13836 (dump_context::begin_next_optinfo): New member function.
13837 (dump_context::end_any_optinfo): New member function.
13838 (dump_context::s_current): New global.
13839 (dump_context::s_default): New global.
13840 (dump_scope_depth): Delete global.
13841 (dumpfile_ensure_any_optinfo_are_flushed): New function.
13842 (dump_symtab_node): New function.
13843 (get_dump_scope_depth): Reimplement in terms of dump_context.
13844 (dump_begin_scope): Likewise.
13845 (dump_end_scope): Likewise.
13846 (selftest::temp_dump_context::temp_dump_context): New ctor.
13847 (selftest::temp_dump_context::~temp_dump_context): New dtor.
13848 (selftest::verify_item): New function.
13849 (ASSERT_IS_TEXT): New macro.
13850 (ASSERT_IS_TREE): New macro.
13851 (ASSERT_IS_GIMPLE): New macro.
13852 (selftest::test_capture_of_dump_calls): New test.
13853 (selftest::dumpfile_c_tests): Call it.
13854 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
13855 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
13856 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
13857 descriptive comment.
13858 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
13859 (dump_node, dump_bb): Move these unrelated decls.
13860 (class dump_manager): Add leading comment.
13861 * optinfo.cc: New file.
13862 * optinfo.h: New file.
13864 2018-07-19 Michael Collison <michael.collison@arm.com>
13865 Richard Henderson <rth@redhat.com>
13867 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
13868 (subti): Handle op1 zero.
13869 (subvti4, usub4ti4): New.
13870 (*sub<GPI>3_compare1_imm): New.
13871 (sub<GPI>3_carryinCV): New.
13872 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
13873 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
13875 2018-07-19 Michael Collison <michael.collison@arm.com>
13876 Richard Henderson <rth@redhat.com>
13878 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
13879 (addti3): Create simpler code if low part is already known to be 0.
13880 (addvti4, uaddvti4): New.
13881 (*add<GPI>3_compareC_cconly_imm): New.
13882 (*add<GPI>3_compareC_cconly): New.
13883 (*add<GPI>3_compareC_imm): New.
13884 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
13885 handle constants within this pattern..
13886 (*add<GPI>3_compareV_cconly_imm): New.
13887 (*add<GPI>3_compareV_cconly): New.
13888 (*add<GPI>3_compareV_imm): New.
13889 (add<GPI>3_compareV): New.
13890 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
13891 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
13892 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
13893 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
13894 with 'comparison' operator.
13895 (*add<GPI>3_compareV_cconly_imm): Ditto.
13896 (*add<GPI>3_compareV_cconly): Ditto.
13897 (*add<GPI>3_compareV_imm): Ditto.
13898 (add<GPI>3_compareV): Ditto.
13899 (add<mode>3_carryinC): Ditto.
13900 (*add<mode>3_carryinC_zero): Ditto.
13901 (*add<mode>3_carryinC): Ditto.
13902 (add<mode>3_carryinV): Ditto.
13903 (*add<mode>3_carryinV_zero): Ditto.
13904 (*add<mode>3_carryinV): Ditto.
13906 2018-07-19 Michael Collison <michael.collison@arm.com>
13907 Richard Henderson <rth@redhat.com>
13909 * config/aarch64/aarch64-modes.def (CC_V): New.
13910 * config/aarch64/aarch64-protos.h
13911 (aarch64_addti_scratch_regs): Declare
13912 (aarch64_subvti_scratch_regs): Declare.
13913 (aarch64_expand_subvti): Declare.
13914 (aarch64_gen_unlikely_cbranch): Declare
13915 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
13916 for signed overflow using CC_Vmode.
13917 (aarch64_get_condition_code_1): Handle CC_Vmode.
13918 (aarch64_gen_unlikely_cbranch): New function.
13919 (aarch64_addti_scratch_regs): New function.
13920 (aarch64_subvti_scratch_regs): New function.
13921 (aarch64_expand_subvti): New function.
13923 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
13925 * config/aarch64/aarch64-option-extensions.def: New entry for profile
13927 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
13928 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
13931 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
13934 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
13935 address check not strict.
13937 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
13939 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
13941 (store_pair_lanes<mode>): Likewise.
13942 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
13943 enum value 'ADDR_QUERY_LDP_STP_N'.
13944 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
13945 (aarch64_print_address_internal): Add declaration.
13946 (aarch64_print_ldpstp_address): Remove.
13947 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
13948 (aarch64_print_operand): Change printing of 'y'.
13949 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
13950 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
13951 'true' rather than '1'.
13952 * config/aarch64/constraints.md (Uml): Likewise.
13953 (Uml): Rename to Umn.
13956 2018-07-19 Richard Biener <rguenther@suse.de>
13958 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
13960 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
13961 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
13962 (vn_reference_hasher): Likewise.
13963 (struct vn_tables_s): Remove obstack and alloc-pool members.
13964 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
13965 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
13966 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
13967 (vn_reference_insert_pieces): Likewise.
13968 (alloc_vn_nary_op_noinit): Adjust.
13969 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
13970 (vn_phi_eq): Adjust.
13971 (shared_lookup_phiargs): Remove.
13972 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
13973 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
13974 (visit_reference_op_call): Likewise.
13975 (copy_nary, copy_phi, copy_reference): Remove.
13976 (process_scc): Rewind the obstack when iterating. Do not
13977 copy the elements to valid_info but just move them from one
13978 hashtable to the other.
13979 (allocate_vn_table): Adjust.
13980 (free_vn_table): Likewise.
13981 (init_scc_vn): Likewise.
13982 (free_scc_vn): Likewise.
13984 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
13987 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
13988 indirect_return as function type attribute.
13989 (ix86_attribute_table): Change indirect_return to function
13991 * doc/extend.texi: Update indirect_return attribute.
13993 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
13995 * wide-int.h (widest2_int): New.
13996 * gimple-fold.c (arith_overflowed_p): Use it.
13997 * tree.h (widest2_int_cst): New.
13998 * tree-vrp.c (wide_int_binop_overflow): Rename from
13999 vrp_int_const_binop.
14000 Rewrite to work on trees.
14001 (extract_range_from_multiplicative_op_1): Abstract code to...
14002 (wide_int_range_min_max): ...here.
14003 (wide_int_range_cross_product): ...and here.
14004 (extract_range_from_binary_expr_1): Abstract overflow code to...
14005 (wide_int_range_mult_wrapping): ...here.
14006 * tree-vrp.h (wide_int_range_cross_product): New.
14007 (wide_int_range_mult_wrapping): New.
14009 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
14010 Julia Koval <julia.koval@intel.com>
14012 * config/i386/x86-tune-costs.h (skylake_memcpy,
14013 skylake_memset): Replace rep_prefix with unrolling for size 512.
14015 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
14017 PR middle-end/86544
14018 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
14019 comparision with EQ_EXPR in last stmt.
14021 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
14023 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
14024 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
14025 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
14026 previously known as "PowerPC AltiVec Built-in Functions". Move
14027 some material to new subsubsections "PowerPC AltiVec Built-in
14028 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
14030 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
14031 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
14032 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
14033 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
14035 2018-07-18 Richard Biener <rguenther@suse.de>
14037 PR tree-optimization/86557
14038 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
14041 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
14043 * config/s390/s390.c (s390_function_profiler): Generate CFI.
14045 2018-07-17 Jeff Law <law@redhat.com>
14047 * config/arm/arm.c (get_label_padding): Update for recent
14048 changes to label_to_alignment.
14050 PR tree-optimization/86010
14051 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
14053 * config/mips/mips.c (vr4130_align_insns): Update for recent
14054 changes to label_to_alignment.
14056 * config/frv/frv.c (frv_label_align): Update for recent changes
14057 to label_to_alignment.
14059 * config/nios2/nios2.c (nios2_label_align): Update for recent
14060 changes which dropped ALIGN_LABELS_LOG.
14062 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
14064 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
14065 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
14067 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14069 * config/arc/arc.c (arc_label_align): Use align_labels instead of
14070 deprecated align_labels_log.
14072 2018-07-17 Richard Biener <rguenther@suse.de>
14075 * dwarf2out.c (init_sections_and_labels): Always generate
14076 a debug_line_str_section for early LTO debug.
14077 (dwarf2out_finish): Reset debug_line_str_hash output early.
14078 Bump counter for extra dwarf5 .debug_loc labels to not conflict
14080 (dwarf2out_early_finish): Output debug_line_str.
14082 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
14084 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
14085 index register on z196 or later.
14087 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
14089 * config/s390/s390.c (s390_default_align): Set default function
14091 (s390_override_options_after_change): Call s390_default align.
14092 (s390_option_override_internal): Call s390_default_align.
14093 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
14095 2018-07-17 Jakub Jelinek <jakub@redhat.com>
14097 PR middle-end/86542
14098 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
14099 to _looptemp_ clauses, other than the first two.
14101 2018-07-17 Martin Liska <mliska@suse.cz>
14103 * opts.c: Do not enable OPT_falign_* for -Os.
14105 2018-07-17 Martin Liska <mliska@suse.cz>
14107 * align.h (MAX_CODE_ALIGN): New.
14108 (MAX_CODE_ALIGN_VALUE): New.
14109 * common/config/i386/i386-common.c (ix86_handle_option):
14110 (MAX_CODE_ALIGN): Moved to align.h.
14111 * final.c (MAX_CODE_ALIGN): Likewise.
14112 * opts.c (parse_and_check_align_values):
14113 (MAX_CODE_ALIGN): Likewise.
14114 (MAX_CODE_ALIGN_VALUE): Likewise.
14116 2018-07-17 Martin Liska <mliska@suse.cz>
14118 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
14119 in order to fulfil coding style.
14120 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
14121 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
14122 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
14123 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
14124 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
14125 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
14126 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
14127 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
14128 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
14130 2018-07-17 Martin Liska <mliska@suse.cz>
14132 * align.h: New file.
14133 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
14135 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
14136 align_flags of label_to_alignment.
14137 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
14139 * config/m68k/m68k.c: Do not use removed align_labels_value and
14141 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
14142 (LOOP_ALIGN): Likewise.
14143 (LABEL_ALIGN): Likewise.
14144 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
14145 Remove not used macro.
14146 (rs6000_loop_align): Change return type to align_flags.
14147 (rs6000_loop_align_max_skip): Remove.
14148 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
14149 Change return type to align_flags.
14150 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
14151 Remove not used macro.
14152 (rs6000_loop_align): Change return type to align_flags.
14153 (rs6000_loop_align_max_skip): Remove.
14154 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
14155 * config/rx/rx-protos.h (rx_align_for_label): Make it
14157 * config/rx/rx.c (rx_align_for_label): Change return type
14159 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
14161 into align_flags class.
14162 (LABEL_ALIGN): Likewise.
14163 (LOOP_ALIGN): Likewise.
14164 * config/s390/s390.c (s390_label_align): Use align_flags
14166 (s390_asm_output_function_label): Likewise.
14167 * config/sh/sh.c (sh_override_options_after_change):
14168 Use align_flags class directly without macros.
14169 (find_barrier): Likewise.
14170 (barrier_align): Likewise.
14171 (sh_loop_align): Likewise.
14172 * config/spu/spu.c (spu_option_override):
14173 Use align_flags_tuple::get_value instead of removed macros.
14174 (spu_sched_init): Likewise.
14175 * config/spu/spu.h (GTY): Likewise.
14176 * config/visium/visium.c (visium_option_override):
14177 Set "8" as default secondary alignment.
14178 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
14179 in order to guarantee secondary alignment of 8.
14180 * coretypes.h: Include align.h header file.
14181 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
14182 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
14183 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
14184 * doc/tm.texi.in: Likewise.
14185 * final.c (struct label_alignment): Remove not used structure.
14186 (LABEL_ALIGN): Change type to align_flags.
14187 (LOOP_ALIGN): Likewise.
14188 (JUMP_ALIGN): Likewise.
14189 (default_loop_align_max_skip): Remove.
14190 (default_label_align_max_skip): Likewise.
14191 (default_jump_align_max_skip): Likewise.
14192 (default_label_align_after_barrier_max_skip):
14193 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
14194 (LABEL_TO_MAX_SKIP): Remove.
14195 (label_to_alignment): Return align_flags type instead of integer.
14196 (label_to_max_skip): Remove.
14197 (align_fuzz): Use align_flags type.
14198 (compute_alignments): Use align_flags type and use align_flags::max
14199 to combine multiple alignments.
14200 (grow_label_align): Grow vec instead of C array.
14201 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
14202 (shorten_branches): Use align_flags type and use align_flags::max
14203 to combine multiple alignments.
14204 (final_scan_insn_1): Remove usage of secondary alignment that comes
14205 from label alignment, but instead use proper secondary alignment
14206 which is computed in grow_label_align.
14207 * flags.h (struct align_flags_tuple): Move to align.h.
14208 (struct align_flags): Likewise.
14209 (state_align_loops): Rename to align_loops.
14210 (state_align_jumps): Rename to align_jumps.
14211 (state_align_labels): Rename to align_labels.
14212 (state_align_functions): Rename to align_functions.
14213 (align_loops_log): Remove.
14214 (align_jumps_log): Remove.
14215 (align_labels_log): Remove.
14216 (align_functions_log): Remove.
14217 (align_loops_max_skip): Remove.
14218 (align_jumps_max_skip): Remove.
14219 (align_labels_max_skip): Remove.
14220 (align_functions_max_skip): Remove.
14221 (align_loops_value): Remove.
14222 (align_jumps_value): Remove.
14223 (align_labels_value): Remove.
14224 (align_functions_value): Remove.
14225 * output.h (label_to_alignment): Change return type to align_flags.
14226 (label_to_max_skip): Remove.
14227 * target.def: Remove loop_align_max_skip, label_align_max_skip,
14228 jump_align_max_skip macros.
14229 * targhooks.h (default_loop_align_max_skip): Remove.
14230 (default_label_align_max_skip): Likewise.
14231 (default_jump_align_max_skip): Likewise.
14232 (default_label_align_after_barrier_max_skip): Remove.
14233 * toplev.c (read_log_maxskip): Use ::normalize function.
14234 (parse_N_M): Remove not used argument and also call ::normalize.
14235 (parse_alignment_opts): Do not pass unused arguments.
14236 * varasm.c (assemble_start_function): Use directly align_functions
14237 instead of removed macros.
14238 * system.h: Do not poison removed macros.
14240 2018-07-17 Jakub Jelinek <jakub@redhat.com>
14242 PR middle-end/86539
14243 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
14244 and cond temporaries don't have reference type if iterator has
14245 pointer type. For init use &for_pre_body instead of pre_p if
14246 for_pre_body is non-empty.
14248 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
14250 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
14251 double-double modes to SFmode directly directly.
14252 (trunc<mode>sf2_fprs): Delete.
14254 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
14256 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
14257 for conversions between IFmode and the decimal floating point modes.
14258 (init_float128_ieee): Use the correct names for conversions between
14259 KFmode and the decimal floating point modes.
14261 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
14263 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
14264 for the conversions between TDmode and IFmode.
14265 (init_float128_ieee): Use more correct names for the conversions
14266 between TDmode and KFmode.
14268 2018-07-16 Jakub Jelinek <jakub@redhat.com>
14270 PR tree-optimization/86526
14271 * builtins.c (expand_builtin_memcmp): Formatting fixes.
14272 (inline_expand_builtin_string_cmp): Likewise.
14273 (inline_string_cmp): Likewise. Use c_readstr instead of
14274 builtin_memcpy_read_str. Add unit_mode temporary.
14276 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
14278 PR middle-end/86528
14279 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
14280 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
14282 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
14284 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
14285 Alphabetize prototypes of built-in functions, separating out
14286 built-in functions that are listed in this section but should be
14287 described elsewhere.
14289 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
14292 * expmed.c (emit_store_flag): Do not emit setcc followed by a
14293 conditional move when trapping comparison was split to a
14294 non-trapping one (and vice versa).
14296 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
14298 * config/s390/s390.c (s390_function_profiler): Generate nops
14299 instead of profiler call sequences.
14300 * config/s390/s390.opt: Add the new option.
14302 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
14304 * config/s390/s390.c (s390_function_profiler): Generate
14305 __mcount_loc section.
14306 * config/s390/s390.opt: Add the new option.
14308 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
14310 * common.opt: Add the new warning.
14311 * config/s390/s390.c (s390_function_profiler): Emit "brasl
14312 %r0,__fentry__" when -mfentry is specified.
14313 (s390_option_override_internal): Disallow -mfentry for 31-bit
14315 * config/s390/s390.opt: Add the new option.
14317 2018-07-16 Richard Biener <rguenther@suse.de>
14320 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
14321 for function-local FUNCTION_DECL and RESULT_DECL immediately.
14323 2018-07-16 Martin Liska <mliska@suse.cz>
14326 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
14329 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
14331 * config/arc/arcHS.md: Update ARCHS scheduling rules.
14333 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
14335 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
14337 * config/arc/arc-cpus.def (hs4x): New cpu.
14339 * config/arc/arc-tables.opt: Regenerate.
14340 * config/arc/arc.c (arc_sched_issue_rate): New function.
14341 (TARGET_SCHED_ISSUE_RATE): Define.
14342 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
14343 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
14345 (attr tune): Add ARCHS4x tune values.
14346 (attr tune_dspmpy): Define.
14347 (*tst): Correct instruction type.
14348 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
14349 * config/arc/arcHS4x.md: New file.
14350 * config/arc/fpu.md: Update instruction type attributes.
14351 * config/arc/t-multilib: Regenerate.
14353 2018-07-16 Tom de Vries <tdevries@suse.de>
14356 * var-tracking.c (vt_initialize): Fix pre_dec handling.
14358 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14360 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
14363 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
14365 PR tree-optimization/86514
14366 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
14367 conversion to a boolean type from a type with greater precision.
14369 2018-07-16 Tom de Vries <tdevries@suse.de>
14371 * var-tracking.c (vt_initialize): Print adjusted insn slim if
14372 dump_flags request TDF_SLIM.
14374 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
14376 * fold-const.c (int_const_binop_1): Abstract...
14377 (wide_int_binop): ...wide int code here.
14378 (poly_int_binop): ...poly int code here.
14379 Abstract the rest of int_const_binop_1 into int_const_binop.
14380 * fold-const.h (wide_int_binop): New.
14381 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
14382 Remove useless PLUS/MINUS_EXPR case.
14383 (zero_nonzero_bits_from_vr): Move wide int code...
14384 (zero_nonzero_bits_from_bounds): ...here.
14385 (extract_range_from_binary_expr_1): Move mask optimization code...
14386 (range_easy_mask_min_max): ...here.
14387 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
14388 (range_easy_mask_min_max): New.
14390 2018-07-15 Jeff Law <law@redhat.com>
14393 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
14396 2018-07-14 Jim Wilson <jimw@sifive.com>
14398 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
14400 2018-07-14 Paul Koning <ni1d@arrl.net>
14402 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
14404 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
14406 * lto-streamer-out.c (copy_function_or_variable): Dump info about
14409 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
14410 Steve Munroe <munroesj52@gmail.com>
14412 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
14413 (_mm_andnot_si128): Likewise.
14414 (_mm_or_si128): Likewise.
14415 (_mm_xor_si128): Likewise.
14417 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
14419 PR middle-end/78809
14420 * builtins.c (expand_builtin_memcmp): Inline the calls first
14421 when result_eq is false.
14422 (expand_builtin_strcmp): Inline the calls first.
14423 (expand_builtin_strncmp): Likewise.
14424 (inline_string_cmp): New routine. Expand a string compare
14425 call by using a sequence of char comparison.
14426 (inline_expand_builtin_string_cmp): New routine. Inline expansion
14427 a call to str(n)cmp/memcmp.
14428 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
14430 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
14432 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
14434 * config/arm/driver-arm.c: Include arm-native.h.
14435 (host_detect_local_cpu): Use auto-generated data tables.
14436 (vendors, arm_cpu_table): Delete. Move part information to ...
14437 * config/arm/arm-cpus.in: ... here.
14438 * config/arm/parsecpu.awk (gen_native): New function.
14439 (vendor, part): New CPU fields.
14440 (END): Add support for building the native CPU detection tables.
14441 * config/arm/t-arm (arm-native.h): Add build rule.
14442 (driver-arm.o): Add dependency on arm-native.h.
14444 2018-07-13 Richard Biener <rguenther@suse.de>
14446 PR middle-end/85974
14447 * match.pd (addr1 - addr2): Allow either of the operand to
14450 2018-07-13 Tom de Vries <tdevries@suse.de>
14452 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
14455 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
14457 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
14458 arrays instead of numbered variables.
14460 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
14462 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
14463 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
14465 2018-07-13 Richard Biener <rguenther@suse.de>
14468 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
14469 instead of get_context_die.
14471 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
14472 Richard Biener <rguenther@suse.de>
14474 PR middle-end/86489
14475 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
14476 that the loop latch destination where phi is defined.
14478 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
14480 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
14481 (riscv_expand_epilogue): Add assertion to check interrupt mode.
14482 (riscv_set_current_function): Extract getting interrupt type to new
14484 (riscv_get_interrupt_type): New function.
14485 (riscv_merge_decl_attributes): New function, checking interrupt type is
14487 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
14489 2018-07-12 Paul Koning <ni1d@arrl.net>
14491 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
14494 2018-07-12 Paul Koning <ni1d@arrl.net>
14496 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
14497 zero reference, add doloop_end instead.
14498 * doc/md.texi (decrement_and_branch_until_zero): Remove.
14499 (Looping patterns): Remove decrement_and_branch_until_zero. Add
14500 detail for doloop_end.
14502 2018-07-12 Martin Sebor <msebor@redhat.com>
14505 * attribs.c (decl_attributes): Reject conflicting attributes before
14506 calling attribute handlers.
14508 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
14510 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
14512 (gcc::dump_manager::get_dump_file_name): likewise.
14513 (dump_begin): Likewise.
14514 * dumpfile.h (dump_begin): Update prototype.
14515 (gcc::dump_manager::get_dump_file_name,
14516 gcc::dump_manager::get_dump_file_name): Update prototype.
14518 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
14520 * internal-fn.h (vectorizable_internal_fn_p): New function.
14521 * tree-vect-slp.c (compatible_calls_p): Likewise.
14522 (vect_build_slp_tree_1): Remove nops argument. Handle calls
14523 to internal functions.
14524 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
14526 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
14528 * fold-const.h (inverse_conditions_p): Declare.
14529 * fold-const.c (inverse_conditions_p): New function.
14530 * match.pd: Use inverse_conditions_p. Add folds of view_converts
14531 that test the inverse condition of a conditional internal function.
14532 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
14533 * internal-fn.c (internal_fn_mask_index): Handle conditional
14534 internal functions.
14535 (vectorized_internal_fn_supported_p): New function.
14536 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
14537 (any_pred_load_store): Replace with...
14538 (need_to_predicate): ...this new variable.
14539 (redundant_ssa_names): New variable.
14540 (ifcvt_can_use_mask_load_store): Move initial checks to...
14541 (ifcvt_can_predicate): ...this new function. Handle tree codes
14542 for which a conditional internal function exists.
14543 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
14544 instead of ifcvt_can_use_mask_load_store. Update after variable
14546 (predicate_load_or_store): New function, split out from
14547 predicate_mem_writes.
14548 (check_redundant_cond_expr): New function.
14549 (value_available_p): Likewise.
14550 (predicate_rhs_code): Likewise.
14551 (predicate_mem_writes): Rename to...
14552 (predicate_statements): ...this. Use predicate_load_or_store
14553 and predicate_rhs_code.
14554 (combine_blocks, tree_if_conversion): Update after above name changes.
14555 (ifcvt_local_dce): Handle redundant_ssa_names.
14556 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
14557 general conditional functions.
14558 * tree-vect-stmts.c (vectorizable_call): Likewise.
14560 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
14561 Alan Hayward <alan.hayward@arm.com>
14562 David Sherwood <david.sherwood@arm.com>
14564 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
14565 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
14566 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
14567 plus and minus and convert them into IFN_COND_FMA-based sequences.
14568 (convert_mult_to_fma): Handle conditional plus and minus.
14570 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
14572 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
14573 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
14574 (cond_fnms_optab): New optabs.
14575 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
14576 internal functions.
14577 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
14578 * internal-fn.h (get_conditional_internal_fn): Declare.
14579 (get_unconditional_internal_fn): Likewise.
14580 * internal-fn.c (cond_ternary_direct): New macro.
14581 (expand_cond_ternary_optab_fn): Likewise.
14582 (direct_cond_ternary_optab_supported_p): Likewise.
14583 (FOR_EACH_COND_FN_PAIR): Likewise.
14584 (get_conditional_internal_fn): New function.
14585 (get_unconditional_internal_fn): Likewise.
14586 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
14587 (gimple_match_op::gimple_match_op): Add a new overload for 5
14589 (gimple_match_op::set_op): Likewise.
14590 (gimple_resimplify5): Declare.
14591 * genmatch.c (decision_tree::gen): Generate simplifications for
14593 * gimple-match-head.c (gimple_simplify): Define an overload for
14594 5 operands. Handle calls with 5 arguments in the top-level overload.
14595 (convert_conditional_op): Handle conversions from unconditional
14596 internal functions to conditional ones.
14597 (gimple_resimplify5): New function.
14598 (build_call_internal): Pass a fifth operand.
14599 (maybe_push_res_to_seq): Likewise.
14600 (try_conditional_simplification): Try converting conditional
14601 internal functions to unconditional internal functions.
14602 Handle 3-operand unconditional forms.
14603 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
14604 Define ternary equivalents of the current rules for binary conditional
14605 internal functions.
14606 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
14607 ternary operations.
14608 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
14609 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
14610 (optab): Handle them.
14611 (SVE_COND_FP_TERNARY): New int iterator.
14612 (sve_fmla_op, sve_fmad_op): New int attributes.
14613 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
14614 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
14615 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
14617 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
14619 * target.def (preferred_else_value): New target hook.
14620 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
14621 * doc/tm.texi: Regenerate.
14622 * targhooks.h (default_preferred_else_value): Declare.
14623 * targhooks.c (default_preferred_else_value): New function.
14624 * internal-fn.h (conditional_internal_fn_code): Declare.
14625 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
14626 (get_conditional_internal_fn): Use it.
14627 (conditional_internal_fn_code): New function.
14628 * gimple-match.h (gimple_match_cond): New struct.
14629 (gimple_match_op): Add a cond member function.
14630 (gimple_match_op::gimple_match_op): Update all forms to take a
14632 * genmatch.c (expr::gen_transform): Use the same condition as res_op
14633 for the suboperation, but don't specify a particular else_value.
14634 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
14635 (visit_nary_op, visit_reference_op_load): Pass
14636 gimple_match_cond::UNCOND to the gimple_match_op constructor.
14637 * gimple-match-head.c: Include tree-eh.h
14638 (convert_conditional_op): New function.
14639 (maybe_resimplify_conditional_op): Likewise.
14640 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
14641 (gimple_resimplify2): Likewise.
14642 (gimple_resimplify3): Likewise.
14643 (gimple_resimplify4): Likewise.
14644 (maybe_push_res_to_seq): Return null for conditional operations.
14645 (try_conditional_simplification): New function.
14646 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
14648 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
14650 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
14652 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
14654 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
14656 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
14658 (hash_tree): Do not hash DECL_FCONTEXT
14659 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
14660 Do not stream DECL_FCONTEXT.
14661 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
14662 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
14664 2018-07-12 Richard Biener <rguenther@suse.de>
14667 * dwarf2out.c (gen_block_die): Only output blocks when they have
14668 at least one !DECL_IGNORED_P variable.
14670 2018-07-12 Richard Biener <rguenther@suse.de>
14673 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
14674 Remove -mieee-fp handling.
14676 2018-07-12 Richard Biener <rguenther@suse.de>
14678 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
14679 left-over from last patch.
14681 2018-07-12 Jakub Jelinek <jakub@redhat.com>
14683 PR tree-optimization/86492
14684 * gimple-ssa-store-merging.c
14685 (imm_store_chain_info::coalesce_immediate_stores): Call
14686 check_no_overlap even for the merge_overlapping case. Formatting fix.
14688 2018-07-12 Richard Biener <rguenther@suse.de>
14690 PR middle-end/86479
14691 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
14692 move possibly trapping operations into the conditional.
14694 2018-07-12 Richard Biener <rguenther@suse.de>
14696 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
14697 (vn_lookup_simplify_result): Remove recursion limit applied
14699 (vn_nary_build_or_lookup_1): Adjust.
14700 (try_to_simplify): Likewise.
14701 * gimple-match-head.c (gimple_resimplify1): Instead apply one
14703 (gimple_resimplify2): Likewise.
14704 (gimple_resimplify3): Likewise.
14705 (gimple_resimplify4): Likewise.
14707 2018-07-11 Jakub Jelinek <jakub@redhat.com>
14709 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
14710 Use __mmask64 type instead of __mmask8 for __M argument.
14711 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
14712 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
14714 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
14715 __mmask16 for __M argument.
14716 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
14717 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
14718 to __mmask16 instead of __mmask8.
14719 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
14720 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
14721 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
14722 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
14723 instead of __mmask16 for __U argument.
14724 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
14725 __mmask16 instead of __mmask8 for __U argument.
14726 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
14728 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
14730 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
14732 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
14734 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
14736 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
14738 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
14740 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
14742 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
14744 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
14746 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
14747 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
14748 return type as well as __M argument type and all casts from __mmask8
14750 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
14751 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
14752 return type as well as __M argument type and all casts from __mmask8
14754 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
14755 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
14756 return type as well as __M argument type and all casts from __mmask8
14758 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
14759 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
14760 return type as well as __M argument type and all casts from __mmask8
14762 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
14763 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
14766 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
14768 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
14769 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
14772 2018-07-11 Paul Koning <ni1d@arrl.net>
14774 * doc/md.texi (define_subst): Document how multiple occurrences of
14775 the same argument in the replacement pattern are handled.
14777 2018-07-11 Paul Koning <ni1d@arrl.net>
14779 * doc/extend.texi (Common Variable Attributes): Move "mode" into
14780 alphabetical order.
14781 (Common Type Attributes): Add "mode" attribute.
14783 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
14785 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
14786 stream DECL_ORIGINAL_TYPE.
14787 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
14788 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
14789 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
14790 Do not walk original type.
14791 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
14793 (write_ts_decl_non_common_tree_pointers): Do not stream
14795 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
14796 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
14798 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
14800 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
14801 thread twice from the same starting edge.
14803 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
14805 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
14806 * gimple.c (gimple_call_nonnull_result_p): ...here...
14807 (gimple_call_nonnull_arg): ...and here.
14808 * gimple.h (gimple_call_nonnull_result_p): New.
14809 (gimple_call_nonnull_arg): New.
14811 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
14813 * config/arm/arm-cpus.in: Move information from fpu field of each
14814 cpu definition to the isa field.
14815 * config/arm/parsecpu.awk (fpu): Delete match rule.
14816 (gen_comm_data): Don't add bits from the CPU's FPU entry.
14818 2018-07-11 Richard Biener <rguenther@suse.de>
14821 * dwarf2out.c (init_sections_and_labels): Use
14822 output_asm_line_debug_info consistently.
14823 (dwarf2out_early_finish): Likewise.
14824 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
14827 2018-07-11 Richard Biener <rguenther@suse.de>
14829 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
14830 Rework father_bb setting in a way to avoid propagating constants
14831 multiple times on a loop body.
14833 2018-07-10 Mark Wielaard <mark@klomp.org>
14836 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
14838 2018-07-10 Richard Biener <rguenther@suse.de>
14840 * hash-map.h (hash_map::iterator::operator*): Return
14841 references to key and value.
14843 2018-07-10 Jakub Jelinek <jakub@redhat.com>
14846 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
14847 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
14848 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
14849 (gimplify_omp_for): For composite loops, move outer
14850 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
14851 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
14852 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
14853 TREE_LIST for both the original class iterator and the "last" helper
14854 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
14855 loop, remember has_decl_expr from outer composite loops for the
14856 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
14858 2018-07-09 Martin Sebor <msebor@redhat.com>
14860 PR middle-end/77357
14861 PR middle-end/86428
14862 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
14863 accessing implicitly initialized array elements.
14864 * expr.c (string_constant): Handle string initializers of
14865 character arrays within aggregates.
14866 * gimple-fold.c (fold_array_ctor_reference): Add argument.
14867 Store element offset. As a special case, handle zero size.
14868 (fold_nonarray_ctor_reference): Same.
14869 (fold_ctor_reference): Add argument. Store subobject offset.
14870 * gimple-fold.h (fold_ctor_reference): Add argument.
14872 2018-07-09 Paul Koning <ni1d@arrl.net>
14874 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
14875 (pdp11_insn_cost): New function.
14876 (pdp11_md_asm_adjust): New function.
14877 (TARGET_INVALID_WITHIN_DOLOOP): Define.
14878 (pdp11_rtx_costs): Update to match machine better.
14879 (output_addr_const_pdp11): Correct format mismatch warnings.
14880 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
14881 * config/pdp11/pdp11.md: General change to add base_cost and/or
14882 length attributes for use by new pdp11_insn_cost function.
14883 (MIN_BRANCH): Correct definition.
14885 (doloop_end): Use standard pattern name for looping pattern.
14886 (doloop_end_nocc): New.
14887 (movsf): Add another constraint alternative.
14888 (zero_extendqihi2): Add constraint alternatives for not in place
14890 (zero_extendhisi2): Remove.
14891 (shift patterns): Add CC handling variants.
14895 (define_peephole2): New peephole to recognize mov that sets CC for
14898 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14900 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
14901 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
14904 2018-07-09 Jakub Jelinek <jakub@redhat.com>
14907 * real.c (real_nextafter): Return true if result is denormal.
14909 2018-07-09 Martin Liska <mliska@suse.cz>
14911 * common.opt: Add back wrongly removed attribute.
14913 2018-07-09 Richard Biener <rguenther@suse.de>
14916 * dwarf2out.c (gen_block_die): For an early generated DIE
14917 always output high/low PC attributes.
14919 2018-07-09 Tom de Vries <tdevries@suse.de>
14921 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
14922 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
14923 onto VAR_DECL with abstract origin.
14925 2018-07-07 Jim Wilson <jimw@sifive.com>
14927 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
14929 2018-07-07 Tom de Vries <tdevries@suse.de>
14931 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
14933 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
14935 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
14937 (combine_bound): Use wide-int overflow calculation instead of
14939 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
14941 * fold-const.c (int_const_binop_2): Same.
14942 (extract_muldiv_1): Same.
14943 (fold_div_compare): Same.
14944 (fold_abs_const): Same.
14946 * poly-int.h (add): Same.
14950 * predict.c (predict_iv_comparison): Same.
14951 * profile-count.c (slow_safe_scale_64bit): Same.
14952 * simplify-rtx.c (simplify_const_binary_operation): Same.
14953 * tree-chrec.c (tree_fold_binomial): Same.
14954 * tree-data-ref.c (split_constant_offset_1): Same.
14955 * tree-if-conv.c (idx_within_array_bound): Same.
14956 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
14957 * tree-ssa-phiopt.c (minmax_replacement): Same.
14958 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
14959 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
14960 * vr-values.c (vr_values::adjust_range_with_scev): Same.
14961 * wide-int.cc (wi::add_large): Same.
14962 (wi::mul_internal): Same.
14963 (wi::sub_large): Same.
14964 (wi::divmod_internal): Same.
14965 * wide-int.h: Change overflow type to overflow_type for neg, add,
14966 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
14967 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
14968 mul_internal, divmod_internal.
14969 (overflow_type): New enum.
14970 (accumulate_overflow): New.
14972 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
14974 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
14975 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
14977 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
14979 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
14980 argument is checked for zero before entering loop, avoid checking again.
14982 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
14984 * gimplify.h (generic_expr_could_trap_p): Set as global function.
14985 * gimplify.h (generic_expr_could_trap_p): Likwise.
14986 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
14988 2018-07-06 Jakub Jelinek <jakub@redhat.com>
14990 PR tree-optimization/86401
14991 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
14992 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
14993 (fold_bit_and_mask): ... here. New helper function for match.pd.
14994 * fold-const.h (fold_bit_and_mask): Declare.
14995 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
14997 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
15000 * target.def (translate_mode_attribute): New hook.
15001 * targhooks.h (default_translate_mode_attribute): Declare.
15002 * targhooks.c (default_translate_mode_attribute): New function.
15003 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
15004 * doc/tm.texi: Regenerate.
15005 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
15006 (rs6000_translate_mode_attribute): New function.
15008 2018-07-06 Paul Koning <ni1d@arrl.net>
15010 * doc/md.texi (define_split): Document DONE and FAIL.
15011 (define_peephole2): Ditto.
15013 2018-07-05 Jeff Law <law@redhat.com>
15015 PR tree-optimization/86010
15016 * tree-ssa-dse.c (compute_trims): More aggressively trim at
15017 both the head and tail of mem* and str* calls.
15019 2018-07-05 Jim Wilson <jimw@sifive.com>
15021 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
15023 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
15025 * config/aarch64/aarch64-simd.md: correct flags text for
15026 MIN_EXPR replacement.
15028 2018-07-05 James Clarke <jrtc27@jrtc27.com>
15030 * configure: Regenerated.
15032 2018-07-05 Carl Love <cel@us.ibm.com>
15034 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
15035 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
15036 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
15037 VSX_BUILTIN_DOUBLEL_V4SF.
15039 2018-07-05 Martin Sebor <msebor@redhat.com>
15042 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
15043 than its domain to compute its the upper bound of a char array.
15045 2018-07-05 Nathan Sidwell <nathan@acm.org>
15047 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
15048 * doc/cpp.texi: Update comment.
15049 * doc/tm.texi: Rebuilt.
15050 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
15051 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
15052 * doc/extend.texi (Backwards Compatibility): Clarify it is system
15053 headers affected by extern "C".
15054 * system.h: Poison NO_IMPLICIT_EXTERN_C.
15055 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
15056 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
15057 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
15058 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
15059 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
15060 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
15061 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
15062 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
15063 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
15064 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
15065 config/sparc/sp64-elf.h, config/spu/spu.h,
15066 config/stormy16/stormy16.h, config/v850/v850.h,
15067 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
15068 define NO_IMPLICIT_EXTERN_C.
15069 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
15071 2018-07-05 Tamar Christina <tamar.christina@arm.com>
15074 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
15075 instead of GET_MODE_SIZE when comparing Units.
15077 2018-07-05 Tamar Christina <tamar.christina@arm.com>
15080 * rtlanal.c (set_noop_p): Constrain on mode change,
15081 include hard-reg-set.h
15083 2018-07-05 Tamar Christina <tamar.christina@arm.com>
15085 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
15087 2018-07-05 Jakub Jelinek <jakub@redhat.com>
15090 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
15093 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
15095 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
15097 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
15100 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
15102 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
15104 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
15106 PR middle-end/86380
15107 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
15109 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
15111 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
15112 neg_*_op* variables.
15114 2018-07-04 Martin Liska <mliska@suse.cz>
15116 * tree-switch-conversion.c: Define
15117 max_ratio_for_speed and max_ratio_for_size constants.
15119 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
15120 Martin Liska <mliska@suse.cz>
15122 PR middle-end/66240
15125 * common.opt: Rename align options with 'str_' prefix.
15126 * common/config/i386/i386-common.c (set_malign_value): New
15128 (ix86_handle_option): Use it to set -falign-* options/
15129 * config/aarch64/aarch64-protos.h (struct tune_params): Change
15130 type from int to string.
15131 * config/aarch64/aarch64.c: Update default values from int
15133 * config/alpha/alpha.c (alpha_override_options_after_change):
15135 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
15136 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
15137 max skip conditionally.
15138 * config/i386/freebsd.h (SUBALIGN_LOG): New.
15139 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
15140 max skip conditionally.
15141 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
15142 max skip conditionally.
15143 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
15144 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
15145 max skip conditionally.
15146 * config/i386/i386.c (struct ptt): Change type from int to
15148 (ix86_default_align): Set default values.
15149 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
15150 max skip conditionally.
15151 * config/i386/iamcu.h (SUBALIGN_LOG): New.
15152 (ASM_OUTPUT_MAX_SKIP_ALIGN):
15153 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
15154 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
15155 max skip conditionally.
15156 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
15157 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
15158 * config/i386/x86-64.h (SUBALIGN_LOG): New.
15159 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
15160 max skip conditionally.
15161 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
15162 * config/ia64/ia64.c (ia64_option_override): Set default values
15163 for alignment options.
15164 * config/m68k/m68k.c: Handle new str_align_* options.
15165 * config/mips/mips.c (mips_set_compression_mode): Change
15167 (mips_option_override): Set default values for options.
15168 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
15170 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15172 * config/rx/rx.c (rx_option_override): Likewise.
15173 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
15174 (LABEL_ALIGN): Use align_labels_log.
15175 (LOOP_ALIGN): Use align_loops_align.
15176 * config/s390/s390.c (s390_asm_output_function_label): Use new
15178 * config/sh/sh.c (sh_override_options_after_change):
15179 Change type of constants.
15180 * config/spu/spu.c (spu_sched_init): Likewise.
15181 * config/sparc/sparc.c (sparc_option_override): Set default
15182 values for options.
15183 * config/visium/visium.c (visium_option_override): Likewise.
15184 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
15185 emit p2align format with last argument if it's not needed.
15186 * doc/invoke.texi: Document extended format of -falign-*.
15187 * final.c: Use align_labels alignment.
15188 * flags.h (struct target_flag_state): Change type to use
15190 (struct align_flags_tuple): New.
15191 (struct align_flags): Likewise.
15192 (align_loops_log): Redefine macro to use new types.
15193 (align_loops_max_skip): Redefine macro to use new types.
15194 (align_jumps_log): Redefine macro to use new types.
15195 (align_jumps_max_skip): Redefine macro to use new types.
15196 (align_labels_log): Redefine macro to use new types.
15197 (align_labels_max_skip): Redefine macro to use new types.
15198 (align_functions_log): Redefine macro to use new types.
15199 (align_loops): Redefine macro to use new types.
15200 (align_jumps): Redefine macro to use new types.
15201 (align_labels): Redefine macro to use new types.
15202 (align_functions): Redefine macro to use new types.
15203 (align_functions_max_skip): Redefine macro to use new types.
15204 (align_loops_value): New macro.
15205 (align_jumps_value): New macro.
15206 (align_labels_value): New macro.
15207 (align_functions_value): New macro.
15208 * function.c (invoke_set_current_function_hook): Propagate
15209 alignment values from flags to global variables default in
15211 * ipa-icf.c (sem_function::equals_wpa): Use
15212 cl_optimization_option_eq instead of memcmp.
15213 * lto-streamer.h (cl_optimization_stream_out): Support streaming
15215 (cl_optimization_stream_in): Likewise.
15216 * optc-save-gen.awk: Support strings in cl_optimization.
15217 * opth-gen.awk: Likewise.
15218 * opts.c (finish_options): Remove error checking of invalid
15220 (MAX_CODE_ALIGN): Remove.
15221 (MAX_CODE_ALIGN_VALUE): Likewise.
15222 (parse_and_check_align_values): New function.
15223 (check_alignment_argument): Likewise.
15224 (common_handle_option): Use check_alignment_argument.
15225 * opts.h (parse_and_check_align_values): Declare.
15226 * toplev.c (init_alignments): Remove.
15227 (read_log_maxskip): New.
15228 (parse_N_M): Likewise.
15229 (parse_alignment_opts): Likewise.
15230 (backend_init_target): Remove usage of init_alignments.
15231 * toplev.h (parse_alignment_opts): Declare.
15232 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
15234 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
15235 * tree.c (cl_option_hasher::equal): New.
15236 * varasm.c: Use new global macros.
15238 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
15240 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
15241 Use a simpler align directive also if MAXSKIP = ALIGN-1.
15242 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15243 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15244 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15245 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
15246 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
15247 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
15248 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
15249 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15250 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15251 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15252 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15254 2018-07-04 Martin Liska <mliska@suse.cz>
15255 Jonathan Wakely <jwakely@redhat.com>
15257 * coverage.c: Use correct type.
15258 * doc/invoke.texi: Language correction.
15260 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
15263 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
15264 ENDBRANCH for non-tail call which may return via indirect branch.
15265 * doc/extend.texi: Document indirect_return attribute.
15267 2018-07-03 Martin Sebor <msebor@redhat.com>
15269 PR tree-optimization/86274
15270 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
15272 (format_floating): Correct handling of infinities and NaNs.
15274 2018-07-03 Martin Sebor <msebor@redhat.com>
15276 * print-tree.c (print_real_cst): New function.
15277 (print_node_brief): Call it.
15278 (print_node): Ditto.
15280 2018-07-03 Jeff Law <law@redhat.com>
15282 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
15283 into a single pattern.
15285 * config/h8300/h8300.md (ors code_iterator): New.
15286 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
15287 a single pattern and single splitter.
15288 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
15289 (iorqi3_1, xorqi3_1): Likewise.
15290 (iorqi3, xorqi3 expanders): Similarly.
15292 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
15293 (movmd_internal) into a single pattern using the P mode iterator.
15294 (movmd splitters): Similarly.
15295 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
15296 (movsd splitters): Similarly.
15298 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
15299 ADDB, ADDW and ADDL into a single ADD attribute which selects the
15300 right table based on the size of the operand.
15301 * config/h8300/h8300.md (length_table): Corresponding changes. All
15302 references to "addb", "addw" and "addl" changed to "add".
15303 (btst patterns): Merge two variants into a single pattern.
15304 (tstqi, tsthi): Likewise.
15305 (addhi3_incdec, addsi3_incdec): Likewise.
15306 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
15307 (mulhi3, mulsi3): Likewise.
15308 (udivhi3, udivsi3): Likewise.
15309 (divhi3, divsi3): Likewise.
15310 (andorqi3, andorhi3, andorsi3): Likewise.
15312 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
15315 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
15316 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
15318 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
15320 PR tree-optimization/85694
15321 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
15322 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
15324 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
15325 (<u>avg<mode>3_ceil): New patterns.
15327 2018-07-03 David Malcolm <dmalcolm@redhat.com>
15329 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
15330 scan-tree-dump directive.
15331 * gcc.dg/vect/slp-perm-2.c: Likewise.
15332 * gcc.dg/vect/slp-perm-3.c: Likewise.
15333 * gcc.dg/vect/slp-perm-5.c: Likewise.
15334 * gcc.dg/vect/slp-perm-6.c: Likewise.
15335 * gcc.dg/vect/slp-perm-7.c: Likewise.
15336 * gcc.dg/vect/slp-perm-8.c: Likewise.
15338 2018-07-03 Marek Polacek <polacek@redhat.com>
15340 PR middle-end/86202
15341 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
15343 2018-07-03 Richard Biener <rguenther@suse.de>
15346 * tree-ssa-structalias.c (find_func_clobbers): Properly
15347 handle indirect calls.
15349 2018-07-03 Jeff Law <law@redhat.com>
15351 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
15352 (shifts): New code iterator.
15353 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
15354 expander. Fix HImode handling on H8/SX.
15355 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
15356 (subqi3, subhi3, subsi3 expanders): Likewise.
15357 (andqi3, andhi3, andsi3 expanders): Likewise.
15358 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
15359 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
15360 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
15361 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
15362 (zero_extendqihi2, zero_extendqisi2): Likewise.
15363 (extendqihi2, extendqisi2): Likewise.
15364 (rotlqi3, rotlhi3, rotlsi3): Likewise.
15365 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
15366 (rotlqi3_1, rotlhi3_1): Likewise.
15367 (logicalhi3_sn, logicalsi3_sn): Likewise.
15368 (logicalhi3, logicalsi3): Likewise.
15370 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
15372 * tree-vect-patterns.c (vect_recog_rotate_pattern)
15373 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
15374 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
15375 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
15376 type to append_pattern_def_seq instead of creating a stmt_vec_info
15378 (build_mask_conversion): Likewise. Remove vinfo argument.
15379 (vect_add_conversion_to_patterm): Likewise, renaming to...
15380 (vect_add_conversion_to_pattern): ...this.
15381 (vect_recog_mask_conversion_pattern): Update call to
15382 build_mask_conversion. Pass the vector type to
15383 append_pattern_def_seq here too.
15384 (vect_recog_gather_scatter_pattern): Update call to
15385 vect_add_conversion_to_pattern.
15387 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
15389 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
15390 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
15391 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
15392 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
15393 STMT_VINFO_PATTERN_DEF_SEQ to null here.
15394 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
15395 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
15396 append_pattern_def_seq instead of new_pattern_def_seq.
15397 (vect_recog_divmod_pattern): Do both of the above.
15398 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
15401 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
15403 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
15404 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
15405 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
15406 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
15407 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
15408 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
15409 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
15410 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
15411 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
15412 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
15413 parameter with a single stmt_vec_info.
15414 (vect_recog_func_ptr): Likewise.
15415 (vect_recog_gather_scatter_pattern): Likewise, folding in...
15416 (vect_try_gather_scatter_pattern): ...this.
15417 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
15418 the stmt_vec_info of the statement to be matched. Don't clear
15419 STMT_VINFO_RELATED_STMT.
15420 (vect_pattern_recog): Update call accordingly.
15422 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
15424 PR tree-optimization/85694
15425 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
15426 (uavgM3_ceil): Document new optabs.
15427 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
15428 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
15430 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
15431 (savg_ceil_optab): New optabs.
15432 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
15433 (vect_vect_recog_func_ptrs): Add it.
15434 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
15435 constant directly from the associated lhs.
15437 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
15439 * tree-vect-patterns.c (vect_split_statement): New function.
15440 (vect_convert_input): Use it to try to split an existing cast.
15442 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
15444 * poly-int.h (print_hex): New function.
15445 * dumpfile.h (dump_dec, dump_hex): Declare.
15446 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
15447 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
15448 min_input_precision, operation_precision and operation_sign.
15449 * tree-vect-patterns.c (vect_get_range_info): New function.
15450 (vect_same_loop_or_bb_p, vect_single_imm_use)
15451 (vect_operation_fits_smaller_type): Delete.
15452 (vect_look_through_possible_promotion): Add an optional
15453 single_use_p parameter.
15454 (vect_recog_over_widening_pattern): Rewrite to use new
15455 stmt_vec_info infomration. Handle one operation at a time.
15456 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
15457 (vect_truncatable_operation_p, vect_set_operation_type)
15458 (vect_set_min_input_precision): New functions.
15459 (vect_determine_min_output_precision_1): Likewise.
15460 (vect_determine_min_output_precision): Likewise.
15461 (vect_determine_precisions_from_range): Likewise.
15462 (vect_determine_precisions_from_users): Likewise.
15463 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
15464 (vect_vect_recog_func_ptrs): Put over_widening first.
15466 (vect_pattern_recog): Call vect_determine_precisions.
15468 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
15470 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
15471 statements that have been replaced by further pattern statements.
15472 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
15474 2018-07-03 Richard Biener <rguenther@suse.de>
15476 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
15477 always set *dt. Dump vectype in vectype overload.
15478 * dumpfile.h (dump_gimple_expr): New function.
15479 (dump_gimple_expr_loc): Likewise.
15480 * dumpfile.c (dump_gimple_expr): New function.
15481 (dump_gimple_expr_loc): Likewise.
15483 2018-07-02 Jeff Law <law@redhat.com>
15485 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
15486 the H8/300, H8/300H and H8/S variants into a single pattern.
15487 (movhi_h8300, movqi_h8300hs): Similarly.
15488 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
15489 (QHI mode iterator): New.
15491 * config/h8300/h8300.md: Remove trailing whitespace.
15493 2018-07-02 Jim Wilson <jimw@sifive.com>
15495 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
15496 instead of emit_insn for interrupt returns.
15497 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
15498 (riscv_sret, riscv_uret): Likewise.
15500 2018-07-02 David Malcolm <dmalcolm@redhat.com>
15502 * pretty-print.c (selftest::test_pp_format): Move save and restore
15503 of quotes to class auto_fix_quotes, and add an instance.
15504 * selftest.c: Include "intl.h".
15505 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
15506 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
15507 * selftest.h (selftest::auto_fix_quotes): New class.
15509 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
15511 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
15512 (aarch64_sve_prepare_conditional_op): Remove.
15513 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
15514 Allow aarch64_simd_reg_or_zero as select operand; remove
15515 the aarch64_sve_prepare_conditional_op call.
15516 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
15517 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
15518 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
15519 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
15520 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
15521 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
15522 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
15523 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
15524 and a splitters to match all of the *_any patterns.
15525 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
15527 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
15528 (SVE_COND_FP_BINARY_REV): Remove.
15529 (sve_int_op_rev, sve_fp_op_rev): New.
15530 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
15531 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
15532 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
15533 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
15534 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
15535 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
15536 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
15537 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
15538 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
15540 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
15541 Remove match_dup 1 from the inner unspec.
15542 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
15544 * config/aarch64/aarch64.md (movprfx): New attr.
15545 (length): Default movprfx to 8.
15546 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
15547 (*madd<SVE_I>, *msub<SVE_I): Likewise.
15548 (*<su>mul<SVE_I>3_highpart): Likewise.
15549 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
15550 (*v<ASHIFT><SVE_I>3): Likewise.
15551 (*<su><MAXMIN><SVE_I>3): Likewise.
15552 (*<su><MAXMIN><SVE_F>3): Likewise.
15553 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
15554 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
15555 (*div<SVE_F>4): Likewise.
15557 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
15559 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
15562 2018-07-02 Richard Biener <rguenther@suse.de>
15564 PR tree-optimization/86363
15565 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
15566 memset argument refers to a non-variable address.
15568 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
15570 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
15571 {PLUS,MINUS}_EXPR code to...
15572 (adjust_symbolic_bound): ...here,
15573 (combine_bound): ...here,
15574 (set_value_range_with_overflow): ...and here.
15576 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
15578 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
15580 (extract_range_from_abs_expr): ...here.
15582 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
15584 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
15585 -fno-omit-frame-pointer when not optimizing.
15587 2018-07-02 Martin Liska <mliska@suse.cz>
15590 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
15591 (propagate_nothrow): Likewise.
15593 2018-07-02 Martin Liska <mliska@suse.cz>
15596 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
15598 2018-07-02 David Malcolm <dmalcolm@redhat.com>
15600 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
15601 function in r262149, changing "loc" param from source_location to
15602 const dump_location_t &.
15603 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
15604 declaration, as above.
15606 2018-07-01 Paul Koning <ni1d@arrl.net>
15608 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
15609 -munit-asm, -mgnu-asm, -mdec-asm.
15610 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
15611 (pdp11_output_labelref): New.
15612 (pdp11_output_def): New.
15613 (pdp11_output_addr_vec_elt): New.
15614 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
15615 %# and %@ format codes.
15616 (pdp11_option_override): New.
15617 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
15618 (pdp11_output_ident): New.
15619 (pdp11_asm_named_section): New.
15620 (pdp11_asm_init_sections): New.
15621 (pdp11_file_start): New.
15622 (pdp11_file_end): New.
15623 (output_ascii): Use .ascii/.asciz for -mdec-asm.
15624 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
15625 %o, like %c but octal.
15626 (pdp11_option_override): New.
15627 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
15629 (DATA_SECTION_ASM_OP): Ditto.
15630 (READONLY_DATA_SECTION_ASM_OP): New.
15631 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
15632 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
15633 (ASM_OUTPUT_LABELREF): Ditto.
15634 (ASM_OUTPUT_DEF): Ditto.
15635 (ASM_OUTPUT_EXTERNAL): New.
15636 (ASM_OUTPUT_SOURCE_FILENAME): New.
15637 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
15638 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
15639 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
15640 %# and %@ format codes.
15641 * config/pdp11/pdp11.opt (mgnu-asm): New.
15642 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
15643 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
15644 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
15646 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
15648 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
15649 dereferencing path[] beyond its length.
15651 (debug_all_paths): New.
15652 (rewire_first_differing_edge): New.
15653 (adjust_paths_after_duplication): New.
15654 (duplicate_thread_path): Call adjust_paths_after_duplication.
15656 (thread_through_all_blocks): Add new argument to
15657 duplicate_thread_path.
15659 2018-06-30 Jim Wilson <jimw@sifive.com>
15661 * config/riscv/predicates.md (p2m1_shift_operand): New.
15662 (high_mask_shift_operand): New.
15663 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
15664 pattern using p2m1_shift_operand.
15665 (lshsi3_zero_extend_3+2): New combiner pattern using
15666 high_mask_shift_operand.
15668 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
15670 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
15672 (vect_recog_rotate_pattern): ...here.
15673 (vect_convert_input): Try to insert casts of invariants in the
15675 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
15676 preheader to be empty.
15678 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
15680 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
15681 vector type. If given, install it in the new statement's
15682 STMT_VINFO_VECTYPE.
15683 (vect_element_precision): New function.
15684 (vect_unpromoted_value): New struct.
15685 (vect_unpromoted_value::vect_unpromoted_value): New function.
15686 (vect_unpromoted_value::set_op): Likewise.
15687 (vect_look_through_possible_promotion): Likewise.
15688 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
15689 (vect_widened_op_tree, vect_convert_input): Likewise.
15690 (vect_convert_inputs, vect_convert_output): Likewise.
15691 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
15692 to handle the optional cast of the multiplication result and
15693 vect_widened_op_tree to detect the widened multiplication itself.
15694 Do not require the input and output of promotion casts to have
15695 the same sign, but base the signedness of the operation on the
15696 input rather than the result. If the pattern includes two
15697 promotions, check that those promotions have the same sign.
15698 Do not restrict the MULT_EXPR handling to a double-width result;
15699 handle quadruple-width results and wider. Use vect_convert_inputs
15700 to convert the inputs to the common type.
15701 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
15702 to handle the optional cast of the ABS result. Also allow a sign
15703 change or a sign extension between the ABS and MINUS.
15704 Use vect_widened_op_tree to detect the widened subtraction and use
15705 vect_convert_inputs to convert the inputs to the common type.
15706 (vect_handle_widen_op_by_const): Delete.
15707 (vect_recog_widen_op_pattern): New function.
15708 (vect_recog_widen_mult_pattern): Use it.
15709 (vect_recog_widen_shift_pattern): Likewise.
15710 (vect_recog_widen_sum_pattern): Use
15711 vect_look_through_possible_promotion to handle the promoted
15714 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
15716 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
15717 the containing gimple_seq *.
15718 * gimple-iterator.h (gsi_for_stmt): Declare it.
15719 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
15720 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
15721 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
15722 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
15723 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
15725 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
15727 (vect_mark_pattern_stmts): ...here. Handle cases in which the
15728 statement being replaced is part of an existing pattern
15729 definition sequence, inserting the new pattern statements before
15731 (vect_pattern_recog_1): Don't return a bool. If the statement
15732 is already part of a pattern, instead apply pattern matching
15733 to the pattern definition statements. Don't clear the
15734 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
15735 (vect_pattern_recog): Don't break after the first match;
15736 continue processing the pattern definition statements instead.
15737 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
15739 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
15741 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
15742 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
15743 (vect_recog_widen_sum_pattern): Use it.
15745 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
15747 * tree-vect-loop.c (vectorizable_reduction): Assert that the
15748 phi is not a pattern statement and has not been replaced by
15749 a pattern statement.
15750 * tree-vect-patterns.c (type_conversion_p): Don't check
15751 STMT_VINFO_IN_PATTERN_P.
15752 (vect_recog_vector_vector_shift_pattern): Likewise.
15753 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
15754 the pattern statement rather than the original statement; check
15755 directly for a WIDEN_MULT_EXPR here.
15756 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
15757 vect_is_simple_use to return the pattern statement rather
15758 than the original statement; use is_pattern_stmt_p to check
15759 for such a pattern statement.
15760 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
15761 to return the pattern statement rather than the original statement;
15762 don't do the same transformation here.
15763 (vect_is_simple_use): If the defining statement has been replaced
15764 by a pattern statement, return the pattern statement instead.
15765 Remove the corresponding (local) transformation from the vectype
15768 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
15770 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
15771 end and default to null.
15772 * tree-vect-loop.c (vect_create_epilog_for_reduction)
15773 (vectorizable_reduction): Update calls accordingly, dropping the
15774 gimple ** argument if the passed-back statement isn't needed.
15775 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
15776 (vect_recog_rotate_pattern): Likewise.
15777 (vect_recog_mask_conversion_pattern): Likewise.
15778 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
15779 (vect_mask_constant_operand_p): Likewise.
15780 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
15781 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
15782 (get_group_load_store_type, get_load_store_type): Likewise.
15783 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
15784 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
15785 (vectorizable_conversion, vectorizable_assignment): Likewise.
15786 (vectorizable_shift, vectorizable_operation): Likewise.
15787 (vectorizable_store, vect_is_simple_cond): Likewise.
15788 (vectorizable_condition, vectorizable_comparison): Likewise.
15789 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
15790 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
15791 and move it to the end. Cope with null def_stmt_outs.
15793 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
15795 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
15797 2018-06-29 Jeff Law <law@redhat.com>
15799 * config/v850/v850.c (v850_legitimate_address_p): Handle large
15800 displacements for TARGET_V850E2V3 and newer.
15801 (TARGET_LRA_P): Remove. Defaults to LRA now.
15802 * config/v850/v850.md (sign23byte_load): Remove.
15803 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
15804 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
15806 2018-06-29 Martin Liska <mliska@suse.cz>
15809 * coverage.c (coverage_init): Mangle full path name.
15810 * doc/invoke.texi: Document the change.
15811 * gcov-io.c (mangle_path): New.
15812 * gcov-io.h (mangle_path): Likewise.
15813 * gcov.c (mangle_name): Use mangle_path for path mangling.
15815 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15817 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
15818 if starting source register is not even.
15820 2018-06-29 Martin Liska <mliska@suse.cz>
15822 PR tree-optimization/86263
15823 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
15824 Make edge redirection.
15826 2018-06-29 David Malcolm <dmalcolm@redhat.com>
15828 * dumpfile.c (dump_loc): Add indentation based on scope depth.
15829 (dump_scope_depth): New variable.
15830 (get_dump_scope_depth): New function.
15831 (dump_begin_scope): New function.
15832 (dump_end_scope): New function.
15833 * dumpfile.h (get_dump_scope_depth): New declaration.
15834 (dump_begin_scope): New declaration.
15835 (dump_end_scope): New declaration.
15836 (class auto_dump_scope): New class.
15837 (AUTO_DUMP_SCOPE): New macro.
15838 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
15841 2018-06-29 Richard Biener <rguenther@suse.de>
15843 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
15844 compute_all_dependences succeeds.
15845 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
15846 exceed --param loop-max-datarefs-for-datadeps.
15848 2018-06-29 Jakub Jelinek <jakub@redhat.com>
15850 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
15852 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
15855 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
15856 alternative 0 in preferred_for_speed attribute.
15858 2018-06-28 Paul Koning <ni1d@arrl.net>
15860 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
15861 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
15862 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
15863 * config/pdp11/pdp11.md: Correct "length" attribute calculation
15864 for shift insn patterns.
15866 2018-06-28 David Malcolm <dmalcolm@redhat.com>
15868 * cgraph.c (cgraph_node::get_body): Replace assignments to
15869 "dump_file" with calls to set_dump_file.
15870 * dumpfile.c (alt_dump_file): Make static, and group with...
15871 (alt_flags): ...this definition.
15872 (dumps_are_enabled): New variable.
15873 (refresh_dumps_are_enabled): New function.
15874 (set_dump_file): New function.
15875 (set_alt_dump_file): New function.
15876 (gcc::dump_manager::dump_start): Replace assignments to
15877 "dump_file" and "alt_dump_file" with calls to set_dump_file and
15879 (gcc::dump_manager::dump_finish): Likewise.
15880 * dumpfile.h (alt_dump_file): Delete decl.
15881 (dumps_are_enabled): New variable decl.
15882 (set_dump_file): New function decl.
15883 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
15885 * tree-nested.c (lower_nested_functions): Replace assignments to
15886 "dump_file" with calls to set_dump_file.
15888 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
15890 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
15891 goto_locus of each outgoing edge of each basic block.
15893 2018-06-28 Richard Biener <rguenther@suse.de>
15895 * dwarf2out.c (decl_scope_table): Remove.
15896 (push_decl_scope): Likewise.
15897 (pop_decl_scope): Likewise.
15898 (gen_type_die_for_member): Do not call push/pop_decl_scope.
15899 (gen_struct_or_union_type_die): Likewise.
15900 (gen_tagged_type_die): Likewise.
15901 (dwarf2out_init): Do not initialize decl_scope_table.
15902 (dwarf2out_c_finalize): Do not free it.
15904 2018-06-28 Richard Biener <rguenther@suse.de>
15906 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
15907 deciding whether to not re-use a DIE.
15909 2018-06-28 Richard Biener <rguenther@suse.de>
15911 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
15912 DW_AT_abstract_origin attribute.
15914 2018-06-28 Martin Liska <mliska@suse.cz>
15916 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
15917 Use newly introduced constants.
15918 * tree-switch-conversion.h (struct jump_table_cluster):
15919 Define max_ratio_for_size and max_ratio_for_speed.
15921 2018-06-28 Martin Liska <mliska@suse.cz>
15923 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
15924 Add new checking assert to catch invalid state.
15925 (jump_table_cluster::can_be_handled): Handle single case
15927 (jump_table_cluster::is_beneficial): Bail out for such case.
15928 (bit_test_cluster::find_bit_tests):
15929 Add new checking assert to catch invalid state.
15930 (bit_test_cluster::can_be_handled): Handle single case
15932 (bit_test_cluster::is_beneficial): Bail out for such case.
15933 (switch_decision_tree::analyze_switch_statement):
15936 2018-06-28 Martin Liska <mliska@suse.cz>
15938 * common.opt: Introduce -completion option.
15939 * gcc.c (driver_handle_option): Handle it.
15940 (driver::main): Print completions if completion
15942 * opt-suggestions.c (option_proposer::get_completions):
15944 (option_proposer::suggest_completion): Likewise.
15945 (option_proposer::find_param_completions): Likewise.
15946 (verify_autocompletions): Likewise.
15947 (test_completion_valid_options): Likewise.
15948 (test_completion_valid_params): Likewise.
15949 (in_completion_p): Likewise.
15950 (empty_completion_p): Likewise.
15951 (test_completion_partial_match): Likewise.
15952 (test_completion_garbage): Likewise.
15953 (opt_proposer_c_tests): Likewise.
15954 * opt-suggestions.h: Declare new functions.
15955 * opts.c (common_handle_option): Handle OPT__completion_.
15956 * selftest-run-tests.c (selftest::run_tests): Add
15957 opt_proposer_c_tests.
15958 * selftest.c (assert_str_startswith): New.
15959 * selftest.h (assert_str_startswith): Likewise.
15960 (opt_proposer_c_tests): New.
15961 (ASSERT_STR_STARTSWITH): Likewise.
15963 2018-06-28 Martin Liska <mliska@suse.cz>
15965 * Makefile.in: Add opt-suggestions.o.
15966 * gcc-main.c: Include opt-suggestions.h.
15967 * gcc.c (driver::driver): Likewise.
15968 (driver::~driver): Remove m_option_suggestions.
15969 (driver::build_option_suggestions): Moved to option_proposer.
15970 (driver::suggest_option): Likewise.
15971 (driver::handle_unrecognized_options): Use option_proposer.
15972 * gcc.h (class driver): Add new memver m_option_proposer.
15973 * opt-suggestions.c: New file.
15974 * opt-suggestions.h: New file.
15976 2018-06-28 Martin Liska <mliska@suse.cz>
15978 * vec.h (class auto_string_vec): New (moved from auto_argvec).
15979 (auto_string_vec::~auto_string_vec): Likewise.
15981 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
15983 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
15984 prevent_decl_creation_for_types fields up and add reset_location field.
15985 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
15986 statement if id->reset_location is true.
15987 (copy_edges_for_bb): Do not set goto_locus on the new edges if
15988 id->reset_location is true.
15989 (copy_phis_for_bb): Force input_location on the arguments if
15990 id->reset_location is true.
15991 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
15992 is set on the function to be inlined.
15994 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
15996 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
15998 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
16000 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
16001 registers for Pmode.
16002 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
16003 hard registers for the clobbered pseudo.
16005 2018-06-27 Paul Koning <ni1d@arrl.net>
16007 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
16008 mutually exclusive options.
16009 * config/pdp11/constraints.md (h): New constraint.
16010 (O): Update definition to match shift code generation.
16011 (D): New constraint.
16012 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
16014 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
16016 (output_jump): Change arguments.
16017 (pdp11_fixed_cc_regs): New function.
16018 (pdp11_cc_mode): Ditto.
16019 (pdp11_expand_shift): Ditto.
16020 (pdp11_assemble_shift): Ditto.
16021 (pdp11_small_shift): Ditto.
16022 (pdp11_branch_cost): Remove.
16023 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
16025 (pdp11_register_move_cost): Update for CC registers.
16026 (pdp11_rtx_costs): Add case for LSHIFTRT.
16027 (pdp11_output_jump): Add CCNZ mode conditional branches.
16028 (notice_update_cc_on_set): Remove.
16029 (pdp11_cc_mode): New function.
16030 (simple_memory_operand): Correct pre/post decrement case.
16031 (no_side_effect_operand): New function.
16032 (pdp11_regno_reg_class): Add CC_REGS class.
16033 (pdp11_fixed_cc_regs): New function.
16034 (pdp11_small_shift): New function.
16035 (pdp11_expand_shift): New function to expand shift insns.
16036 (pdp11_assemble_shift): New function to output shifts.
16037 (pdp11_branch_cost): Remove.
16038 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
16039 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
16040 (WCHAR_TYPE): Ditto.
16041 (PTRDIFF_TYPE): Ditto.
16042 (ADJUST_INSN_LENGTH): New macro.
16043 (FIXED_REGISTERS): Add CC registers.
16044 (CALL_USED_REGISTERS): Ditto.
16045 (reg_class): Ditto.
16046 (REG_CLASS_NAMES): Ditto.
16047 (REG_CLASS_CONTENTS): Ditto.
16048 (SELECT_CC_MODE): Use new function.
16049 (TARGET_FLAGS_REGNUM): New macro.
16050 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
16051 (cc0_reg_rtx): Remove.
16052 (CC_STATUS_MDEP): Remove.
16053 (CC_STATUS_MDEFP_INIT): Remove.
16054 (CC_IN_FPU): Remove.
16055 (NOTICE_UPDATE_CC): Remove.
16056 (REGISTER_NAMES): Add CC registers.
16057 (BRANCH_COST): Change to constant 1.
16058 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
16060 * config/pdp11/pdp11.opt (mbcopy): Remove.
16061 (mbcopy-builtin): Remove.
16062 (mbranch-cheap): Remove.
16063 (mbranch-expensive): Remove.
16064 * config/pdp11/predicates.md (expand_shift_operand): Update to
16065 match shift code generation.
16066 (ccnz_operator): New predicate.
16067 * doc/invoke.texi (PDP-11 Options): Remove deleted options
16068 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
16069 Remove non-existent option -mabshi, -mno-abshi. Document mutually
16071 * doc/md.texi (PDP-11): Document new D and h constraints. Update
16072 description of O constraint.
16074 2018-06-27 Jeff Law <law@redhat.com>
16075 Austin Law <austinklaw@gmail.com>
16077 * config/v850/v850.md (addsi3_set_flags): New pattern.
16078 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
16079 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
16080 (zero_extendhisi2_v850_set_flags): Likewise.
16081 (zero_extendqisi2_v850_set_flags): Likewise.
16082 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
16083 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
16084 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
16086 * config/v850/v850-protos.h (notice_update_cc): Remove.
16087 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
16088 (v850_print_operand): Handle 'D' and "d".
16089 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
16090 Add handling of arithmetic/logical operations compared against zero.
16091 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
16092 Do not look at v850_compare_op, instead get mode from last argument.
16093 (v850_gen_compare): Remove
16094 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
16095 after reload for prologue insns.
16096 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
16098 (construct_save_jarl): Likewise.
16099 (TARGET_FLAGS_REGNUM): Define.
16100 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
16101 (NOTICE_UPDATE_CC): Remove.
16102 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
16103 than cc0. Conditionalize on reload_completed.
16104 (cmpsi_insn, setfcc_insn): Likewise.
16105 (tst1 splitter): Turn into define_and_split which sets the flags
16107 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
16108 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
16109 (cstoresf4, cstoredf4): Clobber the flags.
16110 (cmpsi, cmpsf, cmpdf): Remove expanders.
16111 (setf_insn): Remove pattern.
16112 (addsi3): Turn into define_and_split which clobbers the flags after
16113 reload and a suitable pattern (addsi3_clobber_flags) for use after
16115 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
16116 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
16117 (ashrsi3, ashrsi3_v850e2): Likewise.
16118 (bins): Clobber the flags.
16119 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
16120 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
16121 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
16122 (call_value_internal_short, call_value_internal_long): Likewise.
16123 (callt_save_interrupt, callt_return_interrupt): Likewise.
16124 (save_interrupt, return_interrupt): Likewise.
16125 (callt_save_all_interrupt, save_all_interrupt): Likewise.
16126 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
16127 (restore_all_interrupt, _restore_all_interrupt): Likewise.
16128 (All FP comparisons): Only allow after reload has completed.
16130 (divh, divhu): Tweak output template.
16131 (branch_z_normal, branch_z_invert): Remove
16132 (branch_nz_normal, branch_nz_invert): Likewise.
16133 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
16135 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
16136 * config/v850/v850.c (notice_update_cc): Remove.
16137 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
16138 (CC_NO_CARRY): Likewise.
16139 (NOTICE_UPDATE_CC): Define to nothing.
16140 * config/v850/v850.md: Remove block comment on cc0 handling
16141 Remove "cc" attribute from all patterns. Remove cc_status handling
16142 from all patterns. Minor formatting fixes.
16144 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16146 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
16147 (cortex-a76.cortex-a55): Likewise.
16148 * config/aarch64/aarch64-tune.md: Regenerate.
16149 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
16150 cortex-a76.cortex-a55.
16152 2018-06-27 Jeff Law <law@redhat.com>
16154 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
16155 (MULTILIB_DIRNAMES): Similarly.
16157 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
16159 * gimple.h (gimple_return_retbnd): Delete.
16160 (gimple_return_set_retbnd): Likewise.
16161 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
16162 gimple_return_set_retbnd.
16163 * gimple-pretty-print.c (dump_gimple_return): Remove call to
16164 gimple_return_retbnd and adjust.
16165 * tree-inline.h (struct copy_body_data): Remove retbnd field.
16166 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
16167 Explicitly return NULL in a couple more cases. Move assertion
16168 on debug statements and remove unreachable code.
16169 (reset_debug_binding): Do not test id->retbnd.
16170 (expand_call_inline): Do not set it.
16172 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
16174 * configure.ac: Add --disable-gcov option.
16175 * configure: Regenerate.
16176 * Makefile.in: Honour @enable_gcov@.
16177 * doc/install.texi: Document --disable-gcov.
16179 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16181 * config/arm/arm-cpus.in (cortex-a76): New entry.
16182 (cortex-a76.cortex-a55): Likewise.
16183 * config/arm/arm-tables.opt: Regenerate.
16184 * config/arm/arm-tune.md: Likewise.
16185 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
16186 * doc/invoke.texi (ARM Options): Document cortex-a76 and
16187 cortex-a76.cortex-a55.
16189 2018-06-27 Tamar Christina <tamar.christina@arm.com>
16192 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
16194 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
16196 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
16198 (EPILOGUE_USES): Likewise.
16200 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
16202 * tree-inline.c (remap_location): New function extracted from...
16203 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
16204 (copy_phis_for_bb): ...here. Call remap_location.
16205 (copy_cfg_body): Adjust call to copy_edges_for_bb.
16207 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
16209 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
16210 unaligned vsx for 16B memset.
16212 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
16215 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
16216 ieee128_float_type_node to long_double_type_node unless
16217 TARGET_LONG_DOUBLE_128 is set.
16219 2018-06-26 David Malcolm <dmalcolm@redhat.com>
16221 * cfgloop.c (get_loop_location): Convert return type from
16222 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
16223 by implicit construction from rtx_insn *, and using
16224 dump_user_location_t::from_function_decl for the fallback case.
16225 * cfgloop.h (get_loop_location): Convert return type from
16226 location_t to dump_user_location_t.
16227 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
16228 dump_printf_loc to pass in a dump_location_t rather than a
16229 location_t, via the gimple stmt.
16230 * coverage.c (get_coverage_counts): Update calls to
16231 dump_printf_loc to pass in dump_location_t rather than a
16233 * doc/optinfo.texi (Dump types): Convert example of
16234 dump_printf_loc from taking "locus" to taking "insn". Update
16235 description of the "_loc" calls to cover dump_location_t.
16236 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
16238 (dump_user_location_t::dump_user_location_t): New constructors,
16239 from gimple *stmt and rtx_insn *.
16240 (dump_user_location_t::from_function_decl): New function.
16241 (dump_loc): Make static.
16242 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
16243 const dump_location_t &.
16244 (dump_generic_expr_loc): Delete.
16245 (dump_printf_loc): Convert param "loc" from location_t to
16246 const dump_location_t &.
16247 (selftest::test_impl_location): New function.
16248 (selftest::dumpfile_c_tests): New function.
16249 * dumpfile.h: Include "profile-count.h".
16250 (class dump_user_location_t): New class.
16251 (struct dump_impl_location_t): New struct.
16252 (class dump_location_t): New class.
16253 (dump_printf_loc): Convert 2nd param from source_location to
16254 const dump_location_t &.
16255 (dump_generic_expr_loc): Delete.
16256 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
16257 const dump_location_t &.
16258 * gimple-fold.c (fold_gimple_assign): Update call to
16259 dump_printf_loc to pass in a dump_location_t rather than a
16260 location_t, via the gimple stmt.
16261 (gimple_fold_call): Likewise.
16262 * gimple-loop-interchange.cc
16263 (loop_cand::analyze_iloop_reduction_var): Update for change to
16264 check_reduction_path.
16265 (tree_loop_interchange::interchange): Update for change to
16266 find_loop_location.
16267 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
16268 change in return-type of find_loop_location.
16269 (graphite_regenerate_ast_isl): Likewise.
16270 * graphite-optimize-isl.c (optimize_isl): Likewise.
16271 * graphite.c (graphite_transform_loops): Likewise.
16272 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
16273 pass in a dump_location_t rather than a location_t, via the
16275 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
16276 * ipa.c (walk_polymorphic_call_targets): Likewise.
16277 * loop-unroll.c (report_unroll): Convert "locus" param from
16278 location_t to dump_location_t.
16279 (decide_unrolling): Update for change to get_loop_location's
16281 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
16282 location_t to dump_user_location_t.
16283 (grid_find_single_omp_among_assignments_1): Updates calls to
16284 dump_printf_loc to pass in a dump_location_t rather than a
16285 location_t, via the gimple stmt.
16286 (grid_parallel_clauses_gridifiable): Convert "tloc" from
16287 location_t to dump_location_t. Updates calls to dump_printf_loc
16288 to pass in a dump_location_t rather than a location_t, via the
16290 (grid_inner_loop_gridifiable_p): Likewise.
16291 (grid_dist_follows_simple_pattern): Likewise.
16292 (grid_gfor_follows_tiling_pattern): Likewise.
16293 (grid_target_follows_gridifiable_pattern): Likewise.
16294 (grid_attempt_target_gridification): Convert initialization
16295 of local "grid" from memset to zero-initialization; FIXME: does
16296 this require C++11? Update call to dump_printf_loc to pass in a
16297 optinfo_location rather than a location_t, via the gimple stmt.
16298 * profile.c (read_profile_edge_counts): Updates call to
16299 dump_printf_loc to pass in a dump_location_t rather than a
16301 (compute_branch_probabilities): Likewise.
16302 * selftest-run-tests.c (selftest::run_tests): Call
16304 * selftest.h (dumpfile_c_tests): New decl.
16305 * tree-loop-distribution.c (pass_loop_distribution::execute):
16306 Update for change in return type of find_loop_location.
16307 * tree-parloops.c (parallelize_loops): Likewise.
16308 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
16309 "locus" from location_t to dump_user_location_t.
16310 (canonicalize_loop_induction_variables): Likewise.
16311 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
16312 for change in return type of find_loop_location.
16313 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
16314 to dump_printf_loc to pass in a dump_location_t rather than a
16315 location_t, via the stmt.
16316 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
16318 * tree-vect-loop-manip.c (find_loop_location): Convert return
16319 type from source_location to dump_user_location_t.
16320 (vect_do_peeling): Update for above change.
16321 (vect_loop_versioning): Update for change in type of
16323 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
16324 from location_t to dump_user_location_t.
16325 (vect_estimate_min_profitable_iters): Update for change in type
16327 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
16328 location_t to dump_location_t.
16329 (vect_slp_bb): Update for change in type of vect_location.
16330 * tree-vectorizer.c (vect_location): Convert from source_location
16331 to dump_user_location_t.
16332 (try_vectorize_loop_1): Update for change in vect_location's type.
16333 (vectorize_loops): Likewise.
16334 (increase_alignment): Likewise.
16335 * tree-vectorizer.h (vect_location): Convert from source_location
16336 to dump_user_location_t.
16337 (find_loop_location): Convert return type from source_location to
16338 dump_user_location_t.
16339 (check_reduction_path): Convert 1st param from location_t to
16340 dump_user_location_t.
16341 * value-prof.c (check_counter): Update call to dump_printf_loc to
16342 pass in a dump_user_location_t rather than a location_t; update
16343 call to error_at for change in type of "locus".
16344 (check_ic_target): Update call to dump_printf_loc to
16345 pass in a dump_user_location_t rather than a location_t, via the
16348 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
16350 * config/s390/s390.h (enum processor_flags): Do not use
16351 default tune parameter when -march was specified.
16353 2018-06-26 Jakub Jelinek <jakub@redhat.com>
16356 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
16357 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
16360 2018-06-26 Richard Biener <rguenther@suse.de>
16362 PR tree-optimization/86287
16364 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
16365 (vect_analyze_loop): Initialize n_stmts.
16367 2018-06-26 Richard Biener <rguenther@suse.de>
16369 PR middle-end/86271
16370 * fold-const.c (fold_convertible_p): Pointer extension
16373 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
16376 * dwarf2out.c (loc_list_has_views): Adjust comments.
16377 (dw_loc_list): Split single cross-partition range with
16380 2018-06-25 Jeff Law <law@redhat.com>
16382 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
16383 on -mbig-switch by default.
16385 * config/v850/predicates.md (const_float_1_operand): Fix match_code
16387 (const_float_0_operand): Remove unused predicate.
16388 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
16389 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
16390 (recipsf2): New expander. Original pattern now called
16392 (recipdf2, recipdf2_insn): Similarly.
16393 (rsqrtsf2, rsqrtsf2_insn): Similarly
16394 (rsqrtdf2, rsqrtdf2_insn): Similarly
16396 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
16398 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
16399 Simplify logic for FreeBSD (twice).
16401 2018-06-25 Martin Sebor <msebor@redhat.com>
16403 PR tree-optimization/86204
16404 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
16405 a strnlen result if it's less than the length of the string.
16407 2018-06-25 Martin Sebor <msebor@redhat.com>
16409 PR tree-optimization/85700
16410 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
16411 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
16412 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
16414 2018-06-25 Martin Sebor <msebor@redhat.com>
16416 * doc/extend.texi (Zero-length arrays): Update and clarify.
16418 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
16420 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
16421 added IEEE/IBM long double multilib support on PowerPC little
16422 endian Linux systems.
16423 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
16424 (MULTILIB_DEFAULTS): Likewise.
16425 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16427 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
16428 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
16429 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
16431 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
16433 PR middle-end/86311
16434 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
16435 (REORDER_45): Likewise.
16437 2018-06-25 Jeff Law <law@redhat.com>
16439 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
16440 dividend to 32 bits. Adjust length.
16441 (udivmodhi4): Cleanup output template. Fix length.
16443 2018-06-25 Carl Love <cel@us.ibm.com>
16445 * config/rs6000/vsx.md: Change word selector to prefered location.
16447 2018-06-25 Richard Biener <rguenther@suse.de>
16449 PR tree-optimization/86304
16450 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
16451 epilogue-if-converted loops as well.
16453 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
16455 * lto-section-out.c (lto_begin_section): Do not print section
16456 name for noaddr and unnumbered dumps.
16458 2018-06-25 Richard Biener <rguenther@suse.de>
16460 * tree-vectorizer.h (struct vec_info_shared): New structure
16461 with parts split out from struct vec_info and loop_nest from
16462 struct _loop_vec_info.
16463 (struct vec_info): Adjust accordingly.
16464 (struct _loop_vec_info): Likewise.
16465 (LOOP_VINFO_LOOP_NEST): Adjust.
16466 (LOOP_VINFO_DATAREFS): Likewise.
16467 (LOOP_VINFO_DDRS): Likewise.
16468 (struct _bb_vec_info): Likewise.
16469 (BB_VINFO_DATAREFS): Likewise.
16470 (BB_VINFO_DDRS): Likewise.
16471 (struct _stmt_vec_info): Add dr_aux member.
16472 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
16473 (DR_MISALIGNMENT_UNINITIALIZED): New.
16474 (set_dr_misalignment): Adjust.
16475 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
16476 (vect_analyze_loop): Adjust prototype.
16477 (vect_analyze_loop_form): Likewise.
16478 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
16479 Compute dependences lazily.
16480 (vect_record_base_alignments): Use shared datarefs/ddrs.
16481 (vect_verify_datarefs_alignment): Likewise.
16482 (vect_analyze_data_refs_alignment): Likewise.
16483 (vect_analyze_data_ref_accesses): Likewise.
16484 (vect_analyze_data_refs): Likewise.
16485 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
16486 constructor parameter for shared part.
16487 (vect_analyze_loop_form): Pass in shared part and adjust.
16488 (vect_analyze_loop_2): Pass in storage for the number of
16489 stmts. Move loop nest finding to the caller. Compute
16491 (vect_analyze_loop): Pass in shared part.
16492 (vect_transform_loop): Verify shared datarefs are unchanged.
16493 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
16494 constructor parameter for shared part.
16495 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
16496 (vect_slp_bb): Verify shared datarefs are unchanged before
16498 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
16500 (new_stmt_vec_info): Initialize DR_AUX misalignment to
16501 DR_MISALIGNMENT_UNINITIALIZED.
16502 * tree-vectorizer.c (vec_info::vec_info): Add constructor
16503 parameter for shared part.
16504 (vec_info::~vec_info): Adjust.
16505 (vec_info_shared::vec_info_shared): New.
16506 (vec_info_shared::~vec_info_shared): Likewise.
16507 (vec_info_shared::save_datarefs): Likewise.
16508 (vec_info_shared::check_datarefs): Likewise.
16509 (try_vectorize_loop_1): Construct shared part live for analyses
16510 of a single loop for multiple vector sizes.
16511 * tree-parloops.c (gather_scalar_reductions): Adjust.
16513 2018-06-25 Richard Biener <rguenther@suse.de>
16515 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
16516 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
16517 (vect_analyze_data_refs): Remove similar code from here and
16518 simplify accordingly.
16520 2018-06-25 Richard Biener <rguenther@suse.de>
16522 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
16523 for reverse storage order accesses rather than asserting
16524 they cannot happen here.
16526 2018-06-25 Tom de Vries <tdevries@suse.de>
16529 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
16530 Use data16 instead of .byte for insn prefix.
16532 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
16535 * parser.c (make_char_string_pack): Pass this literal chars
16536 through cpp_interpret_string.
16537 (cp_parser_userdef_numeric_literal): Check the result of
16538 make_char_string_pack.
16540 2018-06-24 Maya Rashish <coypu@sdf.org>
16542 * ginclude/stddef.h: Simplify conditions around avoiding
16543 re-definition of __size_t.
16545 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
16547 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
16548 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
16550 2018-06-22 Maya Rashish <coypu@sdf.org>
16552 * doc/invoke.texi (mno-fancy-math-387): Update for changes
16553 made to OpenBSD and NetBSD through the years.
16555 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
16557 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
16558 behavior of vec_pack (vector double, vector double) to match
16559 behavior of vec_float2 (vector double, vector double).
16561 2018-06-22 Olivier Hainque <hainque@adacore.com>
16563 * gimplify.c (gimplify_function_tree): Prevent creation
16564 of a trampoline for the address of the current function
16565 passed to entry/exit instrumentation hooks.
16567 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
16570 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
16573 2018-06-22 Martin Liska <mliska@suse.cz>
16575 PR tree-optimization/86263
16576 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
16577 Bail out if is_enabled is false.
16578 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
16580 (jump_table_cluster::is_enabled): New function.
16582 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
16584 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
16585 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
16586 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
16587 (lto_input_ts_binfo_tree_pointers): Likewise.
16588 * tree-streamer-out.c (streamer_write_tree_bitfields,
16589 write_ts_binfo_tree_pointers): Likewise.
16590 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
16592 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
16594 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
16596 2018-06-22 Martin Liska <mliska@suse.cz>
16598 * symbol-summary.h (get): Make it pure and inline move
16599 functionality from ::get function.
16600 (get): Remove and inline into ::get and ::get_create.
16601 (get_create): Move code from ::get function.
16603 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16606 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
16607 -x assembler-with-cpp.
16609 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16611 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
16612 _FILE_OFFSET_BITS=64 for C++.
16614 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
16616 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
16617 conversion insn that shows up when pr85657-3.c is compiled using
16618 IEEE 128-bit long double.
16619 (neg<mode>2_internal): Use the correct mode to check whether the
16620 mode is IBM extended.
16621 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
16622 multiply and divide external functions from being created more
16625 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
16627 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
16629 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
16630 the edge can be forwarded.
16631 (cfg_layout_merge_blocks): Likewise.
16633 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
16635 * except.c (finish_eh_generation): Commit edge insertions only after
16636 the EH edges have been redirected from post-landing to landing pads.
16638 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
16640 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
16641 create_tmp_var_for to create the FRAME decl.
16642 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
16644 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
16646 * tree-inline.c (copy_edges_for_bb): Minor tweak.
16647 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
16648 debug statement when resetting its value.
16649 (expand_call_inline): Copy the locus of the call onto the assignment
16650 of the return value, if any. Use local variable in more cases.
16652 2018-06-21 Martin Liska <mliska@suse.cz>
16654 * ipa-pure-const.c (propagate_nothrow): Use
16655 funct_state_summaries->get.
16656 (dump_malloc_lattice): Likewise.
16657 (propagate_malloc): Likewise.
16659 2018-06-21 Richard Biener <rguenther@suse.de>
16661 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
16662 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
16663 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
16665 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
16666 BLOCK_ABSTRACT_ORIGIN unconditionally.
16668 2018-06-21 David Malcolm <dmalcolm@redhat.com>
16670 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
16672 * ipa-reference.c (ipa_reference_c_finalize): Delete
16673 ipa_ref_opt_sum_summaries and set it to NULL.
16675 2018-06-21 Tom de Vries <tdevries@suse.de>
16677 PR tree-optimization/85859
16678 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
16679 test with comment from bb_no_side_effects_p.
16681 2018-06-21 Richard Biener <rguenther@suse.de>
16683 PR tree-optimization/86232
16684 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
16685 max for constant niter.
16687 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
16689 * config/aarch64/aarch64-simd.md
16690 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
16692 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
16694 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
16695 Make opernads of the unspec commutative.
16697 2018-06-21 Richard Biener <rguenther@suse.de>
16699 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
16700 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
16701 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
16702 (vect_analyze_data_ref_dependence): Re-order checks to deal with
16704 (vect_record_base_alignments): Do not record base alignment
16705 for gathers or scatters.
16706 (vect_compute_data_ref_alignment): Drop return value that is always
16707 true. Bail out early for gathers or scatters.
16708 (vect_enhance_data_refs_alignment): Bail out early for gathers
16710 (vect_find_same_alignment_drs): Likewise.
16711 (vect_analyze_data_refs_alignment): Remove dead code.
16712 (vect_slp_analyze_and_verify_node_alignment): Likewise.
16713 (vect_analyze_data_refs): For possible gathers or scatters do
16714 not create an alternate DR, just check their possible validity
16715 and mark them. Adjust DECL_NONALIASED handling to not rely
16716 on DR_BASE_ADDRESS.
16717 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
16718 update inits of gathers or scatters.
16719 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
16720 Also copy gather/scatter flag to pattern vinfo.
16722 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
16724 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
16725 behavior of vec_packsu (vector unsigned long long, vector unsigned
16726 long long) to match behavior of vec_packs with same signature.
16728 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
16729 Thomas Schwinge <thomas@codesourcery.com>
16730 Cesar Philippidis <cesar@codesourcery.com>
16732 * gimplify.c (gimplify_scan_omp_clauses): Add support for
16733 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
16734 (gimplify_adjust_omp_clauses): Likewise.
16735 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
16736 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
16737 (gimplify_omp_target_update): Update handling of acc update and
16739 * omp-low.c (install_var_field): Remove unused parameter
16740 base_pointers_restrict.
16741 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
16742 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
16744 (omp_target_base_pointers_restrict_p): Delete.
16745 (scan_omp_target): Update call to scan_sharing_clauses.
16746 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
16748 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
16749 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
16750 (convert_local_omp_clauses): Likewise.
16751 * tree-pretty-print.c (dump_omp_clause): Likewise.
16752 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
16754 (omp_clause_code_name): Likewise.
16756 2018-06-20 Jakub Jelinek <jakub@redhat.com>
16759 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
16762 PR tree-optimization/86231
16763 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
16764 anti-range don't overwrite *vr0min before using it to compute *vr0max.
16766 2018-06-20 Tom de Vries <tdevries@suse.de>
16768 PR tree-optimization/86097
16769 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
16770 iv type if signedness of iv type is not the same as that of *nit.
16772 2018-06-20 Jakub Jelinek <jakub@redhat.com>
16774 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
16775 EDGE_EH edges, verify they are all EDGE_EH.
16777 2018-06-20 Maya Rashish <coypu@sdf.org>
16779 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
16781 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16783 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
16784 * config/aarch64/aarch64.c (xgene1_tunings): Add
16785 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
16786 (aarch64_mode_valid_for_sched_fusion_p):
16787 Allow 16-byte modes.
16788 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
16789 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
16791 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
16793 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
16794 * config/aarch64/iterators.md (VQ2): New mode iterator.
16796 2018-06-20 Martin Liska <mliska@suse.cz>
16798 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
16799 Change default ratio from 10 to 8.
16801 2018-06-20 Martin Liska <mliska@suse.cz>
16803 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
16805 (bit_test_cluster::find_bit_tests): Likewise.
16806 (switch_decision_tree::analyze_switch_statement): Find clusters.
16807 * tree-switch-conversion.h (struct jump_table_cluster): Document
16810 2018-06-20 Martin Liska <mliska@suse.cz>
16812 * tree-switch-conversion.c (switch_conversion::collect):
16813 Record m_uniq property.
16814 (switch_conversion::expand): Bail out for special conditions.
16815 (group_cluster::~group_cluster): New.
16816 (group_cluster::group_cluster): Likewise.
16817 (group_cluster::dump): Likewise.
16818 (jump_table_cluster::emit): New.
16819 (switch_decision_tree::fix_phi_operands_for_edges): New.
16820 (struct case_node): Remove struct.
16821 (jump_table_cluster::can_be_handled): New.
16822 (case_values_threshold): Moved to header.
16823 (reset_out_edges_aux): Likewise.
16824 (jump_table_cluster::is_beneficial): New.
16825 (bit_test_cluster::can_be_handled): Likewise.
16826 (add_case_node): Remove.
16827 (bit_test_cluster::is_beneficial): New.
16828 (case_bit_test::cmp): New.
16829 (bit_test_cluster::emit): New.
16830 (expand_switch_as_decision_tree_p): Remove.
16831 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
16832 (fix_phi_operands_for_edge): Likewise.
16833 (switch_decision_tree::analyze_switch_statement): New.
16834 (compute_cases_per_edge): Move ...
16835 (switch_decision_tree::compute_cases_per_edge): ... here.
16836 (try_switch_expansion): Likewise.
16837 (switch_decision_tree::try_switch_expansion): Likewise.
16838 (record_phi_operand_mapping): Likewise.
16839 (switch_decision_tree::record_phi_operand_mapping): Likewise.
16840 (emit_case_decision_tree): Likewise.
16841 (switch_decision_tree::emit): Likewise.
16842 (balance_case_nodes): Likewise.
16843 (switch_decision_tree::balance_case_nodes): Likewise.
16844 (dump_case_nodes): Likewise.
16845 (switch_decision_tree::dump_case_nodes): Likewise.
16846 (emit_jump): Likewise.
16847 (switch_decision_tree::emit_jump): Likewise.
16848 (emit_cmp_and_jump_insns): Likewise.
16849 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
16850 (emit_case_nodes): Likewise.
16851 (switch_decision_tree::emit_case_nodes): Likewise.
16852 (conditional_probability): Remove.
16853 * tree-switch-conversion.h (enum cluster_type): New.
16855 (struct cluster): Likewise.
16856 (cluster::cluster): Likewise.
16857 (struct simple_cluster): Likewise.
16858 (simple_cluster::simple_cluster): Likewise.
16859 (struct group_cluster): Likewise.
16860 (struct jump_table_cluster): Likewise.
16861 (struct bit_test_cluster): Likewise.
16862 (struct min_cluster_item): Likewise.
16863 (struct case_tree_node): Likewise.
16864 (case_tree_node::case_tree_node): Likewise.
16865 (jump_table_cluster::case_values_threshold): Likewise.
16866 (struct case_bit_test): Likewise.
16867 (struct switch_decision_tree): Likewise.
16868 (struct switch_conversion): Likewise.
16869 (switch_decision_tree::reset_out_edges_aux): Likewise.
16871 2018-06-20 Martin Liska <mliska@suse.cz>
16873 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
16874 (hoist_edge_and_branch_if_true): Likewise.
16875 (expand_switch_using_bit_tests_p): Likewise.
16876 (struct case_bit_test): Likewise.
16877 (case_bit_test_cmp): Likewise.
16878 (emit_case_bit_tests): Likewise.
16879 (switch_conversion::switch_conversion): New class.
16880 (struct switch_conv_info): Remove old struct.
16881 (collect_switch_conv_info): More to ...
16882 (switch_conversion::collect): ... this.
16883 (check_range): Likewise.
16884 (switch_conversion::check_range): Likewise.
16885 (check_all_empty_except_final): Likewise.
16886 (switch_conversion::check_all_empty_except_final): Likewise.
16887 (check_final_bb): Likewise.
16888 (switch_conversion::check_final_bb): Likewise.
16889 (create_temp_arrays): Likewise.
16890 (switch_conversion::create_temp_arrays): Likewise.
16891 (free_temp_arrays): Likewise.
16892 (gather_default_values): Likewise.
16893 (switch_conversion::gather_default_values): Likewise.
16894 (build_constructors): Likewise.
16895 (switch_conversion::build_constructors): Likewise.
16896 (constructor_contains_same_values_p): Likewise.
16897 (switch_conversion::contains_same_values_p): Likewise.
16898 (array_value_type): Likewise.
16899 (switch_conversion::array_value_type): Likewise.
16900 (build_one_array): Likewise.
16901 (switch_conversion::build_one_array): Likewise.
16902 (build_arrays): Likewise.
16903 (switch_conversion::build_arrays): Likewise.
16904 (gen_def_assigns): Likewise.
16905 (switch_conversion::gen_def_assigns): Likewise.
16906 (prune_bbs): Likewise.
16907 (switch_conversion::prune_bbs): Likewise.
16908 (fix_phi_nodes): Likewise.
16909 (switch_conversion::fix_phi_nodes): Likewise.
16910 (gen_inbound_check): Likewise.
16911 (switch_conversion::gen_inbound_check): Likewise.
16912 (process_switch): Use the newly created class.
16913 (switch_conversion::expand): New.
16914 (switch_conversion::~switch_conversion): New.
16915 * tree-switch-conversion.h: New file.
16917 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
16919 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
16920 tree-vect-patterns.c.
16921 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
16922 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
16923 (vect_recog_sad_pattern): Likewise.
16924 (vect_recog_widen_sum_pattern): Likewise.
16925 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
16926 (vect_recog_widen_shift_pattern): Remove the type_in argument.
16927 (vect_recog_rotate_pattern): Likewise.
16928 (vect_recog_mult_pattern): Likewise.
16929 (vect_recog_vector_vector_shift_pattern): Likewise.
16930 (vect_recog_divmod_pattern): Likewise.
16931 (vect_recog_mixed_size_cond_pattern): Likewise.
16932 (vect_recog_bool_pattern): Likewise.
16933 (vect_recog_mask_conversion_pattern): Likewise.
16934 (vect_try_gather_scatter_pattern): Likewise.
16935 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
16936 (vect_recog_over_widening_pattern): Likewise.
16937 (vect_recog_gather_scatter_pattern): Likewise.
16938 (vect_recog_func_ptr): Move from tree-vectorizer.h
16939 (vect_vect_recog_func_ptrs): Move further down the file.
16940 (vect_recog_func): Likewise. Remove the third argument.
16941 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
16942 (vect_pattern_recog_1): Expect the pattern function to do any
16943 necessary target tests. Also expect it to provide a vector type.
16944 Remove the type_in handling.
16946 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
16948 * tree-vect-patterns.c (vect_pattern_detected): New function.
16949 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
16950 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
16951 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
16952 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
16953 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
16954 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
16955 (vect_recog_mask_conversion_pattern)
16956 (vect_try_gather_scatter_pattern): Likewise.
16958 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
16960 * tree-vect-patterns.c (vect_get_internal_def): New function.
16961 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
16962 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
16963 (search_type_for_mask_1): Use it.
16965 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
16967 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
16968 redundant WIDEN_SUM_EXPR handling.
16969 (vect_recog_sad_pattern): Likewise.
16971 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
16973 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
16974 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
16975 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
16976 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
16977 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
16979 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
16981 * tree-vect-stmts.c (vectorizable_call): Make sure that we
16982 use the stmt_vec_info of the original bb statement for the
16983 new zero assignment, even if the call is part of a pattern.
16985 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
16987 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
16988 that the sequence is attached to the original statement rather
16989 than the pattern statement.
16990 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
16991 PATTERN_DEF_SEQ from the original statement rather than
16992 the main pattern statement.
16993 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
16994 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
16995 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
16997 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
16999 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
17000 definition statements before the early exit for statements that aren't
17002 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
17004 (vect_transform_loop): ...here. Process pattern definition
17005 statements without first checking whether the main pattern
17006 statement is live or relevant.
17008 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
17010 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
17011 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
17013 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
17015 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
17016 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
17017 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
17018 (expand_block_compare): Change select_block_compare_mode call.
17019 (expand_strncmp_align_check): Use new functions, fix comment.
17020 (emit_final_str_compare_gpr): New function.
17021 (expand_strn_compare): Refactor and clean up code.
17022 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
17024 2018-06-19 Tony Reix <tony.reix@atos.com>
17025 Damien Bergamini <damien.bergamini@atos.com>
17026 David Edelsohn <dje.gcc@gmail.com>
17028 * collect2.c (static_obj): New variable.
17029 (static_libs): New variable.
17030 (is_in_list): Uncomment declaration.
17031 (main): Track AIX libraries linked statically.
17032 (is_in_list): Uncomment definition.
17033 (scan_prog_file): Don't add AIX shared libraries initializer
17034 to constructor list if linking statically.
17036 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
17038 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
17040 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
17042 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
17044 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
17047 2018-06-19 Martin Liska <mliska@suse.cz>
17049 * config/i386/i386.c (ix86_can_inline_p): Do not use
17050 ipa_fn_summaries::get_create.
17051 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
17053 (devirtualization_time_bonus): Likewise.
17054 (ipcp_propagate_stage): Likewise.
17055 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
17056 (edge_set_predicate): Likewise.
17057 (evaluate_conditions_for_known_args): Likewise.
17058 (evaluate_properties_for_edge): Likewise.
17059 (ipa_call_summary::reset): Tranform to ...
17060 (ipa_call_summary::~ipa_call_summary): ... this.
17061 (ipa_fn_summary::reset): Transform to ...
17062 (ipa_fn_summary::~ipa_fn_summary): ... this.
17063 (ipa_fn_summary_t::remove): Rename to ...
17064 (ipa_fn_summary_t::remove_callees): ... this.
17065 (ipa_fn_summary_t::duplicate): Use placement new
17066 instead of memory copy.
17067 (ipa_call_summary_t::duplicate): Likewise.
17068 (ipa_call_summary_t::remove): Remove.
17069 (dump_ipa_call_summary): Change get_create to get.
17070 (ipa_dump_fn_summary): Dump only when summary exists.
17071 (analyze_function_body): Use symbol_summary::get instead
17073 (compute_fn_summary): Likewise.
17074 (estimate_edge_devirt_benefit): Likewise.
17075 (estimate_edge_size_and_time): Likewise.
17076 (inline_update_callee_summaries): Likewise.
17077 (remap_edge_change_prob): Likewise.
17078 (remap_edge_summaries): Likewise.
17079 (ipa_merge_fn_summary_after_inlining): Likewise.
17080 (write_ipa_call_summary): Likewise.
17081 (ipa_fn_summary_write): Likewise.
17082 (ipa_free_fn_summary): Likewise.
17083 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
17084 (struct ipa_call_summary): Likewise.
17085 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
17087 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
17088 (estimate_size_after_inlining): Likewise.
17089 (estimate_growth): Likewise.
17090 (growth_likely_positive): Likewise.
17091 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
17092 (inline_call): Likewise.
17093 * ipa-inline.c (caller_growth_limits): Likewise.
17094 (can_inline_edge_p): Likewise.
17095 (can_inline_edge_by_limits_p): Likewise.
17096 (compute_uninlined_call_time): Likewise.
17097 (compute_inlined_call_time): Likewise.
17098 (want_inline_small_function_p): Likewise.
17099 (edge_badness): Likewise.
17100 (update_caller_keys): Likewise.
17101 (update_callee_keys): Likewise.
17102 (inline_small_functions): Likewise.
17103 (inline_to_all_callers_1): Likewise.
17104 (dump_overall_stats): Likewise.
17105 (early_inline_small_functions): Likewise.
17106 (early_inliner): Likewise.
17107 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
17108 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
17109 * ipa-pure-const.c (malloc_candidate_p): Likewise.
17110 * ipa-split.c (execute_split_functions): Likewise.
17111 * symbol-summary.h: Likewise.
17112 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
17114 2018-06-19 Richard Biener <rguenther@suse.de>
17116 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
17117 (vectorize_loops): ... here. Fix dbgcnt handling.
17118 (try_vectorize_loop): Wrap try_vectorize_loop_1.
17120 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
17123 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
17124 ieee128 argument takes up only one (vector) register, not two (floating
17127 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
17129 * gimplify.c (gimplify_init_constructor): Really never clear for an
17130 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
17132 2018-06-19 Richard Biener <rguenther@suse.de>
17134 PR tree-optimization/86179
17135 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
17136 after failed recognition.
17138 2018-06-18 Martin Sebor <msebor@redhat.com>
17140 PR middle-end/85602
17141 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
17142 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
17143 Handle integer subtraction.
17144 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
17145 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
17147 2018-06-18 David Malcolm <dmalcolm@redhat.com>
17149 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
17150 param from rtx to rtx_insn *.
17151 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
17153 (frv_ifcvt_modify_insn): Likwise.
17154 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
17155 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
17156 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
17157 as_a <rtx_insn *> cast to local "unprotected_region" once
17158 it's been established that it's not NULL or pc_rtx.
17159 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
17160 param "sethi" from rtx to rtx_insn *.
17161 (nds32_group_float_insns): Likewise for param "insn".
17162 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
17164 (vax_output_int_subtract): Likewise.
17165 * config/vax/vax.c (vax_output_int_add): Likewise for param
17167 (vax_output_int_subtract): Likewise.
17168 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
17169 (emit_pattern_after): Likewise for param "after".
17170 (emit_insn_after): Likewise.
17171 (emit_jump_insn_after): Likewise.
17172 (emit_call_insn_after): Likewise.
17173 (emit_debug_insn_after): Likewise.
17174 (emit_pattern_before): Likewise for param "before".
17175 (emit_insn_before): Likewise.
17176 (emit_jump_insn_before): Likewise.
17177 * final.c (get_insn_template): Likewise for param "insn", removing
17179 * output.h (get_insn_template): Likewise for 2nd param.
17180 * rtl.h (emit_insn_before): Likewise.
17181 (emit_jump_insn_before): Likewise.
17182 (emit_debug_insn_before_noloc): Likewise.
17183 (emit_insn_after): Likewise.
17184 (emit_jump_insn_after): Likewise.
17185 (emit_call_insn_after): Likewise.
17186 (emit_debug_insn_after): Likewise.
17187 (set_insn_deleted): Likewise for param.
17189 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
17192 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
17193 floating point modes, so that IFmode is numerically greater than
17194 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
17195 to declare the ordering. This prevents IFmode from being
17196 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
17197 machine. Include rs6000-modes.h to share the fractional values
17198 between genmodes* and the rest of the compiler.
17199 (IFmode): Likewise.
17200 (KFmode): Likewise.
17201 (TFmode): Likewise.
17202 * config/rs6000/rs6000-modes.h: New file.
17203 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
17204 meaning of rs6000_long_double_size so that 126..128 selects an
17205 appropriate 128-bit floating point type.
17206 (rs6000_option_override_internal): Likewise.
17207 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
17208 (TARGET_LONG_DOUBLE_128): Change the meaning of
17209 rs6000_long_double_size so that 126..128 selects an appropriate
17210 128-bit floating point type.
17211 (LONG_DOUBLE_TYPE_SIZE): Update comment.
17212 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
17213 source and destination to match the standard usage.
17214 (truncifkf2): Likewise.
17215 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
17216 ISA 2.07 to use an explicit clobber, instead of passing in a
17218 (copysign<mode>3_soft): Likewise.
17220 2018-06-18 David Malcolm <dmalcolm@redhat.com>
17222 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
17223 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
17224 (vect_slp_analyze_instance_dependence): Likewise.
17225 (vect_enhance_data_refs_alignment): Likewise.
17226 (vect_analyze_data_refs_alignment): Likewise.
17227 (vect_slp_analyze_and_verify_instance_alignment
17228 (vect_analyze_data_ref_accesses): Likewise.
17229 (vect_prune_runtime_alias_test_list): Likewise.
17230 (vect_analyze_data_refs): Likewise.
17231 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
17232 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
17233 (vect_analyze_scalar_cycles_1): Likewise.
17234 (vect_get_loop_niters): Likewise.
17235 (vect_analyze_loop_form_1): Likewise.
17236 (vect_update_vf_for_slp): Likewise.
17237 (vect_analyze_loop_operations): Likewise.
17238 (vect_analyze_loop): Likewise.
17239 (vectorizable_induction): Likewise.
17240 (vect_transform_loop): Likewise.
17241 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
17242 * tree-vect-slp.c (vect_analyze_slp): Likewise.
17243 (vect_make_slp_decision): Likewise.
17244 (vect_detect_hybrid_slp): Likewise.
17245 (vect_slp_analyze_operations): Likewise.
17246 (vect_slp_bb): Likewise.
17247 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
17248 (vectorizable_bswap): Likewise.
17249 (vectorizable_call): Likewise.
17250 (vectorizable_simd_clone_call): Likewise.
17251 (vectorizable_conversion): Likewise.
17252 (vectorizable_assignment): Likewise.
17253 (vectorizable_shift): Likewise.
17254 (vectorizable_operation): Likewise.
17255 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
17257 2018-06-18 Martin Sebor <msebor@redhat.com>
17259 PR tree-optimization/81384
17260 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
17261 * builtins.c (expand_builtin_strnlen): New function.
17262 (expand_builtin): Call it.
17263 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
17264 * builtins.def (BUILT_IN_STRNLEN): New.
17265 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
17266 Warn for bounds in excess of maximum object size.
17267 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
17268 single-value ranges. Handle strnlen.
17269 (handle_builtin_strlen): Handle strnlen.
17270 (strlen_check_and_optimize_stmt): Same.
17271 * doc/extend.texi (Other Builtins): Document strnlen.
17273 2018-06-18 Maya Rashish <coypu@sdf.org>
17275 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
17276 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
17277 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
17279 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
17281 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
17283 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17285 * tree.c (escaped_string::escape): Replace cast to char * by
17286 const_cast<char *> (unescaped).
17288 2018-06-18 Nick Clifton <nickc@redhat.com>
17291 * tree.c (escaped_string): New class. Converts an unescaped
17292 string into its escaped equivalent.
17293 (warn_deprecated_use): Use the new class to convert the
17294 deprecation message, if present.
17295 (test_escaped_strings): New self test.
17296 (test_c_tests): Add test_escaped_strings.
17297 * doc/extend.texi (deprecated): Add a note that the
17298 deprecation message is affected by the -fmessage-length
17299 option, and that control characters will be escaped.
17300 (#pragma GCC error): Document this pragma.
17301 (#pragma GCC warning): Likewise.
17302 * doc/invoke.texi (-fmessage-length): Document this option's
17303 effect on the #warning and #error preprocessor directives and
17304 the deprecated attribute.
17306 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
17308 * tree.c (decl_value_expr_lookup): Revert latest change.
17309 (decl_value_expr_insert): Likewise.
17311 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
17313 * gimplify.c (nonlocal_vlas): Delete.
17314 (nonlocal_vla_vars): Likewise.
17315 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
17317 (gimplify_body): Do not create and destroy nonlocal_vlas.
17318 * tree-nested.c: Include diagnostic.h.
17319 (use_pointer_in_frame): Tweak.
17320 (lookup_field_for_decl): Add assertion and declare the transformation.
17321 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
17322 internal error when the reference is in a wrong context. Do not
17323 create a debug decl by default.
17324 (note_nonlocal_block_vlas): Delete.
17325 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
17326 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
17327 create a debug decl by default.
17328 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
17329 call is in a wrong context.
17330 (fixup_vla_decls): New function.
17331 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
17332 debug variables were created.
17333 * tree.c (decl_value_expr_lookup): Add checking assertion.
17334 (decl_value_expr_insert): Likewise.
17336 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
17338 PR middle-end/82479
17339 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
17340 * tree-scalar-evolution.c (interpret_expr): Likewise.
17341 (expression_expensive_p): Likewise.
17342 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
17343 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
17344 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
17345 (ssa_defined_by_minus_one_stmt_p): New.
17347 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
17349 PR middle-end/64946
17350 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
17351 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
17352 * dojump.c (do_jump): Likewise.
17353 * expr.c (expand_expr_real_2): Check operand type's sign.
17354 * fold-const.c (const_unop): Handle ABSU_EXPR.
17355 (fold_abs_const): Likewise.
17356 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
17357 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
17358 (strip_sign_op_1): Likesise.
17359 * match.pd: Add new pattern to generate ABSU_EXPR.
17360 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
17361 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
17362 * tree-eh.c (operation_could_trap_helper_p): Likewise.
17363 * tree-inline.c (estimate_operator_cost): Likewise.
17364 * tree-pretty-print.c (dump_generic_node): Likewise.
17365 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
17366 * tree.def (ABSU_EXPR): New.
17368 2018-06-16 Jakub Jelinek <jakub@redhat.com>
17370 PR middle-end/86095
17371 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
17372 documented as preserved for backward compatibility only.
17373 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
17375 PR rtl-optimization/86108
17376 * bb-reorder.c (create_forwarder_block): Renamed to ...
17377 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
17378 jump from new landing pad to the second part.
17379 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
17382 2018-06-15 Jakub Jelinek <jakub@redhat.com>
17384 PR middle-end/85878
17385 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
17386 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
17387 Only call store_expr for halves if the mode is the same.
17389 PR middle-end/86123
17390 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
17391 Fix up comment formatting.
17393 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
17395 * typed-splay-tree.h (typed_splay_tree::remove): New function.
17396 (typed_splay_tree::closure,
17397 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
17398 (typed_splay_tree::typed_splay_tree,
17399 typed_splay_tree::operator =): Declared private.
17400 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
17401 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
17402 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
17403 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
17404 typed_splay_tree::splay_tree_splay,
17405 typed_splay_tree::splay_tree_foreach_helper,
17406 typed_splay_tree::splay_tree_insert,
17407 typed_splay_tree::splay_tree_remove,
17408 typed_splay_tree::splay_tree_lookup,
17409 typed_splay_tree::splay_tree_predecessor,
17410 typed_splay_tree::splay_tree_successor,
17411 typed_splay_tree::splay_tree_min,
17412 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
17413 (typed_splay_tree::root, typed_splay_tree::comp,
17414 typed_splay_tree::delete_key,
17415 typed_splay_tree::delete_value): New data members.
17416 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
17417 typed_splay_tree::remove.
17419 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
17421 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
17422 -mginv and -mno-ginv to the assembler.
17423 * config/mips/mips.opt (-mcrc): New option.
17424 (-mginv): Likewise.
17425 * doc/invoke.text (-mcrc): Document.
17426 (-mginv): Likewise.
17428 2018-06-15 Nick Clifton <nickc@redhat.com>
17431 * tree.c (escaped_string): New class. Converts an unescaped
17432 string into its escaped equivalent.
17433 (warn_deprecated_use): Use the new class to convert the
17434 deprecation message, if present.
17435 (test_escaped_strings): New self test.
17436 (test_c_tests): Add test_escaped_strings.
17437 * doc/extend.texi (deprecated): Add a note that the
17438 deprecation message is affected by the -fmessage-length
17439 option, and that control characters will be escaped.
17440 (#pragma GCC error): Document this pragma.
17441 (#pragma GCC warning): Likewise.
17442 * doc/invoke.texi (-fmessage-length): Document this option's
17443 effect on the #warning and #error preprocessor directives and
17444 the deprecated attribute.
17446 2018-06-15 Richard Biener <rguenther@suse.de>
17448 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
17449 here, also noting vector size used.
17450 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
17451 size used in MSG_OPTIMIZED_LOCATIONS dump.
17452 (pass_slp_vectorize::execute): Adjust.
17454 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
17457 * config/arc/arc.c (arc_return_address_register): Fix
17460 2018-06-15 Richard Biener <rguenther@suse.de>
17462 PR middle-end/86159
17463 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
17464 leave useless conversion stripping to force_gimple_operand_gsi.
17465 (gimplify_build2): Likewise.
17466 (gimplify_build1): Likewise.
17468 2018-06-15 Richard Biener <rguenther@suse.de>
17470 PR middle-end/86076
17471 * tree-cfg.c (move_stmt_op): unshare invariant addresses
17472 before adjusting their block.
17474 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
17476 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
17477 multilibs for *-*-rtems*.
17478 * config/riscv/t-rtems: New file.
17480 2018-06-14 Jakub Jelinek <jakub@redhat.com>
17482 PR middle-end/86122
17483 * match.pd ((A +- CST1) +- CST2): Punt if last resort
17484 unsigned_type_for returns NULL.
17487 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
17488 subregs of multi-word pseudos unless the float mode has word size.
17490 2018-06-14 Richard Biener <rguenther@suse.de>
17492 PR middle-end/86139
17493 * tree-vect-generic.c (build_word_mode_vector_type): Remove
17494 duplicate and harmful type_hash_canon.
17495 * tree.c (type_hash_canon): Assert we didn't find ourselves.
17497 2018-06-14 Richard Biener <rguenther@suse.de>
17500 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
17503 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
17505 * config/rtems.h (STDINT_LONG32): Define.
17507 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
17508 Prachi Godbole <prachi.godbole@imgtec.com>
17510 * config/mips/mips-cpus.def: Define P6600.
17511 * config/mips/mips-tables.opt: Regenerate.
17512 * config/mips/mips.c (mips_ucbranch_type): New enum.
17513 (mips_rtx_cost_data): Add support for P6600.
17514 (mips_issue_rate): Likewise.
17515 (mips_multipass_dfa_lookahead): Likewise.
17516 (mips_avoid_hazard): Likewise.
17517 (mips_reorg_process_insns): Likewise.
17518 (mips_classify_branch_p6600): New function.
17519 * config/mips/mips.h (TUNE_P6600): New define.
17520 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
17521 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
17522 * config/mips/mips.md: Include p6600.md.
17523 (processor): Add p6600.
17524 * config/mips/p6600.md: New file.
17525 * doc/invoke.texi: Add p6600 to supported architectures.
17527 2018-06-13 Martin Sebor <msebor@redhat.com>
17529 PR tree-optimization/86114
17530 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
17532 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
17534 2018-06-13 Richard Biener <rguenther@suse.de>
17536 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
17537 Properly set vector type of the intermediate stmt.
17538 * tree-vect-stmts.c (vectorizable_operation): The destination
17539 var always has vectype_out type.
17541 2018-06-13 Jeff Law <law@redhat.com>
17543 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
17544 integer 0 for argument to print_rtl_with_bb.
17545 (rl78_reorg): Likewise.
17547 2018-06-13 David Malcolm <dmalcolm@redhat.com>
17549 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
17550 from rtx to rtx_insn *.
17551 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
17553 (add_sched_insns_for_speculation): Likewise for local "target",
17554 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
17555 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
17556 from rtx_insn ** to rtx_code_label **.
17557 (reorg_emit_nops): Likewise.
17558 (c6x_reorg): Likewise for local "call_labels".
17559 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
17561 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
17562 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
17563 the loops over LABEL_REFS.
17564 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
17566 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
17567 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
17568 (split_branches): Strengthen local "olabel" from rtx to
17569 rtx_insn *, adding a safe_as_a cast.
17570 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
17572 (add_insn_after): Likewise for first two params.
17573 (add_insn_before): Likewise.
17574 (remove_insn): Likewise for param.
17575 (emit_pattern_before_noloc): Likewise for second and third params.
17576 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
17577 (emit_call_insn_before_noloc): Likewise.
17578 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
17580 (emit_barrier_before): Likewise.
17581 (emit_label_before): Strengthen "label" param from "rtx" to
17582 "rtx_code_label *". Strengthen "before" param from "rtx" to
17584 (emit_insn_after_1): Strengthen "after" param from "rtx" to
17586 (emit_pattern_after_noloc): Likewise.
17587 (emit_insn_after_noloc): Likewise.
17588 (emit_jump_insn_after_noloc): Likewise.
17589 (emit_call_insn_after_noloc): Likewise.
17590 (emit_debug_insn_after_noloc): Likewise.
17591 (emit_barrier_after): Likewise.
17592 (emit_label_after): Likewise for both params.
17593 (emit_pattern_after_setloc): Likewise for "after" param. Convert
17594 "loc" param from "int" to "location_t".
17595 (emit_insn_after_setloc): Likewise.
17596 (emit_jump_insn_after_setloc): Likewise.
17597 (emit_call_insn_after_setloc): Likewise.
17598 (emit_debug_insn_after_setloc): Likewise.
17599 (emit_pattern_before_setloc): Likewise for "before" param. Convert
17600 "loc" param from "int" to "location_t".
17601 (emit_pattern_before): Convert NULL_RTX to NULL.
17602 (emit_insn_before_setloc): Convert "loc" param from "int" to
17604 (emit_jump_insn_before_setloc): Likewise.
17605 (emit_call_insn_before_setloc): Likewise.
17606 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
17607 rtx_insn *. Convert "loc" param from "int" to "location_t".
17608 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
17609 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
17610 Convert 3rd param from "int" to "location_t".
17611 (emit_barrier_before, emit_barrier_after, next_real_insn):
17612 Strengthen param from rtx to rtx_insn *.
17613 (emit_label_before): Strengthen 1st param from "rtx" to
17614 "rtx_code_label *". Strengthen 2nd param from "rtx" to
17616 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
17617 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
17618 Strengthen 2nd param from "rtx" to "rtx_insn *".
17619 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
17620 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
17621 Likewise. Convert 3rd param from "int" to "location_t".
17622 (emit_label_after): Strengthen 1st param from "rtx" to
17623 "rtx_code_label *".
17624 (next_real_insn, remove_insn): Strengthen param from "rtx" to
17626 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
17627 from "rtx" to "rtx_insn *".
17629 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
17631 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
17632 bodies streamed in with -Q.
17633 * dumpfile.c (dump_files): Add lto-stream-out dump file.
17634 * dumpfile.h (tree_dump_index): Add lto_stream_out.
17635 * gimple-streamer-out.c: Include gimple-pretty-print.h
17636 (output_bb): Dump stmts streamed.
17637 * lto-section-out.c: Include print-tree.h
17638 (lto_begin_section): Dump sections created.
17639 (lto_output_decl_index): Dump decl encoded.
17640 * lto-streamer-out.c: Include print-tree.h
17641 (create_output_block): Dump output block created.
17642 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
17643 (output_function): Dump function output.
17644 (output_constructor): Dump constructor streamed.
17645 (write_global_stream): Output indexes encoded.
17646 (produce_asm_for_decls): Dump streams encoded.
17647 * lto-streamer.c (streamer_dump_file): New global var.
17648 * lto-streamer.h (streamer_dump_file): Declare.
17649 * passes.c (ipa_write_summaries): Initialize streamer dump.
17650 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
17653 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
17656 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
17657 offsets for register save directives. Emit a second batch of save
17658 directives, if need be, when the function accesses prior frames.
17660 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
17662 * config/arc/fpu.md (fmasf4): Force operand to register.
17663 (fnmasf4): Likewise.
17665 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
17667 * config/arc/arc-protos.h (arc_pad_return): Remove.
17668 * config/arc/arc.c (machine_function): Remove force_short_suffix
17670 (arc_print_operand): Adjust printing of '&'.
17671 (arc_verify_short): Remove conditional printing of short suffix.
17672 (arc_final_prescan_insn): Remove reference to size_reason.
17673 (pad_return): New function.
17674 (arc_reorg): Call pad_return.
17675 (arc_pad_return): Remove.
17676 (arc_init_machine_status): Remove reference to force_short_suffix.
17677 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
17678 (attr length): When attribute iscompact is true force to 2
17679 regardless; in the case of maybe check if we want to force the
17680 instruction to have 4 bytes length.
17681 (nopv): Change it to generate 4 byte long nop as well.
17682 (blockage): New pattern.
17683 (simple_return): Remove call to arc_pad_return.
17684 (p_return_i): Likewise.
17686 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
17688 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
17690 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
17692 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
17695 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
17697 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
17698 for ARC700 and ARCv2.
17700 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
17703 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
17704 operands[2] instead of operands[1].
17707 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
17709 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
17710 case, check whether the outer register overlaps an unallocatable
17711 register, not just whether it fits the required class.
17713 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
17715 * poly-int.h (can_div_trunc_p): Add new overload in which all values
17717 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
17718 (memrefs_conflict_p): Likewise.
17719 (init_alias_analysis): Likewise.
17720 * cfgexpand.c (expand_debug_expr): Likewise.
17721 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
17722 * cse.c (fold_rtx): Likewise.
17723 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
17724 * expr.c (emit_block_move_hints): Likewise.
17725 (clear_storage_hints, push_block, emit_push_insn): Likewise.
17726 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
17727 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
17728 (emit_group_store): Likewise.
17729 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
17730 to read the PRE/POST_MODIFY increment.
17731 * calls.c (store_one_arg): Use strip_offset.
17732 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
17734 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
17736 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
17737 (simplify_binary_operation_1): Extend CONST_INT handling to
17739 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
17740 than a HOST_WIDE_INT.
17741 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
17743 (adjust_mems, add_stores): Update accodingly.
17744 (vt_canonicalize_addr): Track polynomial offsets.
17745 (emit_note_insn_var_location): Likewise.
17746 (vt_add_function_parameter): Likewise.
17747 (vt_initialize): Likewise.
17749 2018-06-12 Jeff Law <law@redhat.com>
17751 * config.gcc (alpha*-*-freebsd*): Remove.
17752 * config/alpha/freebsd.h: Remove.
17754 2018-06-12 David Malcolm <dmalcolm@redhat.com>
17757 * spellcheck-tree.c (levenshtein_distance): Rename to...
17758 (get_edit_distance): ...this, and update for underlying renaming.
17759 * spellcheck-tree.h (levenshtein_distance): Rename to...
17760 (get_edit_distance): ...this.
17761 * spellcheck.c (levenshtein_distance): Rename to...
17762 (get_edit_distance): ...this. Convert from Levenshtein distance
17763 to Damerau-Levenshtein distance by supporting transpositions of
17764 adjacent characters. Rename "v1" to "v_next" and "v0" to
17766 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
17767 (selftest::test_edit_distance_unit_test_oneway): ...this, and
17768 update for underlying renaming.
17769 (selftest::levenshtein_distance_unit_test): Rename to...
17770 (selftest::test_get_edit_distance_unit): ...this, and update for
17771 underlying renaming.
17772 (selftest::test_find_closest_string): Add example from PR 69968
17773 where transposition helps
17774 (selftest::test_metric_conditions): Update for renaming.
17775 (selftest::test_metric_conditions): Likewise.
17776 (selftest::spellcheck_c_tests): Likewise.
17777 * spellcheck.h (levenshtein_distance): Rename both overloads to...
17778 (get_edit_distance): ...this.
17779 (best_match::consider): Update for renaming.
17781 2018-06-12 Martin Sebor <msebor@redhat.com>
17783 PR tree-optimization/85259
17784 * builtins.c (compute_objsize): Handle constant offsets.
17785 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
17786 true iff a warning has been issued.
17787 * gimple.h (gimple_nonartificial_location): New function.
17788 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
17789 gimple_nonartificial_location and handle -Wno-system-headers.
17790 (handle_builtin_stxncpy): Same.
17792 2018-06-12 Martin Sebor <msebor@redhat.com>
17795 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
17797 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17799 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17800 BUILTIN_VEC_XST entries for pointer to double and long long.
17802 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
17805 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
17807 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
17810 2018-06-12 Martin Liska <mliska@suse.cz>
17812 * doc/options.texi: Document IntegerRange.
17814 2018-06-12 Martin Liska <mliska@suse.cz>
17816 * config/i386/i386.opt: Make MPX-related options as Deprecated.
17817 * opt-functions.awk: Handle Deprecated flag.
17818 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
17820 (read_cmdline_option): Report warning for a deprecated option.
17821 * opts.h (struct cl_option): Add new field cl_deprecated.
17822 (CL_ERR_DEPRECATED): New.
17824 2018-06-12 Martin Liska <mliska@suse.cz>
17826 * doc/options.texi: Document Deprecated option flag.
17828 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
17830 * config/arc/arc-arch.h (arc_extras): New enum.
17831 (arc_cpu_t):Add field extra.
17832 (arc_cpu_types): Consider the extras.
17833 * config/arc/arc-cpus.def: Add extras info.
17834 * config/arc/arc-opts.h (processor_type): Consider extra field.
17835 * config/arc/arc.c (arc_override_options): Handle extra field.
17837 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
17839 * config/arc/arc-arch.h: Update ARC_OPTX macro.
17840 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
17842 * config/arc/arc.c (arc_init): Update pic warning.
17843 (irq_range): Update irq range parsing warnings.
17844 (arc_override_options): Update various warning messages.
17845 (arc_handle_aux_attribute): Likewise.
17847 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
17849 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
17851 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
17853 * doc/sourcebuild.texi: Document usage of line number 0 in verify
17854 compiler messages directives.
17856 2018-06-12 Matthew Fortune <mfortune@gmail.com>
17858 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
17859 * config/mips/mips-tables.opt: Regenerate.
17860 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
17862 * doc/invoke.texi: Document -march=i6500.
17864 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
17866 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
17867 (i6400_gpmul): Add cpu_unit.
17868 (i6400_gpdiv): Likewise.
17869 (i6400_msa_add_d): Update reservations.
17870 (i6400_msa_int_add) Likewise.
17871 (i6400_msa_short_logic3) Likewise.
17872 (i6400_msa_short_logic2) Likewise.
17873 (i6400_msa_short_logic) Likewise.
17874 (i6400_msa_move) Likewise.
17875 (i6400_msa_cmp) Likewise.
17876 (i6400_msa_short_float2) Likewise.
17877 (i6400_msa_div_d) Likewise.
17878 (i6400_msa_long_logic1) Likewise.
17879 (i6400_msa_long_logic2) Likewise.
17880 (i6400_msa_mult) Likewise.
17881 (i6400_msa_long_float2) Likewise.
17882 (i6400_msa_long_float4) Likewise.
17883 (i6400_msa_long_float5) Likewise.
17884 (i6400_msa_long_float8) Likewise.
17885 (i6400_fpu_fadd): Include frint type.
17886 (i6400_fpu_store): New define_insn_reservation.
17887 (i6400_fpu_load): Likewise.
17888 (i6400_fpu_move): Likewise.
17889 (i6400_fpu_fcmp): Likewise.
17890 (i6400_fpu_fmadd): Likewise.
17891 (i6400_int_mult): Include imul3nc type and update reservation.
17892 (i6400_int_div): Include idiv3 type and update reservation.
17893 (i6400_int_load): Update to check type not move_type.
17894 (i6400_int_store): Likewise.
17895 (i6400_int_prefetch): Set zero latency.
17897 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
17899 * gcc.c: Document new %@{...} sequence.
17900 (LINK_COMMAND_SPEC): Use it for the -L switches.
17901 (cpp_unique_options): Use it for the -I switches.
17902 (at_file_argbuf): New global variable.
17903 (in_at_file): Likewise.
17904 (alloc_args): Create at_file_argbuf.
17905 (clear_args): Truncate at_file_argbuf.
17906 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
17907 (open_at_file): New function.
17908 (close_at_file): Likewise.
17909 (create_at_file): Delete.
17910 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
17913 (validate_switches_from_spec): Deal with %@{...} sequence.
17914 (validate_switches): Likewise.
17915 (driver::finalize): Call clear_args.
17917 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
17919 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
17921 2018-06-11 Martin Sebor <msebor@redhat.com>
17923 * doc/invoke.texi (-Wall): List -Wc++17-compat.
17924 (Wno-class-memaccess): Add @opindex.
17925 (Wno-templates, Wno-multiple-inheritance): Same.
17926 (Wno-virtual-inheritance, Wno-namespaces): Same.
17927 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
17928 (Wno-format-overflow, Wno-format-truncation): Same.
17929 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
17930 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
17931 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
17932 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
17933 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
17934 (Wno-misspelled-isr): Same.
17936 2018-06-11 Martin Sebor <msebor@redhat.com>
17938 * PR tree-optimization/86083
17939 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
17941 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
17943 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
17945 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
17948 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
17949 on the correct operand.
17950 (*movdi_internal64): Ditto.
17952 2018-06-11 Martin Liska <mliska@suse.cz>
17954 PR tree-optimization/86089
17955 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
17957 2018-06-11 Julia Koval <julia.koval@intel.com>
17959 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
17960 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
17961 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
17963 2018-06-11 Olivier Hainque <hainque@adacore.com>
17965 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
17966 for Ada with strict dwarf2.
17968 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
17971 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
17974 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
17976 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
17978 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
17980 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
17983 2018-06-08 Martin Liska <mliska@suse.cz>
17985 * tree-cfg.h (debug_function): Fix argument type to match
17988 2018-06-08 Martin Liska <mliska@suse.cz>
17990 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
17991 Remove usage of MPX-related (and removed) fields.
17992 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
17994 2018-06-08 David Malcolm <dmalcolm@redhat.com>
17996 * cfg.c (debug): Use TDF_NONE rather than 0.
17997 * cfghooks.c (debug): Likewise.
17998 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
17999 (struct dump_option_value_info): Convert to...
18000 (struct kv_pair): ...this template type.
18001 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
18003 (optinfo_verbosity_options): Likewise.
18004 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
18006 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
18007 than int for "optgroup_flags" param.
18008 (dump_generic_expr_loc): Use dump_flags_t rather than int for
18010 (dump_dec): Likewise.
18011 (dump_finish): Use TDF_NONE rather than 0.
18012 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
18013 rather than int for "optgroup_flags" param. Use TDF_NONE rather
18014 than 0. Update for change to option_ptr.
18015 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
18016 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
18017 0. Update for changes to optinfo_verbosity_options and
18019 (opt_info_switch_p): Convert optgroup_flags from int to
18021 (dump_basic_block): Use dump_flags_t rather than int
18022 for "dump_kind" param.
18023 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
18024 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
18025 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
18026 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
18027 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
18028 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
18029 TDF_NONE): Convert from macros to...
18030 (enum dump_flag): ...this new enum.
18031 (dump_flags_t): Update to use enum.
18032 (operator|, operator&, operator~, operator|=, operator&=):
18033 Implement for dump_flags_t.
18034 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
18035 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
18036 Convert from macros to...
18037 (enum optgroup_flag): ...this new enum.
18038 (optgroup_flags_t): New typedef.
18039 (operator|, operator|=): Implement for optgroup_flags_t.
18040 (struct dump_file_info): Convert field "alt_flags" to
18041 dump_flags_t. Convert field "optgroup_flags" to
18043 (dump_basic_block): Use dump_flags_t rather than int for param.
18044 (dump_generic_expr_loc): Likewise.
18045 (dump_dec): Likewise.
18046 (dump_register): Convert param "optgroup_flags" to
18048 (opt_info_enable_passes): Likewise.
18049 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
18051 * gimple-pretty-print.c (debug): Likewise.
18052 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
18053 (merged_store_group::apply_stores): Likewise.
18054 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
18055 * gimple.c (verify_gimple_pp): Likewise.
18056 * graphite-poly.c (print_pbb_body): Likewise.
18057 * passes.c (pass_manager::register_one_dump_file): Convert
18058 local "optgroup_flags" to optgroup_flags_t.
18059 * print-tree.c (print_node): Use TDF_NONE rather than 0.
18061 (debug_body): Likewise.
18062 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
18063 to optgroup_flags_t.
18064 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
18066 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
18067 (convert_mult_to_fma): Likewise.
18068 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
18069 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
18070 * tree-vect-data-refs.c (dump_lower_bound): Convert param
18071 "dump_kind" to dump_flags_t.
18073 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
18075 * config/rs6000/rs6000.c (min, max): Delete.
18077 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
18079 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
18080 -mabi=spe and -mabi=no-spe.
18082 2018-06-08 Martin Liska <mliska@suse.cz>
18084 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
18085 where we expect an existing summary.
18087 2018-06-08 Martin Liska <mliska@suse.cz>
18089 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
18090 * ipa-inline.h (estimate_edge_growth): Likewise.
18092 2018-06-08 Martin Liska <mliska@suse.cz>
18094 * cgraph.c (function_version_hasher::hash): Use
18095 cgraph_node::get_uid ().
18096 (function_version_hasher::equal):
18097 * cgraph.h (cgraph_node::get_uid): New method.
18098 * ipa-inline.c (update_caller_keys): Use
18099 cgraph_node::get_uid ().
18100 (update_callee_keys): Likewise.
18101 * ipa-utils.c (searchc): Likewise.
18102 (ipa_reduced_postorder): Likewise.
18103 * lto-cgraph.c (input_node): Likewise.
18104 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
18105 * symbol-summary.h (symtab_insertion): Likewise.
18106 (symtab_removal): Likewise.
18107 (symtab_duplication): Likewise.
18108 * tree-pretty-print.c (dump_function_header): Likewise.
18109 * tree-sra.c (convert_callers_for_node): Likewise.
18111 2018-06-08 Martin Liska <mliska@suse.cz>
18113 * cgraph.c (symbol_table::create_edge): Always assign a new
18115 (symbol_table::free_edge): Do not recycle numbers.
18116 * cgraph.h (cgraph_edge::get): New method.
18117 * symbol-summary.h (symtab_removal): Use it.
18118 (symtab_duplication): Likewise.
18119 (call_summary::hashable_uid): Remove.
18121 2018-06-08 Martin Liska <mliska@suse.cz>
18123 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
18124 (initialize_growth_caches): Remove.
18125 (free_growth_caches): Likewise.
18126 (do_estimate_edge_time): Use edge_growth_cache.
18127 (do_estimate_edge_size): Likewise.
18128 (do_estimate_edge_hints): Likewise.
18129 * ipa-inline.c (reset_edge_caches): Likewise.
18130 (recursive_inlining): Likewise.
18131 (inline_small_functions): Likewise.
18132 * ipa-inline.h (initialize_growth_caches): Remove.
18133 (estimate_edge_size): Likewise.
18134 (estimate_edge_time): Likewise.
18135 (estimate_edge_hints): Likewise.
18136 (reset_edge_growth_cache): Likewise.
18137 * symbol-summary.h (call_summary::remove): New method.
18139 2018-06-08 Martin Liska <mliska@suse.cz>
18141 * ipa-cp.c (class edge_clone_summary): New summary.
18142 (grow_edge_clone_vectors): Remove.
18143 (ipcp_edge_duplication_hook): Remove.
18144 (class edge_clone_summary_t): New call_summary class.
18145 (ipcp_edge_removal_hook): Remove.
18146 (edge_clone_summary_t::duplicate): New function.
18147 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
18148 (create_specialized_node): Likewise.
18149 (ipcp_driver): Initialize edge_clone_summaries and do not
18152 2018-06-08 Martin Liska <mliska@suse.cz>
18154 * symbol-summary.h (get): New function.
18155 (call_summary::m_initialize_when_cloning): New class member.
18157 2018-06-08 Martin Liska <mliska@suse.cz>
18159 * cgraph.c (cgraph_node::remove): Do not recycle uid.
18160 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
18161 (symbol_table::allocate_cgraph_symbol): Do not set uid.
18162 * passes.c (uid_hash_t): Record removed_nodes by their uids.
18163 (remove_cgraph_node_from_order): Use the removed_nodes set.
18164 (do_per_function_toporder): Likwise.
18165 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
18166 instead of summary_uid.
18167 (symtab_removal): Likewise.
18168 (symtab_duplication): Likewise.
18170 2018-06-08 Martin Liska <mliska@suse.cz>
18172 * ipa-cp.c (ipcp_store_bits_results): Use
18173 ipcp_transformation_sum.
18174 (ipcp_store_vr_results): Likewise.
18175 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
18177 (ipcp_transformation_initialize): ... this.
18178 (ipa_set_node_agg_value_chain):
18179 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
18180 (write_ipcp_transformation_info): Likewise.
18181 (read_ipcp_transformation_info): Likewise.
18182 (ipcp_update_bits): Likewise.
18183 (ipcp_update_vr): Likewise.
18184 (ipcp_transform_function): Likewise.
18185 * ipa-prop.h: Rename ipcp_transformation_summary to
18186 ipcp_transformation.
18187 (class ipcp_transformation_t): New function summary.
18188 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
18189 (ipa_get_agg_replacements_for_node): Likewise.
18191 2018-06-08 Martin Liska <mliska@suse.cz>
18193 * ipa-pure-const.c (struct funct_state_d): Do it class instead
18195 (class funct_state_summary_t): New function_summary class.
18196 (has_function_state): Remove.
18197 (get_function_state): Likewise.
18198 (set_function_state): Likewise.
18199 (add_new_function): Likewise.
18200 (funct_state_summary_t::insert): New function.
18201 (duplicate_node_data): Remove.
18202 (remove_node_data): Remove.
18203 (funct_state_summary_t::duplicate): New function.
18204 (register_hooks): Create new funct_state_summaries.
18205 (pure_const_generate_summary): Use it.
18206 (pure_const_write_summary): Likewise.
18207 (pure_const_read_summary): Likewise.
18208 (propagate_pure_const): Likewise.
18209 (propagate_nothrow): Likewise.
18210 (dump_malloc_lattice): Likewise.
18211 (propagate_malloc): Likewise.
18212 (execute): Do not register hooks, just remove summary
18214 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
18217 2018-06-08 Martin Liska <mliska@suse.cz>
18219 * ipa-reference.c (remove_node_data): Remove.
18220 (duplicate_node_data): Likewise.
18221 (class ipa_ref_var_info_summary_t): New class.
18222 (class ipa_ref_opt_summary_t): Likewise.
18223 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
18224 (get_reference_optimization_summary): Use
18225 ipa_ref_opt_sum_summaries.
18226 (set_reference_vars_info): Remove.
18227 (set_reference_optimization_summary): Likewise.
18228 (ipa_init): Create summaries.
18229 (init_function_info): Use function summary.
18230 (ipa_ref_opt_summary_t::duplicate): New function.
18231 (ipa_ref_opt_summary_t::remove): New function.
18232 (get_read_write_all_from_node): Fix GNU coding style.
18233 (propagate): Use function summary.
18234 (write_node_summary_p): Fix GNU coding style.
18235 (stream_out_bitmap): Likewise.
18236 (ipa_reference_read_optimization_summary): Use function summary.
18237 (ipa_reference_c_finalize): Do not release hooks.
18239 2018-06-08 Martin Liska <mliska@suse.cz>
18241 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
18242 (analyze_function_body): Extract multiple calls of get_create.
18243 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
18244 * ipa-inline.c (recursive_inlining): Use ::get method.
18245 * ipa-inline.h (estimate_edge_growth): Likewise.
18247 2018-06-08 Martin Liska <mliska@suse.cz>
18249 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
18251 (hsa_function_summary::hsa_function_summary): Use the new enum
18253 (hsa_gpu_implementation_p): Use hsa_summaries::get.
18254 * hsa-gen.c (hsa_get_host_function): Likewise.
18255 (get_brig_function_name): Likewise.
18256 * ipa-hsa.c (process_hsa_functions): Likewise.
18257 (ipa_hsa_write_summary): Likewise.
18258 * symbol-summary.h (symtab_duplication): Use ::get function/
18259 (get): New function.
18261 2018-06-08 Martin Liska <mliska@suse.cz>
18263 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
18265 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
18266 (hsa_register_kernel): Likewise.
18267 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
18268 * hsa-gen.c (hsa_get_host_function): Likewise.
18269 (get_brig_function_name): Likewise.
18270 (generate_hsa): Likewise.
18271 (pass_gen_hsail::execute): Likewise.
18272 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
18273 (devirtualization_time_bonus): Likewise.
18274 (ipcp_propagate_stage): Likewise.
18275 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
18276 (edge_set_predicate): Likewise.
18277 (evaluate_conditions_for_known_args): Likewise.
18278 (evaluate_properties_for_edge): Likewise.
18279 (ipa_fn_summary::reset): Likewise.
18280 (ipa_fn_summary_t::duplicate): Likewise.
18281 (dump_ipa_call_summary): Likewise.
18282 (ipa_dump_fn_summary): Likewise.
18283 (analyze_function_body): Likewise.
18284 (compute_fn_summary): Likewise.
18285 (estimate_edge_devirt_benefit): Likewise.
18286 (estimate_edge_size_and_time): Likewise.
18287 (estimate_calls_size_and_time): Likewise.
18288 (estimate_node_size_and_time): Likewise.
18289 (inline_update_callee_summaries): Likewise.
18290 (remap_edge_change_prob): Likewise.
18291 (remap_edge_summaries): Likewise.
18292 (ipa_merge_fn_summary_after_inlining): Likewise.
18293 (ipa_update_overall_fn_summary): Likewise.
18294 (read_ipa_call_summary): Likewise.
18295 (inline_read_section): Likewise.
18296 (write_ipa_call_summary): Likewise.
18297 (ipa_fn_summary_write): Likewise.
18298 (ipa_free_fn_summary): Likewise.
18299 * ipa-hsa.c (process_hsa_functions): Likewise.
18300 (ipa_hsa_write_summary): Likewise.
18301 (ipa_hsa_read_section): Likewise.
18302 * ipa-icf.c (sem_function::merge): Likewise.
18303 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
18304 (do_estimate_edge_time): Likewise.
18305 (estimate_size_after_inlining): Likewise.
18306 (estimate_growth): Likewise.
18307 (growth_likely_positive): Likewise.
18308 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
18309 (inline_call): Likewise.
18310 * ipa-inline.c (caller_growth_limits): Likewise.
18311 (can_inline_edge_p): Likewise.
18312 (can_inline_edge_by_limits_p): Likewise.
18313 (compute_uninlined_call_time): Likewise.
18314 (compute_inlined_call_time): Likewise.
18315 (want_inline_small_function_p): Likewise.
18316 (edge_badness): Likewise.
18317 (update_caller_keys): Likewise.
18318 (update_callee_keys): Likewise.
18319 (recursive_inlining): Likewise.
18320 (inline_small_functions): Likewise.
18321 (inline_to_all_callers_1): Likewise.
18322 (dump_overall_stats): Likewise.
18323 (early_inline_small_functions): Likewise.
18324 (early_inliner): Likewise.
18325 * ipa-inline.h (estimate_edge_growth): Likewise.
18326 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
18327 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
18328 * ipa-prop.h (IPA_NODE_REF): Likewise.
18329 (IPA_EDGE_REF): Likewise.
18330 * ipa-pure-const.c (malloc_candidate_p): Likewise.
18331 (propagate_malloc): Likewise.
18332 * ipa-split.c (execute_split_functions): Likewise.
18333 * symbol-summary.h: Rename get to get_create.
18335 (get_create): Likewise.
18336 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
18338 2018-06-08 Martin Liska <mliska@suse.cz>
18340 * symbol-summary.h (release): Move definition out of class
18342 (symtab_removal): Likewise.
18343 (symtab_duplication): Likewise.
18345 2018-06-08 Martin Liska <mliska@suse.cz>
18347 * symbol-summary.h (function_summary): Move constructor
18348 implementation out of class declaration.
18349 (release): Likewise.
18350 (symtab_insertion): Likewise.
18351 (symtab_removal): Likewise.
18352 (symtab_duplication): Likewise.
18355 2018-06-08 Martin Liska <mliska@suse.cz>
18357 * Makefile.in: Remove support for MPX (macros, related functions,
18358 fields in cgraph_node, ...).
18359 * builtin-types.def (BT_BND): Likewise.
18360 (BT_FN_BND_CONST_PTR): Likewise.
18361 (BT_FN_CONST_PTR_BND): Likewise.
18362 (BT_FN_VOID_PTR_BND): Likewise.
18363 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
18364 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
18365 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
18366 (expand_builtin_mempcpy_with_bounds): Likewise.
18367 (expand_builtin_memset_with_bounds): Likewise.
18368 (expand_builtin_memset_args): Likewise.
18369 (std_expand_builtin_va_start): Likewise.
18370 (expand_builtin): Likewise.
18371 (expand_builtin_with_bounds): Likewise.
18372 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
18373 (DEF_LIB_BUILTIN_CHKP): Likewise.
18374 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
18375 (DEF_CHKP_BUILTIN): Likewise.
18376 (BUILT_IN_MEMCPY): Likewise.
18377 (BUILT_IN_MEMMOVE): Likewise.
18378 (BUILT_IN_MEMPCPY): Likewise.
18379 (BUILT_IN_MEMSET): Likewise.
18380 (BUILT_IN_STPCPY): Likewise.
18381 (BUILT_IN_STRCAT): Likewise.
18382 (BUILT_IN_STRCHR): Likewise.
18383 (BUILT_IN_STRCPY): Likewise.
18384 (BUILT_IN_STRLEN): Likewise.
18385 (BUILT_IN_MEMCPY_CHK): Likewise.
18386 (BUILT_IN_MEMMOVE_CHK): Likewise.
18387 (BUILT_IN_MEMPCPY_CHK): Likewise.
18388 (BUILT_IN_MEMSET_CHK): Likewise.
18389 (BUILT_IN_STPCPY_CHK): Likewise.
18390 (BUILT_IN_STRCAT_CHK): Likewise.
18391 (BUILT_IN_STRCPY_CHK): Likewise.
18392 * calls.c (store_bounds): Likewise.
18393 (emit_call_1): Likewise.
18394 (special_function_p): Likewise.
18395 (maybe_warn_nonstring_arg): Likewise.
18396 (initialize_argument_information): Likewise.
18397 (finalize_must_preallocate): Likewise.
18398 (compute_argument_addresses): Likewise.
18399 (expand_call): Likewise.
18400 * cfgexpand.c (expand_call_stmt): Likewise.
18401 (expand_return): Likewise.
18402 (expand_gimple_stmt_1): Likewise.
18403 (pass_expand::execute): Likewise.
18404 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
18405 (cgraph_node::remove): Likewise.
18406 (cgraph_node::dump): Likewise.
18407 (cgraph_node::verify_node): Likewise.
18408 * cgraph.h (chkp_function_instrumented_p): Likewise.
18409 (symtab_node::get_alias_target): Likewise.
18410 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
18411 (cgraph_local_p): Likewise.
18412 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
18413 (cgraph_edge::rebuild_references): Likewise.
18414 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
18415 (walk_polymorphic_call_targets): Likewise.
18416 (cgraph_node::expand_thunk): Likewise.
18417 (symbol_table::output_weakrefs): Likewise.
18418 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
18419 (ix86_handle_option): Likewise.
18420 * config/i386/constraints.md: Likewise.
18421 * config/i386/i386-builtin-types.def (BND): Likewise.
18425 * config/i386/i386-builtin.def (BDESC_END): Likewise.
18426 (BDESC_FIRST): Likewise.
18428 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
18429 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
18430 * config/i386/i386.c (enum reg_class): Likewise.
18431 (ix86_target_string): Likewise.
18432 (ix86_option_override_internal): Likewise.
18433 (ix86_conditional_register_usage): Likewise.
18434 (ix86_valid_target_attribute_inner_p): Likewise.
18435 (ix86_set_indirect_branch_type): Likewise.
18436 (ix86_set_current_function): Likewise.
18437 (ix86_function_arg_regno_p): Likewise.
18438 (init_cumulative_args): Likewise.
18439 (ix86_function_arg_advance): Likewise.
18440 (ix86_function_arg): Likewise.
18441 (ix86_pass_by_reference): Likewise.
18442 (ix86_function_value_regno_p): Likewise.
18443 (ix86_function_value_1): Likewise.
18444 (ix86_function_value_bounds): Likewise.
18445 (ix86_return_in_memory): Likewise.
18446 (ix86_setup_incoming_vararg_bounds): Likewise.
18447 (ix86_va_start): Likewise.
18448 (indirect_thunk_need_prefix): Likewise.
18449 (print_reg): Likewise.
18450 (ix86_print_operand): Likewise.
18451 (ix86_expand_call): Likewise.
18452 (ix86_output_function_return): Likewise.
18453 (reg_encoded_number): Likewise.
18454 (BDESC_VERIFYS): Likewise.
18455 (ix86_init_mpx_builtins): Likewise.
18456 (ix86_init_builtins): Likewise.
18457 (ix86_emit_cmove): Likewise.
18458 (ix86_emit_move_max): Likewise.
18459 (ix86_expand_builtin): Likewise.
18460 (ix86_builtin_mpx_function): Likewise.
18461 (ix86_get_arg_address_for_bt): Likewise.
18462 (ix86_load_bounds): Likewise.
18463 (ix86_store_bounds): Likewise.
18464 (ix86_load_returned_bounds): Likewise.
18465 (ix86_store_returned_bounds): Likewise.
18466 (ix86_class_likely_spilled_p): Likewise.
18467 (ix86_hard_regno_mode_ok): Likewise.
18468 (x86_order_regs_for_local_alloc): Likewise.
18469 (ix86_mitigate_rop): Likewise.
18470 (ix86_bnd_prefixed_insn_p): Likewise.
18471 (ix86_mpx_bound_mode): Likewise.
18472 (ix86_make_bounds_constant): Likewise.
18473 (ix86_initialize_bounds): Likewise.
18474 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
18475 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
18476 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
18477 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
18478 (TARGET_CHKP_BOUND_MODE): Likewise.
18479 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
18480 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
18481 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
18482 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
18483 * config/i386/i386.h (TARGET_MPX): Likewise.
18484 (TARGET_MPX_P): Likewise.
18485 (VALID_BND_REG_MODE): Likewise.
18486 (FIRST_BND_REG): Likewise.
18487 (LAST_BND_REG): Likewise.
18488 (enum reg_class): Likewise.
18489 (BND_REG_P): Likewise.
18490 (BND_REGNO_P): Likewise.
18491 (BNDmode): Likewise.
18492 (ADJUST_INSN_LENGTH): Likewise.
18493 * config/i386/i386.md: Likewise.
18494 * config/i386/i386.opt: Likewise.
18495 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
18496 (defined): Likewise.
18497 (LINK_MPX): Likewise.
18498 (MPX_SPEC): Likewise.
18499 (LIBMPX_SPEC): Likewise.
18500 (LIBMPXWRAPPERS_SPEC): Likewise.
18501 (CHKP_SPEC): Likewise.
18502 * config/i386/predicates.md: Likewise.
18503 * dbxout.c (dbxout_type): Likewise.
18504 * doc/extend.texi: Likewise.
18505 * doc/invoke.texi: Likewise.
18506 * doc/md.texi: Likewise.
18507 * doc/tm.texi: Likewise.
18508 * doc/tm.texi.in: Likewise.
18509 * dwarf2out.c (is_base_type): Likewise.
18510 (gen_formal_types_die): Likewise.
18511 (gen_subprogram_die): Likewise.
18512 (gen_type_die_with_usage): Likewise.
18513 (gen_decl_die): Likewise.
18514 (dwarf2out_late_global_decl): Likewise.
18515 * expr.c (expand_assignment): Likewise.
18516 (emit_storent_insn): Likewise.
18517 (store_expr_with_bounds): Likewise.
18518 (store_expr): Likewise.
18519 (expand_expr_real_1): Likewise.
18520 * expr.h (store_expr_with_bounds): Likewise.
18521 * function.c (use_register_for_decl): Likewise.
18522 (struct bounds_parm_data): Likewise.
18523 (assign_parms_augmented_arg_list): Likewise.
18524 (assign_parm_find_entry_rtl): Likewise.
18525 (assign_parm_is_stack_parm): Likewise.
18526 (assign_parm_load_bounds): Likewise.
18527 (assign_bounds): Likewise.
18528 (assign_parms): Likewise.
18529 (expand_function_start): Likewise.
18530 * gcc.c (CHKP_SPEC): Likewise.
18531 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
18532 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
18533 (wrestrict_dom_walker::check_call): Likewise.
18534 * gimple.c (gimple_build_call_from_tree): Likewise.
18535 * gimple.h (enum gf_mask): Likewise.
18536 (gimple_call_with_bounds_p): Likewise.
18537 (gimple_call_set_with_bounds): Likewise.
18538 * gimplify.c (gimplify_init_constructor): Likewise.
18539 * ipa-cp.c (initialize_node_lattices): Likewise.
18540 (propagate_constants_across_call): Likewise.
18541 (find_more_scalar_values_for_callers_subset): Likewise.
18542 * ipa-hsa.c (process_hsa_functions): Likewise.
18543 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
18544 * ipa-icf.c (sem_function::merge): Likewise.
18545 * ipa-inline.c (early_inliner): Likewise.
18546 * ipa-pure-const.c (warn_function_noreturn): Likewise.
18547 (warn_function_cold): Likewise.
18548 (propagate_pure_const): Likewise.
18549 * ipa-ref.h (enum GTY): Likewise.
18550 * ipa-split.c (find_retbnd): Likewise.
18551 (consider_split): Likewise.
18552 (split_function): Likewise.
18553 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
18554 * ipa.c (walk_polymorphic_call_targets): Likewise.
18555 (symbol_table::remove_unreachable_nodes): Likewise.
18556 (process_references): Likewise.
18557 (cgraph_build_static_cdtor_1): Likewise.
18558 * lto-cgraph.c (lto_output_node): Likewise.
18559 (output_refs): Likewise.
18560 (compute_ltrans_boundary): Likewise.
18561 (input_overwrite_node): Likewise.
18562 (input_node): Likewise.
18563 (input_cgraph_1): Likewise.
18564 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
18565 * passes.c (pass_manager::execute_early_local_passes): Likewise.
18566 (class pass_chkp_instrumentation_passes): Likewise.
18567 (make_pass_chkp_instrumentation_passes): Likewise.
18568 * passes.def: Likewise.
18569 * rtl.h (struct GTY): Likewise.
18570 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
18571 * stor-layout.c (layout_type): Likewise.
18572 * symtab.c: Likewise.
18573 * target.def: Likewise.
18574 * targhooks.c (default_chkp_bound_type): Likewise.
18575 (default_chkp_bound_mode): Likewise.
18576 (default_builtin_chkp_function): Likewise.
18577 (default_chkp_function_value_bounds): Likewise.
18578 (default_chkp_make_bounds_constant): Likewise.
18579 (default_chkp_initialize_bounds): Likewise.
18580 * targhooks.h (default_chkp_bound_type): Likewise.
18581 (default_chkp_bound_mode): Likewise.
18582 (default_builtin_chkp_function): Likewise.
18583 (default_chkp_function_value_bounds): Likewise.
18584 (default_chkp_make_bounds_constant): Likewise.
18585 (default_chkp_initialize_bounds): Likewise.
18586 * toplev.c (compile_file): Likewise.
18587 (process_options): Likewise.
18588 * tree-core.h (DEF_BUILTIN): Likewise.
18589 (DEF_BUILTIN_CHKP): Likewise.
18590 * tree-inline.c (declare_return_variable): Likewise.
18591 (remap_gimple_stmt): Likewise.
18592 (copy_bb): Likewise.
18593 (initialize_inlined_parameters): Likewise.
18594 (expand_call_inline): Likewise.
18595 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
18596 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
18597 (make_pass_ipa_chkp_produce_thunks): Likewise.
18598 (make_pass_chkp): Likewise.
18599 (make_pass_chkp_opt): Likewise.
18600 (make_pass_chkp_instrumentation_passes): Likewise.
18601 * tree-pretty-print.c (dump_generic_node): Likewise.
18602 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
18603 * tree-ssa-dce.c (propagate_necessity): Likewise.
18604 (eliminate_unnecessary_stmts): Likewise.
18605 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
18606 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
18607 * tree-ssa-sccvn.h: Likewise.
18608 * tree-ssa-strlen.c (get_string_length): Likewise.
18609 (valid_builtin_call): Likewise.
18610 (adjust_last_stmt): Likewise.
18611 (handle_builtin_strchr): Likewise.
18612 (handle_builtin_strcpy): Likewise.
18613 (handle_builtin_stxncpy): Likewise.
18614 (handle_builtin_memcpy): Likewise.
18615 (handle_builtin_strcat): Likewise.
18616 (strlen_check_and_optimize_stmt): Likewise.
18617 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
18618 * tree-streamer-in.c: Likewise.
18619 * tree-streamer.c (record_common_node): Likewise.
18620 * tree.c (tree_code_size): Likewise.
18621 (wide_int_to_tree_1): Likewise.
18622 (type_contains_placeholder_1): Likewise.
18623 (build_common_tree_nodes): Likewise.
18624 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
18625 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
18626 (POINTER_BOUNDS_P): Likewise.
18627 (BOUNDED_TYPE_P): Likewise.
18628 (BOUNDED_P): Likewise.
18629 (CALL_WITH_BOUNDS_P): Likewise.
18630 (pointer_bounds_type_node): Likewise.
18631 * value-prof.c (gimple_ic): Likewise.
18632 * var-tracking.c (vt_add_function_parameters): Likewise.
18633 * varasm.c (make_decl_rtl): Likewise.
18634 (assemble_start_function): Likewise.
18635 (output_constant): Likewise.
18636 (maybe_assemble_visibility): Likewise.
18637 * varpool.c (ctor_for_folding): Likewise.
18638 * chkp-builtins.def: Remove.
18639 * ipa-chkp.c: Remove.
18640 * ipa-chkp.h: Remove.
18641 * rtl-chkp.c: Remove.
18642 * rtl-chkp.h: Remove.
18643 * tree-chkp-opt.c: Remove.
18644 * tree-chkp.c: Remove.
18645 * tree-chkp.h: Remove.
18647 2018-06-07 Carl Love <cel@us.ibm.com>
18649 * config/rs6000/vsx.md (vextract_fp_from_shorth,
18650 vextract_fp_from_shortl): Add BE support.
18652 2018-06-07 Paul Koning <ni1d@arrl.net>
18654 * compare-elim.c (try_merge_compare): Don't merge compare if
18655 address contains a side effect.
18656 (try_eliminate_compare): Likewise.
18658 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
18660 * config.gcc: Support "tremont".
18661 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
18662 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
18664 * config/i386/i386.c (m_TREMONT): Define.
18665 (processor_target_table): Add "tremont".
18666 (PTA_TREMONT): Define.
18667 (ix86_lea_outperforms): Add TARGET_TREMONT.
18668 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
18669 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
18670 and M_INTEL_GOLDMONT_PLUS.
18671 (fold_builtin_cpu): Add "tremont".
18672 (ix86_add_stmt_cost): Add TARGET_TREMONT.
18673 (ix86_option_override_internal): Add "tremont".
18674 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
18675 (processor_type): Add PROCESSOR_TREMONT.
18676 * config/i386/x86-tune.def: Add m_TREMONT.
18677 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
18679 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
18681 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
18682 symbol defined for msp430i* devices to be lower case.
18684 2018-06-07 Richard Biener <rguenther@suse.de>
18686 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
18687 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
18688 Properly wrap signed arithmetic if overflow wraps.
18690 2018-06-07 Jakub Jelinek <jakub@redhat.com>
18692 PR tree-optimization/69615
18693 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
18694 of a cast from a same precision integral SSA_NAME in a bb dominated
18695 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
18696 cast to utype if rhs2 has already a compatible type.
18698 2018-06-07 Richard Biener <rguenther@suse.de>
18700 PR tree-optimization/85935
18701 * graphite-scop-detection.c (find_params_in_bb): Analyze
18702 condition operands with respect to the correct loop. Assert
18703 the analysis doesn't fail.
18705 2018-06-04 Carl Love <cel@us.ibm.com>
18707 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
18708 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
18709 as it is slightly cheaper.
18710 (first_match_or_eos_index_<mode>):
18711 Calculate index using natural element order.
18712 (first_match_index_<mode>):
18713 Calculate index using natural element order.
18714 (first_match_or_eos_index_<mode>):
18715 Calculate index using natural order.
18716 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
18717 for BE and LE modes.
18718 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
18719 P9V_BUILTIN_VCLZLSBB_V16QI.
18720 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
18723 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
18725 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
18726 indentation and line wrap for many prototypes. Add missing
18727 @smallexample directives around block of prototypes for vec_xl and
18730 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
18732 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
18733 track if we pass or return IEEE 128-bit floating point.
18734 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
18735 C++ mangling that is compatible with GCC 8.1.
18736 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
18737 (init_cumulative_args): Note if we pass or return IEEE 128-bit
18738 floating point types.
18739 (rs6000_function_arg_advance_1): Likewise.
18740 (rs6000_mangle_type): Optionally generate mangled names that match
18741 what GCC 8.1 generated for IEEE 128-bit floating point types.
18742 (rs6000_globalize_decl_name): If we have an external function that
18743 passes or returns IEEE 128-bit types, generate a weak reference
18744 from the mangled name used in GCC 8.1 to the current mangled
18746 (rs6000_init_builtins): Make __ibm128 use the long double type if
18747 long double is IBM extended double. Make __float128 use the long
18748 double type if long double is IEEE 128-bit.
18751 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
18752 macro for __ibm128 built-in functions.
18753 (PACK_IF): Add __ibm128 pack/unpack functions.
18754 (UNPACK_IF): Likewise.
18755 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
18756 enable long double built-in functions if long double is IEEE
18757 128-bit floating point.
18758 (rs6000_invalid_builtin): Update long double built-in function
18760 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
18761 functions, adjust the built-in function to use the long double
18762 built-in function if __ibm128 and long double are the same type.
18763 * doc/extend.texi (PowerPC builtins): Update documention for
18764 __builtin_{,un}pack_longdouble. Add documentation for
18765 __builtin_{,un}pack_ibm128.
18767 2018-06-06 Jim Wilson <jimw@sifive.com>
18769 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
18770 (struct machine_function): New field interrupt_mode.
18771 (riscv_handle_type_attribute): New function. Add forward declaration.
18772 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
18773 (riscv_expand_epilogue): Check interrupt_mode field.
18774 (riscv_set_current_function): Check interrupt attribute args and
18775 set interrupt_mode field.
18776 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
18777 (riscv_sret, riscv_uret): New.
18778 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
18779 new arguments to interrupt attribute.
18781 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
18784 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
18785 Don't handle -mcpu=power8 if -mpower9-vector is also used.
18787 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
18789 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
18790 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
18791 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
18792 several redundant entries.
18794 2018-06-06 David Malcolm <dmalcolm@redhat.com>
18796 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
18797 type from "rtx" to "rtx_insn *".
18798 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
18799 for local "call_insn", removing cast.
18800 (ix86_expand_call): Likewise, introducing a "call_insn" local.
18802 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
18804 PR tree-optimization/86066
18805 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
18806 for BIT_INSERT_EXPR stores.
18808 2018-06-06 Richard Biener <rguenther@suse.de>
18810 PR tree-optimization/86062
18811 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
18812 component refs ontop
18813 of to be offsetted base.
18815 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
18817 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
18818 to be static and remove check on interrupt attribute name.
18820 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
18822 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
18823 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
18825 2018-06-05 Steve Ellcey <sellcey@cavium.com>
18828 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
18830 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
18831 Remove second argument, change how error is called.
18832 (aarch64_layout_arg): Remove second argument from
18833 aarch64_err_no_fpadvsimd call.
18834 (aarch64_init_cumulative_args): Ditto.
18835 (aarch64_gimplify_va_arg_expr): Ditto.
18836 * config/aarch64/aarch64.md (mov<mode>): Ditto.
18838 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
18840 * config/i386/i386.md (simple_return_indirect_internal): New expander.
18841 (*simple_return_indirect_internal<mode>): Rename from
18842 simple_return_indirect_internal. Use W mode iterator.
18843 (rstorssp): New expander.
18844 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
18845 (clrssbsy): New expander.
18846 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
18848 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
18850 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
18852 (cmse_check_pointed_object): Likewise.
18854 2018-06-05 Martin Liska <mliska@suse.cz>
18856 PR gcov-profile/47618
18857 * doc/invoke.texi: Document how -fprofile-dir format
18860 2018-06-05 Richard Biener <rguenther@suse.de>
18862 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
18863 removal pretend DOM info isn't available so we do not update
18864 it and only remove edges, not dominated blocks. Actually free
18865 DOM info in case we removed something. Remove unreachable blocks.
18866 (mfb_keep_latches): Work with either DOM info or marked backedges.
18867 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
18868 first. Mark backedges if DOM info isn't available.
18869 (Re-)compute DOM info after cleanup_control_flow_pre.
18871 2018-06-05 Richard Biener <rguenther@suse.de>
18873 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
18874 (locus_discrim_hasher::hash): Adjust.
18875 (locus_discrim_hasher::equal): Likewise.
18876 (next_discriminator_for_locus): Work on line directly.
18877 (same_line_p): Pass in expanded locus1 as well.
18878 (assign_discriminators): Avoid redundant location expansions.
18880 2018-06-05 Richard Biener <rguenther@suse.de>
18882 PR tree-optimization/86046
18883 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
18884 if required after clearing TREE_ADDRESSABLE.
18886 2018-06-05 Richard Biener <rguenther@suse.de>
18888 PR tree-optimization/86047
18889 * tree-ssa-loop.c (for_each_index): Glob handling of all
18890 decls and constants and really handle all of them.
18892 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18895 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
18896 qualifier_void_pointer and qualifier_const_void_pointer.
18897 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
18898 (arm_init_builtins): Handle the above.
18899 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
18900 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
18901 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
18904 2018-06-05 Martin Liska <mliska@suse.cz>
18906 * auto-profile.c (read_autofdo_file): Do not use
18907 gcov_ctr_summary struct.
18908 (afdo_callsite_hot_enough_for_early_inline): Likewise.
18909 * coverage.c (struct counts_entry): Likewise.
18910 (read_counts_file): Read just single summary entry.
18911 (get_coverage_counts): Use gcov_summary struct.
18912 * coverage.h (get_coverage_counts): Likewise.
18913 * gcov-dump.c (dump_working_sets): Likewise.
18914 (tag_summary): Dump just single summary.
18915 * gcov-io.c (gcov_write_summary): Write just histogram
18917 (gcov_read_summary): Read just single summary.
18918 (compute_working_sets): Use gcov_summary struct.
18919 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
18920 of GCOV_COUNTERS_SUMMABLE.
18921 (GCOV_COUNTERS_SUMMABLE): Remove.
18922 (GCOV_FIRST_VALUE_COUNTER): Replace with
18923 GCOV_COUNTER_V_INTERVAL.
18924 (struct gcov_ctr_summary): Remove.
18925 (struct gcov_summary): Directly use fields of former
18927 (compute_working_sets): Use gcov_summary struct.
18928 * gcov.c (read_count_file): Do not use ctrs fields.
18929 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
18931 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
18933 * profile.c: Likewise.
18934 * profile.h: Likewise.
18936 2018-06-05 Martin Liska <mliska@suse.cz>
18938 PR gcov-profile/84846
18939 * gcov.c (output_lines): Print working directory only
18940 in intermediate format.
18942 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
18944 * config/s390/s390-builtin-types.def: Add void function type.
18945 * config/s390/s390-builtins.def: Use the function type for the
18948 2018-06-04 Jim Wilson <jimw@sifive.com>
18950 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
18952 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
18953 and maybe_eh_return. Change regno to unsigned int. Use new args to
18954 handle EH_RETURN_DATA_REGNO registers properly.
18955 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
18956 (riscv_expand_epilogue): Update comment. Change argument name and
18957 type. Update code to use new name and type. Pass new args to
18958 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
18960 * config/riscv/riscv.md (NORMAL_RETURN): New.
18961 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
18962 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
18963 (eh_return): Call gen_eh_return_internal and emit barrier.
18964 (eh_return_internal): Call riscv_expand_epilogue.
18966 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
18968 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
18969 bit_insertion field and declare can_be_merged_into method.
18970 (merged_store_group::can_be_merged_into): New method.
18971 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
18972 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
18973 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
18975 2018-06-04 Richard Biener <rguenther@suse.de>
18977 PR tree-optimization/85955
18978 * builtins.c (fold_builtin_sincos): Convert pointers to
18979 destination to appropriate type before dereferencing.
18981 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
18983 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
18985 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
18987 * expr.c (expand_expr_real_1): Force the operand into memory if
18988 its TYPE_MODE is BLKmode and if there is no integer mode for
18989 the number of bits being extracted.
18991 2018-06-04 Jakub Jelinek <jakub@redhat.com>
18995 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
18996 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
18998 2018-06-04 Richard Biener <rguenther@suse.de>
19000 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
19001 (cleanup_tree_cfg_noloop): ... single caller. Do
19002 start_recording_case_labels later.
19004 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
19006 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
19007 to _IMMINTRIN_H_INCLUDED.
19008 * config/i386/pconfigintrin.h: Ditto.
19009 * config/i386/waitpkgintrin.h: Ditto.
19010 * config/i386/immintrin.h: Add includes for sgxintrin.h,
19011 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
19012 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
19013 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
19014 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
19015 waitpkgintrin.h and cldemoteintrin.h.
19017 2018-06-04 Richard Biener <rguenther@suse.de>
19019 PR tree-optimization/86038
19020 * tracer.c (find_best_successor): Check probability for
19021 being initialized, bail out if not.
19023 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
19026 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
19027 of bits to ignore when comparing architectures.
19029 2018-06-04 Jakub Jelinek <jakub@redhat.com>
19031 PR tree-optimization/69615
19032 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
19033 maximum or minimum of the type, try to merge it also as if
19034 range1 is + [-, x - 1] or + [x + 1, -].
19037 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
19039 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
19041 PR tree-optimization/86034
19042 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
19043 the unsigned bitfield type in a bit insertion sequence if it does not
19044 have a larger precision than the bitfield size.
19045 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
19047 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
19049 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
19051 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
19053 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
19054 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
19055 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
19056 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
19058 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
19060 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
19061 Disable -fdelete-null-pointer-checks for ELF toolchain.
19063 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
19064 Kito Cheng <kito.cheng@gmail.com>
19066 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
19067 (nds32le-*-*, nds32be-*-*): Integrate checking process.
19068 (nds32*-*-*): Add glibc and uclibc conditions.
19069 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
19070 (TARGET_EXCEPT_UNWIND_INFO): Define.
19071 * config/nds32/elf.h: New file.
19072 * config/nds32/linux.h: New file.
19073 * config/nds32/nds32-elf.opt: New file.
19074 * config/nds32/nds32-linux.opt: New file.
19075 * config/nds32/nds32-fp-as-gp.c
19076 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
19077 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
19079 (nds32_asm_file_end): Ditto.
19080 (nds32_print_operand): Ditto.
19081 (nds32_insert_attributes): Ditto.
19082 (nds32_init_libfuncs): New function.
19083 (TARGET_HAVE_TLS): Define.
19084 (TARGET_INIT_LIBFUNCS): Define.
19085 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
19087 (TARGET_ELF): Apply different mcmodel setting.
19088 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
19089 been migrated into elf.h and linux.h files.
19090 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
19091 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
19092 (mcmodel): The content has been migrated into nds32-elf.opt and
19093 nds32-linux.opt files.
19094 * config/nds32/t-elf: New file.
19095 * config/nds32/t-linux: New file.
19097 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
19098 Shiva Chen <shiva0217@gmail.com>
19100 * config/nds32/constants.md (unspec_volatile_element): Add
19101 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
19102 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
19104 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
19105 (make_pass_nds32_fp_as_gp): Declare.
19106 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
19108 (nds32_asm_function_end_prologue): Remove unused asm output.
19109 (nds32_asm_function_begin_epilogue): Remove unused asm output.
19110 (nds32_asm_file_start): Output necessary fp_as_gp information.
19111 (nds32_option_override): Adjust register usage.
19112 (nds32_expand_prologue): Consider fp_as_gp situation.
19113 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
19114 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
19117 (simple_return): Ditto.
19118 (omit_fp_begin): Output special directive for fp_as_gp.
19119 (omit_fp_end): Output special directive for fp_as_gp.
19120 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
19121 mforbid-fp-as-gp): New options.
19123 2018-06-01 Mark Wielaard <mark@klomp.org>
19125 * dwarf2out.c (dwarf2out_finish): Remove generation of
19126 DW_AT_loclists_base.
19128 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
19130 * gimple-ssa-store-merging.c: Include gimple-fold.h.
19131 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
19132 (struct merged_store_group): Add bit_insertion field.
19133 (dump_char_array): Use standard hexadecimal format.
19134 (merged_store_group::merged_store_group): Set bit_insertion to false.
19135 (merged_store_group::apply_stores): Use optimal buffer size. Deal
19136 with BIT_INSERT_EXPR stores. Move up code updating the mask and
19137 also print the mask in the dump file.
19138 (pass_store_merging::gate): Minor tweak.
19139 (imm_store_chain_info::coalesce_immediate): Fix wrong association
19140 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
19141 stores with INTEGER_CST stores.
19142 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
19143 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
19144 and use it throughout. Generate bit insertion sequences if need be.
19145 (pass_store_merging::process_store): Remove redundant condition.
19146 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
19148 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
19150 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
19151 the 128-bit floating point types. Fix function comment.
19153 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19155 * config/aarch64/aarch64-simd.md
19156 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
19158 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
19161 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
19163 PR tree-optimization/85989
19164 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
19166 (backprop::intersect_uses): Check it when deciding whether this
19167 is a backedge reference.
19168 (backprop::process_block): Add each phi to m_visited_phis
19169 after visiting it, then clear it at the end.
19171 2018-06-01 Richard Biener <rguenther@suse.de>
19173 * tree-vectorizer.h (vect_dr_stmt): New function.
19174 (vect_get_load_cost): Adjust.
19175 (vect_get_store_cost): Likewise.
19176 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
19177 Use vect_dr_stmt instead of DR_SMTT.
19178 (vect_record_base_alignments): Likewise.
19179 (vect_calculate_target_alignment): Likewise.
19180 (vect_compute_data_ref_alignment): Likewise and make static.
19181 (vect_update_misalignment_for_peel): Likewise.
19182 (vect_verify_datarefs_alignment): Likewise.
19183 (vector_alignment_reachable_p): Likewise.
19184 (vect_get_data_access_cost): Likewise. Pass down
19185 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
19186 (vect_get_peeling_costs_all_drs): Likewise.
19187 (vect_peeling_hash_get_lowest_cost): Likewise.
19188 (vect_enhance_data_refs_alignment): Likewise.
19189 (vect_find_same_alignment_drs): Likewise.
19190 (vect_analyze_data_refs_alignment): Likewise.
19191 (vect_analyze_group_access_1): Likewise.
19192 (vect_analyze_group_access): Likewise.
19193 (vect_analyze_data_ref_access): Likewise.
19194 (vect_analyze_data_ref_accesses): Likewise.
19195 (vect_vfa_segment_size): Likewise.
19196 (vect_small_gap_p): Likewise.
19197 (vectorizable_with_step_bound_p): Likewise.
19198 (vect_prune_runtime_alias_test_list): Likewise.
19199 (vect_analyze_data_refs): Likewise.
19200 (vect_supportable_dr_alignment): Likewise.
19201 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
19202 (vect_gen_prolog_loop_niters): Likewise.
19203 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
19204 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
19206 (vect_recog_mask_conversion_pattern): Likewise.
19207 (vect_try_gather_scatter_pattern): Likewise.
19208 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
19209 to vect_get_store_cost.
19210 (vect_get_store_cost): Get stmt_info instead of DR.
19211 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
19212 (vect_get_load_cost): Get stmt_info instead of DR.
19214 2018-06-01 Richard Biener <rguenther@suse.de>
19216 PR middle-end/86017
19217 * gimple-fold.c (var_decl_component_p): Also allow offsetted
19218 vars wrapped in MEM_REFs.
19220 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
19222 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
19223 Fix subreg tests so that we only return a choice between
19224 GENERAL_REGS and FP_REGS if the original classes included both.
19226 2018-06-01 Richard Biener <rguenther@suse.de>
19229 * tree-ssa-structalias.c (get_function_part_constraint):
19230 Handle NULL fi->decl.
19231 (find_func_aliases_for_call): Properly handle indirect
19232 fi from direct call.
19233 (find_func_clobbers): Likewise.
19234 (ipa_pta_execute): Likewise.
19235 (create_variable_info_for): For functions that are ifunc_resolver
19236 resolve to a varinfo that contains the result of the resolver call.
19237 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
19240 2018-05-31 Michael Collison <michael.collison@arm.com>
19242 * config/aarch64/aarch64.md:
19243 (*fix_to_zero_extenddfdi2): New pattern.
19244 * gcc.target/aarch64/fix_extend1.c: New testcase.
19246 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
19248 PR middle-end/78809
19249 PR middle-end/83026
19250 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
19251 and BUILT_IN_STRNCMP_EQ.
19252 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
19253 BUILT_IN_STRNCMP_EQ.
19254 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
19255 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
19256 (gimple_fold_builtin): Likewise.
19257 * tree-ssa-strlen.c (compute_string_length): New function.
19258 (determine_min_obsize): New function.
19259 (handle_builtin_string_cmp): New function to handle calls to
19260 string compare functions.
19261 (strlen_optimize_stmt): Add handling to builtin string compare
19263 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
19264 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
19265 * tree.c (build_common_builtin_nodes): Add new defines of
19266 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
19268 2018-05-31 Jakub Jelinek <jakub@redhat.com>
19271 * bb-reorder.c (pass_partition_blocks::gate): Return false for
19272 functions with naked attribute.
19274 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
19276 * config/i386/sse.md (avx_vec_concat<mode>):
19277 Substitute concat_tg_mode mode attribute with xtg_mode.
19278 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
19279 (concat_tg_mode): Remove mode attribute.
19281 2018-05-31 Martin Sebor <msebor@redhat.com>
19284 * calls.c (alloc_max_size): Correct a logic error/typo.
19285 Treat excessive arguments as infinite. Warn for invalid arguments.
19286 * doc/invoke.texi (-Walloc-size-larger-than): Update.
19288 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
19291 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
19292 and movx for Haswell.
19294 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
19295 Cesar Philippidis <cesar@codesourcery.com>
19297 PR middle-end/85879
19298 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
19299 when emitting error on private/firstprivate reductions.
19300 * omp-low.c (lower_omp_target): Avoid reference-type processing
19301 on pointers for firstprivate clause.
19303 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
19305 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
19308 * config/aarch64/aarch64-simd.md
19309 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
19310 (aarch64_ld1_x3_<mode>): Likewise
19311 (aarch64_st1x2<VALLDIF:mode>): Likewise
19312 (aarch64_st1_x2_<mode>): Likewise
19313 (aarch64_st1x3<VALLDIF:mode>): Likewise
19314 (aarch64_st1_x3_<mode>): Likewise
19315 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
19316 (vld1_s8_x3): Likewise.
19317 (vld1_u16_x3): Likewise.
19318 (vld1_s16_x3): Likewise.
19319 (vld1_u32_x3): Likewise.
19320 (vld1_s32_x3): Likewise.
19321 (vld1_u64_x3): Likewise.
19322 (vld1_s64_x3): Likewise.
19323 (vld1_f16_x3): Likewise.
19324 (vld1_f32_x3): Likewise.
19325 (vld1_f64_x3): Likewise.
19326 (vld1_p8_x3): Likewise.
19327 (vld1_p16_x3): Likewise.
19328 (vld1_p64_x3): Likewise.
19329 (vld1q_u8_x3): Likewise.
19330 (vld1q_s8_x3): Likewise.
19331 (vld1q_u16_x3): Likewise.
19332 (vld1q_s16_x3): Likewise.
19333 (vld1q_u32_x3): Likewise.
19334 (vld1q_s32_x3): Likewise.
19335 (vld1q_u64_x3): Likewise.
19336 (vld1q_s64_x3): Likewise.
19337 (vld1q_f16_x3): Likewise.
19338 (vld1q_f32_x3): Likewise.
19339 (vld1q_f64_x3): Likewise.
19340 (vld1q_p8_x3): Likewise.
19341 (vld1q_p16_x3): Likewise.
19342 (vld1q_p64_x3): Likewise.
19343 (vst1_s64_x2): Likewise.
19344 (vst1_u64_x2): Likewise.
19345 (vst1_f64_x2): Likewise.
19346 (vst1_s8_x2): Likewise.
19347 (vst1_p8_x2): Likewise.
19348 (vst1_s16_x2): Likewise.
19349 (vst1_p16_x2): Likewise.
19350 (vst1_s32_x2): Likewise.
19351 (vst1_u8_x2): Likewise.
19352 (vst1_u16_x2): Likewise.
19353 (vst1_u32_x2): Likewise.
19354 (vst1_f16_x2): Likewise.
19355 (vst1_f32_x2): Likewise.
19356 (vst1_p64_x2): Likewise.
19357 (vst1q_s8_x2): Likewise.
19358 (vst1q_p8_x2): Likewise.
19359 (vst1q_s16_x2): Likewise.
19360 (vst1q_p16_x2): Likewise.
19361 (vst1q_s32_x2): Likewise.
19362 (vst1q_s64_x2): Likewise.
19363 (vst1q_u8_x2): Likewise.
19364 (vst1q_u16_x2): Likewise.
19365 (vst1q_u32_x2): Likewise.
19366 (vst1q_u64_x2): Likewise.
19367 (vst1q_f16_x2): Likewise.
19368 (vst1q_f32_x2): Likewise.
19369 (vst1q_f64_x2): Likewise.
19370 (vst1q_p64_x2): Likewise.
19371 (vst1_s64_x3): Likewise.
19372 (vst1_u64_x3): Likewise.
19373 (vst1_f64_x3): Likewise.
19374 (vst1_s8_x3): Likewise.
19375 (vst1_p8_x3): Likewise.
19376 (vst1_s16_x3): Likewise.
19377 (vst1_p16_x3): Likewise.
19378 (vst1_s32_x3): Likewise.
19379 (vst1_u8_x3): Likewise.
19380 (vst1_u16_x3): Likewise.
19381 (vst1_u32_x3): Likewise.
19382 (vst1_f16_x3): Likewise.
19383 (vst1_f32_x3): Likewise.
19384 (vst1_p64_x3): Likewise.
19385 (vst1q_s8_x3): Likewise.
19386 (vst1q_p8_x3): Likewise.
19387 (vst1q_s16_x3): Likewise.
19388 (vst1q_p16_x3): Likewise.
19389 (vst1q_s32_x3): Likewise.
19390 (vst1q_s64_x3): Likewise.
19391 (vst1q_u8_x3): Likewise.
19392 (vst1q_u16_x3): Likewise.
19393 (vst1q_u32_x3): Likewise.
19394 (vst1q_u64_x3): Likewise.
19395 (vst1q_f16_x3): Likewise.
19396 (vst1q_f32_x3): Likewise.
19397 (vst1q_f64_x3): Likewise.
19398 (vst1q_p64_x3): Likewise.
19400 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
19402 * config/msp430/msp430.c (msp430_output_labelref): Prepend
19403 user_label_prefix to name.
19405 * tree-core.h: Update comment about the format of NAME string
19406 passed to handler in attribute_spec.
19408 * config/msp430/msp430.md: Remove erroneous subreg expression from
19409 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
19410 zero_extend{q,h}isi2.
19412 2018-05-30 Borislav Petkov <bp@suse.de>
19414 * doc/extend.texi: Document some architecture specific
19415 constraints and sort entries.
19417 2018-05-30 Martin Sebor <msebor@redhat.com>
19419 PR middle-end/85369
19420 * builtins.c (expand_builtin_stpcpy_1): New function.
19421 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
19422 only if the former succeeds.
19424 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
19426 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
19429 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
19431 * doc/invoke.texi (-flinker-output): Document
19433 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
19435 * passes.c (ipa_write_summaries): Only modify statements if body
19437 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
19438 incrementally linking.
19439 (ipa_passes): Likewise.
19440 * lto-cgraph.c (lto_output_node): When incrementally linking do not
19441 pass down resolution info.
19442 * common.opt (flag_incremental_link): Update info.
19443 * gcc.c (plugin specs): Turn flinker-output=* to
19444 -plugin-opt=-linker-output-known
19445 * toplev.c (compile_file): Also cut compilation when doing incremental
19447 * flag-types. (enum lto_partition_model): Add
19448 LTO_LINKER_OUTPUT_NOLTOREL.
19449 (invoke.texi): Add -flinker-output docs.
19450 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
19451 link same way as WPA; do not stream in dead initializers.
19453 * dwarf2out.c (dwarf2out_die_ref_for_decl,
19454 darf2out_register_external_decl): Support incremental link.
19456 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
19458 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
19460 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
19462 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
19463 it down to simple_object_copy_lto_debug_sections.
19464 (run_gcc): Determine incremental LTO link time and configure
19465 lto1 into non-wpa mode, disable renaming of debug sections.
19467 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
19469 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
19470 descriptions of various incorrectly documented functions.
19472 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
19475 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
19476 address check not strict.
19478 2018-05-30 Richard Biener <rguenther@suse.de>
19480 PR tree-optimization/85964
19481 * tracer.c (better_p): Drop initialized count check, we only
19482 call the function with initialized counts now.
19483 (find_best_successor): Do find a best edge if one
19484 has uninitialized count.
19485 (find_best_predecessor): Likewise. Do BB frequency check only
19486 if count is initialized.
19488 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
19490 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
19491 (aarch64_ldrstr_offset_compare): New.
19492 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
19493 load/store orderings.
19494 (aarch64_gen_adjusted_ldpstp): Likewise.
19496 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
19498 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
19499 Check for subset of GENERAL_REGS and FP_REGS.
19500 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
19503 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
19505 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
19506 and wi::to_poly_offset. Add the current offset and then check
19507 whether the sum fits, rather than using an unchecked addition of
19508 a checked term. Check for a shwi rather than a uhwi.
19509 * expr.c (get_bit_range): Use tree_to_poly_uint64.
19510 (store_constructor): Use poly_int_tree_p.
19511 (expand_expr_real_1): Likewise.
19512 * function.c (assign_temp): Likewise.
19513 * fold-const.c (const_binop): Use poly_int_tree_p and
19514 wi::to_poly_offset.
19515 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
19517 * ipa-icf-gimple.c (func_checker::compare_operand): Use
19518 to_poly_offset for MEM offsets.
19519 * ipa-icf.c (sem_variable::equals): Likewise.
19520 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
19521 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
19522 wi::to_poly_offset for BIT_FIELD_REF offsets.
19523 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
19524 wi::to_poly_offset.
19525 * var-tracking.c (emit_note_insn_var_location): Use
19526 tree_to_poly_uint64.
19528 2018-05-29 Jim Wilson <jimw@sifive.com>
19530 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
19532 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
19535 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
19536 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
19538 (sse4_1_round<mode>2): Use nonimmediate_operand
19539 for operand 1 predicate.
19541 2018-05-29 Martin Sebor <msebor@redhat.com>
19542 Richard Biener <rguenther@suse.de>
19545 * calls.c (get_size_range): Call determine_value_range instead
19546 of get_value_range..
19547 * tree-vrp.h (determine_value_range): Declared new function.
19548 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
19550 2018-05-29 Richard Biener <rguenther@suse.de>
19552 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
19553 sure to use non-pattern stmts for get_earlier_stmt arguments.
19554 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
19555 called on pattern stmts.
19556 (get_later_stmt): Likewise.
19558 2018-05-29 Martin Liska <mliska@suse.cz>
19560 PR gcov-profile/85759
19561 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
19564 2018-05-29 Jakub Jelinek <jakub@redhat.com>
19566 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
19568 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
19572 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
19573 VEC_PACK_FLOAT_EXPR): New tree codes.
19574 * tree-pretty-print.c (op_code_prio): Handle
19575 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
19576 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
19577 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
19578 * tree-inline.c (estimate_operator_cost): Likewise.
19579 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
19580 * fold-const.c (const_binop): Likewise.
19581 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
19582 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
19583 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
19584 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
19585 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
19586 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
19587 * expr.c (expand_expr_real_2): Likewise.
19588 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
19589 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
19590 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
19592 * optabs.c (expand_widen_pattern_expr): For
19593 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
19594 sign from result type rather than operand's type.
19595 (expand_binop_directly): For vec_packu_float_optab and
19596 vec_packs_float_optab allow result type to be different from operand's
19598 * optabs-tree.c (optab_for_tree_code): Handle
19599 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
19600 VEC_PACK_FLOAT_EXPR. Formatting fixes.
19601 * tree-vect-generic.c (expand_vector_operations_1): Handle
19602 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
19603 VEC_PACK_FLOAT_EXPR.
19604 * tree-vect-stmts.c (supportable_widening_operation): Handle
19606 (supportable_narrowing_operation): Handle FLOAT_EXPR.
19607 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
19608 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
19609 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
19610 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
19612 (vec_pack<floatprefix>_float_<mode>): New expander.
19613 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
19615 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
19616 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
19617 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
19618 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
19619 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
19621 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
19622 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
19623 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
19624 VEC_PACK_FLOAT_EXPR): Document.
19626 2018-05-29 Richard Biener <rguenther@suse.de>
19628 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
19630 (stmt_vec_info_vec): Make pointer.
19631 (init_stmt_vec_info_vec): Remove.
19632 (free_stmt_vec_info_vec): Likewise.
19633 (set_stmt_vec_info_vec): New function.
19634 (free_stmt_vec_infos): Likewise.
19635 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
19636 (set_vinfo_for_stmt): Likewise.
19637 (get_earlier_stmt): Likewise.
19638 (get_later_stmt): Likewise.
19639 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
19640 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
19641 (vec_info::~vec_info): Free stmt_vec_infos.
19642 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
19643 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
19644 (pass_slp_vectorize::execute): Likewise.
19645 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
19646 (free_stmt_vec_info_vec): Likewise.
19647 (set_stmt_vec_info_vec): New function.
19648 (free_stmt_vec_infos): Likewise.
19649 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
19650 the global stmt_vec_info_vec.
19651 * tree-parloops.c (gather_scalar_reductions): Use
19652 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
19655 2018-05-29 Richard Biener <rguenther@suse.de>
19657 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
19659 2018-05-29 Martin Liska <mliska@suse.cz>
19660 David Malcolm <dmalcolm@redhat.com>
19662 * vec.c (test_reverse): New.
19663 (vec_c_tests): Add new test.
19664 * vec.h (vl_ptr>::reverse): New function.
19666 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
19668 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
19670 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
19673 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
19675 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
19677 2018-05-28 Richard Biener <rguenther@suse.de>
19679 PR tree-optimization/85933
19680 * tree-vect-data-refs.c (vect_record_base_alignments): Only
19681 look at stmts marked as vectorizable.
19683 2018-05-28 Richard Biener <rguenther@suse.de>
19685 PR tree-optimization/85934
19686 * tree-vect-generic.c (expand_vector_operations_1): Hoist
19687 vector boolean check before scalar optimization.
19689 2018-05-28 Jakub Jelinek <jakub@redhat.com>
19691 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
19694 2018-05-28 Mark Wielaard <mark@klomp.org>
19696 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
19697 if it is an expression containing a minus sign.
19699 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
19701 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
19703 2018-05-27 Paul Koning <ni1d@arrl.net>
19705 * config/pdp11/pdp11.md (truncsihi2): Remove.
19707 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
19708 Chung-Ju Wu <jasonwucj@gmail.com>
19710 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
19712 (unaligned_store_dw): Ditto.
19713 * config/nds32/nds32-memory-manipulation.c
19714 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
19715 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
19716 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
19717 (emit_setmem_word_loop): Rename to ...
19718 (emit_setmem_doubleword_loop): ... this.
19719 (nds32_gen_dup_4_byte_to_word_value): New function.
19720 (nds32_gen_dup_8_byte_to_double_word_value): New function.
19721 (nds32_expand_setmem_loop): Refine implementation.
19722 (nds32_expand_setmem_loop_v3m): Ditto.
19723 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
19726 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
19728 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
19730 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
19732 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
19733 (nds32_init_machine_status): Initialize machine->attr_naked_p and
19734 machine->attr_no_prologue_p.
19735 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
19736 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
19737 (nds32_expand_epilogue): Consider attr_naked_p.
19738 (nds32_expand_epilogue_v3pop): Likewise.
19739 (nds32_can_use_return_insn): Likewise.
19740 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
19741 attr_no_prologue_p fields.
19742 * config/nds32/nds32.opt (mret-in-naked-func): New option.
19744 2018-05-27 Jakub Jelinek <jakub@redhat.com>
19747 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
19749 * config/i386/sse.md
19750 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
19752 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
19754 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
19756 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
19758 (*<floatsuffix>floatv2div2sf2): Rename to ...
19759 (*float<floatunssuffix>v2div2sf2): ... this.
19760 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
19761 (float<floatunssuffix>v2div2sf2_mask): ... this.
19762 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
19763 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
19764 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
19766 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
19768 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
19770 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
19772 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
19774 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
19776 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
19777 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
19778 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
19779 gen_ufix_truncv8dfv8si2.
19780 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
19781 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
19782 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
19783 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
19784 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
19785 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
19786 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
19787 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
19789 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
19793 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
19795 2018-05-25 Jim Wilson <jimw@sifive.com>
19797 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
19798 * config/riscv/riscv.c (struct machine_function): Add
19799 interrupt_handler_p and attribute_checked_p fields.
19800 (riscv_attribute_table): Add interrupt.
19801 (riscv_interrupt_type_p): New.
19802 (riscv_save_reg_p): Save extra regs for interrupt handler.
19803 (riscv_use_save_libcall): Return false for interrupt handler.
19804 (riscv_first_stack_step): Add forward declaration.
19805 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
19806 for interrupt handler with large frame. Use it for saved reg list.
19807 (riscv_expand_prologue): Move flag_stack_usage_info support to
19808 eliminate duplication.
19809 (riscv_expand_epilogue): Generate mret for interrupt handler.
19810 (riscv_epilogue_uses): New.
19811 (riscv_can_use_return_insn): Return false for interrupt handler.
19812 (riscv_function_ok_for_sibcall): Likewise.
19813 (riscv_set_current_function): Add interrupt handler support.
19814 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
19815 * config/riscv/riscv.md (UNSPECV_MRET): New.
19817 (riscv_frflags, riscv_fsflags): Use tab after opcode.
19819 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
19821 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
19823 PR tree-optimization/85712
19824 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
19825 this candidate has already been replaced in-situ by a copy.
19827 2018-05-25 Jason Merrill <jason@redhat.com>
19829 PR c++/80485 - inline function non-zero address.
19830 * symtab.c (nonzero_address): Check DECL_COMDAT.
19832 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
19835 * config/alpha/alpha.md (ashlsi3): New insn pattern.
19836 (*ashlsi_se): Rename from *ashldi_se. Define as sign
19837 extension of SImode operation. Use const123_operand predicate.
19838 (*saddsi_1): Remove.
19839 (*saddl_se_1): Ditto.
19840 (*ssubsi_1): Ditto.
19841 (*ssubl_se_1): Ditto.
19842 * config/alpha/predicates.md (const123_operand): New predicate.
19843 * config/alpha/constraints.md (P): Use IN_RANGE.
19845 2018-05-25 Richard Biener <rguenther@suse.de>
19847 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
19849 (ref_maybe_used_by_stmt_p): Likewise.
19850 (stmt_may_clobber_ref_p): Likewise.
19851 (stmt_may_clobber_ref_p_1): Likewise.
19852 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
19854 (ref_maybe_used_by_stmt_p): Likewise.
19855 (stmt_may_clobber_ref_p): Likewise.
19856 (stmt_may_clobber_ref_p_1): Likewise.
19857 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
19858 the alias oracle to disambiguate DRs with stmts DR analysis
19860 (vect_analyze_data_refs): Do not give up on not analyzable
19861 DRs for BB vectorization. Remove code truncating the dataref
19864 2018-05-25 Jakub Jelinek <jakub@redhat.com>
19867 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
19868 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
19869 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
19871 2018-05-25 Richard Biener <rguenther@suse.de>
19873 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
19874 function, combining stmt data ref gathering and fatal analysis
19876 (vect_analyze_data_refs): Remove now redudnant code and simplify.
19877 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
19878 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
19879 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
19880 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
19882 2018-05-25 Bin Cheng <bin.cheng@arm.com>
19884 PR tree-optimization/85720
19885 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
19886 SCC if all partitions are builtins.
19887 (version_loop_by_alias_check): New parameter. Generate cancelable
19888 runtime alias check if all partitions are builtins.
19889 (distribute_loop): Update call to above function.
19891 2018-05-25 Bin Cheng <bin.cheng@arm.com>
19893 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
19894 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
19895 (parm_default_def_partition_arg): Ditto.
19896 (set_parm_default_def_partition): Ditto.
19897 (get_parm_default_def_partitions): Ditto and make it static.
19898 (get_undefined_value_partitions): Ditto and make it static.
19899 (remove_ssa_form): Refactor call to init_var_map here.
19900 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
19901 computation for loop region.
19902 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
19903 (register_default_def): Delete.
19904 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
19905 (parm_default_def_partition_arg): Ditto.
19906 (set_parm_default_def_partition): Ditto.
19907 (get_parm_default_def_partitions): Ditto and make it static.
19908 (get_undefined_value_partitions): Ditto and make it static.
19909 (coalesce_with_default, coalesce_with_default): Update comment.
19910 (create_coalesce_list_for_region): New func factored out from
19911 create_outofssa_var_map.
19912 (populate_coalesce_list_for_outofssa): New func factored out from
19913 create_outofssa_var_map and coalesce_ssa_name.
19914 (create_outofssa_var_map): Delete.
19915 (coalesce_ssa_name): Refactor to support live range computation.
19916 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
19917 (get_parm_default_def_partitions): Delete.
19918 (get_undefined_value_partitions): Ditto.
19919 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
19920 computation for loop region.
19921 (new_tree_live_info, loe_visit_block): Ditto.
19922 (live_worklist, set_var_live_on_entry): Ditto.
19923 (calculate_live_on_exit, verify_live_on_entry): Ditto.
19924 * tree-ssa-live.h (struct _var_map): New fields.
19925 (init_var_map): Change decl.
19926 (region_contains_p): New.
19928 2018-05-25 Bin Cheng <bin.cheng@arm.com>
19930 * tree-ssa-live.h (live_merge_and_clear): Delete.
19932 2018-05-25 Richard Biener <rguenther@suse.de>
19935 * tree-dump.c (dequeue_and_dump): Remove access to removed
19936 operand 2 of a SWITCH_EXPR.
19938 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
19940 * doc/sourcebuild.texi (vect_double_cond_arith): Include
19941 multiplication and division.
19942 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
19943 (cond_udiv@var{m}, cond_umod@var{m}): Document.
19944 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
19945 (cond_udiv_optab, cond_umod_optab): New optabs.
19946 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
19947 (IFN_COND_RDIV): New internal functions.
19948 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
19949 TRUNC_MOD_EXPR and RDIV_EXPR.
19950 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
19951 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
19953 (SVE_INT_BINARY): Include mult.
19954 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
19955 (optab, sve_int_op): Handle mult.
19956 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
19958 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
19959 for SVE_INT_BINARY_SD.
19961 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
19963 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
19964 (optab, sve_int_op): Handle div and udiv.
19965 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
19966 for SVE_INT_BINARY_SD.
19967 (*<optab><mode>3): New insn for the same.
19969 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
19971 * tree-vect-patterns.c: Include predict.h.
19972 (vect_recog_divmod_pattern): Restrict check for division support
19973 to when optimizing for size.
19975 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
19977 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
19978 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
19979 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
19980 (gimple_match_op::set_op): Likewise.
19981 (gimple_resimplify4): Declare.
19982 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
19983 (expr::gen_transform): Likewise.
19984 (decision_tree::gen): Generate a simplification routine for 4 operands.
19985 * gimple-match-head.c (gimple_simplify): Add an overload for
19986 4 operands. In the top-level function, handle up to 4 call
19987 arguments and call gimple_resimplify4.
19988 (gimple_resimplify4): New function.
19989 (build_call_internal): Pass a fourth operand.
19990 (maybe_push_to_seq): Likewise.
19991 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
19992 Fold VEC_COND_EXPRs of an operation and a default value into
19993 an IFN_COND_* function if possible.
19994 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
19996 (SVE_COND_FP_BINARY): Include them.
19997 (optab, sve_fp_op): Handle them.
19998 (SVE_INT_BINARY_REV): New code iterator.
19999 (SVE_COND_FP_BINARY_REV): New int iterator.
20000 (commutative): New int attribute.
20001 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
20003 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
20005 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
20006 (*cond_<optab><mode>): New patterns for reversed operands.
20008 2018-05-25 Richard Biener <rguenther@suse.de>
20010 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
20011 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
20012 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
20013 (STMT_VINFO_GROUPED_ACCESS): Adjust.
20014 * tree-vect-data-refs.c (everywhere): Adjust users.
20015 * tree-vect-loop.c (everywhere): Likewise.
20016 * tree-vect-slp.c (everywhere): Likewise.
20017 * tree-vect-stmts.c (everywhere): Likewise.
20018 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
20020 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20022 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
20024 (gcc_cv_as_section_exclude): ... this.
20025 Try Solaris as #exclude syntax.
20026 * configure: Regenerate.
20027 * config.in: Regenerate.
20028 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
20030 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
20031 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
20033 * varasm.c (default_elf_asm_named_section): Don't check if
20034 HAVE_GAS_SECTION_EXCLUDE is defined.
20036 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
20038 * doc/md.texi: Update the documentation of the cond_* optabs
20039 to mention the new final operand. Fix GET_MODE_NUNITS call.
20040 Describe the scalar case too.
20041 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
20042 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
20044 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
20045 (get_conditional_internal_fn): Update comment.
20046 * tree-vect-loop.c (vectorizable_reduction): Pass the original
20047 accumulator value as a final argument to conditional functions.
20048 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
20049 a define_expand and add an "else" operand. Assert for now that
20050 the else operand is equal to operand 2. Use SVE_INT_BINARY and
20051 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
20052 (*cond_<optab><mode>): New patterns.
20053 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
20054 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
20055 (UNSPEC_COND_EOR): Delete.
20056 (optab): Remove associated mappings.
20057 (SVE_INT_BINARY): New code iterator.
20058 (sve_int_op): Remove int attribute and add "minus" to the code
20060 (SVE_COND_INT_OP): Delete.
20061 (SVE_COND_FP_OP): Rename to...
20062 (SVE_COND_FP_BINARY): ...this.
20064 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
20066 * optabs.c (can_reuse_operands_p): New function.
20067 (maybe_legitimize_operands): Try to reuse the results for
20070 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
20072 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
20073 Add {q} suffix to insn mnemonic.
20075 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
20077 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
20078 (msp430_warn_func_return): New.
20080 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
20082 * fold-const.c (tree_nonzero_bits): New function.
20083 * fold-const.h (tree_nonzero_bits): Likewise.
20084 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
20085 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
20087 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
20091 * varasm.c (assemble_alias): Check ifunc_resolver only on
20094 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
20097 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
20098 when memory input operand is handled.
20100 2018-05-24 Luis Machado <luis.machado@linaro.org>
20102 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
20104 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
20106 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
20108 * match.pd: Delay FMA folds until after vectorization.
20110 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
20113 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
20114 address check not strict.
20116 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
20118 * gimple-match.h (gimple_match_op): New class.
20119 (mprts_hook): Replace parameters with a gimple_match_op *.
20120 (maybe_build_generic_op): Likewise.
20121 (gimple_simplified_result_is_gimple_val): Replace parameters with
20122 a const gimple_match_op *.
20123 (gimple_simplify): Replace code_helper * and tree * parameters with
20124 a gimple_match_op * parameter.
20125 (gimple_resimplify1): Replace code_helper *, tree and tree *
20126 parameters with a gimple_match_op * parameter.
20127 (gimple_resimplify2): Likewise.
20128 (gimple_resimplify3): Likewise.
20129 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
20130 parameters with a gimple_match_op * parameter.
20131 * gimple-match-head.c (gimple_simplify): Change prototypes of
20132 auto-generated functions to take a gimple_match_op * instead of
20133 separate code_helper * and tree * parameters. Make the same
20134 change in the top-level overload and update calls to the
20135 gimple_resimplify routines. Update calls to the auto-generated
20136 functions and to maybe_push_res_to_seq in the publicly-facing
20137 operation-specific gimple_simplify overloads.
20138 (gimple_match_op::MAX_NUM_OPS): Define.
20139 (gimple_resimplify1): Replace rcode and ops with a single res_op
20140 parameter. Update call to gimple_simplify.
20141 (gimple_resimplify2): Likewise.
20142 (gimple_resimplify3): Likewise.
20143 (mprts_hook): Replace parameters with a gimple_match_op *.
20144 (maybe_build_generic_op): Likewise.
20145 (build_call_internal): Replace type, nargs and ops with
20146 a gimple_match_op *.
20147 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
20148 with a single gimple_match_op *. Update calls to mprts_hook,
20149 build_call_internal and gimple_simplified_result_is_gimple_val.
20150 Factor out code that is common to the tree_code and combined_fn cases.
20151 * genmatch.c (expr::gen_transform): Replace tem_code and
20152 tem_ops with a gimple_match_op called tem_op. Update calls
20153 to the gimple_resimplify functions and maybe_push_res_to_seq.
20154 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
20155 res_ops. Update call to the gimple_resimplify functions.
20156 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
20157 (decision_tree::gen): Make the functions take a gimple_match_op *
20158 called res_op instead of separate res_code and res_ops parameters.
20159 Update call accordingly.
20160 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
20161 and ops with a single res_op parameter. Update calls to
20162 maybe_build_generic_op and maybe_push_res_to_seq.
20163 (fold_stmt_1): Update calls to gimple_simplify and
20164 replace_stmt_with_simplification.
20165 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
20166 and gimple_simplified_result_is_gimple_val.
20167 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
20169 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
20170 with a gimple_match_op *.
20171 (vn_nary_build_or_lookup): Likewise. Update call to
20172 vn_nary_build_or_lookup_1.
20173 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
20174 gimple_match_op *. Update calls to the gimple_resimplify routines
20175 and to gimple_simplified_result_is_gimple_val.
20176 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
20177 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
20178 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
20179 (visit_nary_op): Likewise.
20180 (visit_reference_op_load): Likewise.
20182 2018-05-23 Luis Machado <luis.machado@linaro.org>
20184 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
20185 modifier for printing the step amount.
20187 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20190 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
20193 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
20195 * doc/sourcebuild.texi (Endianness): New subsubsection.
20197 2018-05-23 Luis Machado <luis.machado@linaro.org>
20199 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
20200 <prefetch_dynamic_strides>: New const bool field.
20201 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
20202 prefetch_dynamic_strides.
20203 (exynosm1_prefetch_tune): Likewise.
20204 (thunderxt88_prefetch_tune): Likewise.
20205 (thunderx_prefetch_tune): Likewise.
20206 (thunderx2t99_prefetch_tune): Likewise.
20207 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
20209 (aarch64_override_options_internal): Update to set
20210 PARAM_PREFETCH_DYNAMIC_STRIDES.
20211 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
20212 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
20213 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
20214 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
20215 prefetch-dynamic-strides setting.
20217 2018-05-23 Luis Machado <luis.machado@linaro.org>
20219 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
20220 <minimum_stride>: New const int field.
20221 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
20222 minimum_stride field defaulting to -1.
20223 (exynosm1_prefetch_tune): Likewise.
20224 (thunderxt88_prefetch_tune): Likewise.
20225 (thunderx_prefetch_tune): Likewise.
20226 (thunderx2t99_prefetch_tune): Likewise.
20227 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
20228 <default_opt_level>: Set to 3.
20229 (aarch64_override_options_internal): Update to set
20230 PARAM_PREFETCH_MINIMUM_STRIDE.
20231 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
20232 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
20233 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
20234 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
20235 stride is constant and is below the minimum stride threshold.
20237 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20239 * config/arm/arm-cpus.in (mode26): Delete.
20240 (armv4): Delete mode26 reference.
20241 * config/arm/arm.c (arm_configure_build_target): Delete use of
20244 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
20246 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
20248 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
20249 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
20251 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
20252 (floatunsdidf2): Ditto.
20254 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
20256 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
20257 (fixuns_trunc<mode>si2_avx512f): Ditto.
20258 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
20259 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
20260 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
20262 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
20264 PR rtl-optimization/79985
20265 * df-scan.c (df_insn_refs_collect): Remove special case for
20266 global registers and asm statements.
20268 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
20270 * extend.texi (Global Register Variables): Rewrite the bullet list.
20271 Note that the register is available for allocation. Note that access
20272 via inline asm must use constraints. Add note about async-signal
20273 handlers. Remove paragraph about automagic register selection.
20275 2018-05-23 Richard Biener <rguenther@suse.de>
20277 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
20278 of fixed offset from memset VN.
20280 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
20282 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
20283 first_interp field.
20284 (alloc_cand_and_find_basis): Initialize first_interp field.
20285 (slsr_process_mul): Modify first_interp field.
20286 (slsr_process_add): Likewise.
20287 (slsr_process_cast): Modify first_interp field for each new
20289 (slsr_process_copy): Likewise.
20290 (dump_candidate): Dump first_interp field.
20291 (replace_mult_candidate): Process all interpretations, not just
20293 (replace_rhs_if_not_dup): Likewise.
20294 (replace_one_candidate): Likewise.
20296 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
20298 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
20300 (aarch64_needs_frame_chain): New function.
20301 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
20303 2018-05-23 Sudakshina Das <sudi.das@arm.com>
20306 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
20307 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
20308 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
20309 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
20310 as true for strict-align.
20311 (aarch64_can_inline_p): Perform checks even when callee has no
20312 attributes to check for strict alignment.
20313 * doc/extend.texi (AArch64 Function Attributes): Document
20315 * doc/invoke.texi: (AArch64 Options): Likewise.
20317 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
20319 PR tree-optimization/85853
20320 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
20321 the handling of the root of the node to...
20322 (vect_slp_analyze_node_operations_1): ...this new function,
20323 and run the whole thing with the child nodes' def types
20324 set according to their SLP node's def type.
20326 2018-05-23 Richard Biener <rguenther@suse.de>
20328 PR middle-end/85874
20329 * tree-data-ref.c (create_runtime_alias_checks): Defer
20330 and ignore overflow warnings.
20332 2018-05-23 Yury Gribov <tetra2005@gmail.com>
20334 PR tree-optimization/85822
20335 * tree-vrp.c (is_masked_range_test): Fix handling of negative
20338 2018-05-23 Richard Biener <rguenther@suse.de>
20340 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
20341 memset constants via native_interpret_expr.
20343 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
20346 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
20348 (cgraph_node::create_alias): Likewise.
20349 (cgraph_node::get_availability): Check ifunc_resolver instead
20350 of looking up ifunc attribute.
20351 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
20352 * varasm.c (do_assemble_alias): Likewise.
20353 (assemble_alias): Likewise.
20354 (default_binds_local_p_3): Likewise.
20355 * cgraph.h (cgraph_node): Add ifunc_resolver.
20356 (cgraph_node::only_called_directly_or_aliased_p): Return false
20357 for IFUNC resolver.
20358 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
20360 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
20361 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
20362 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
20363 instead of looking up ifunc attribute.
20365 2018-05-22 Luis Machado <luis.machado@linaro.org>
20367 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
20369 2018-05-22 Martin Sebor <msebor@redhat.com>
20371 PR middle-end/85359
20372 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
20373 only when expasion succeeds.
20374 (expand_builtin_strcmp): Same.
20375 (expand_builtin_strncmp): Same.
20377 2018-05-22 Martin Sebor <msebor@redhat.com>
20379 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
20381 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
20382 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20384 * config/aarch64/aarch64-ldpstp.md: Replace uses of
20385 aarch64_mem_pair_operand with memory_operand and delete operand swapping
20387 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
20388 Add check for legitimate_address.
20389 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
20390 (aarch64_swap_ldrstr_operands): New.
20391 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
20394 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
20395 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20397 * config/aarch64/aarch64.md: New patterns to generate stp
20399 (store_pair_sw, store_pair_dw): New patterns to generate stp for
20400 single words and double words.
20401 (load_pair_sw, load_pair_dw): Likewise.
20402 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
20404 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
20406 * config/aarch64/aarch64-ldpstp.md: Modify peephole
20407 for different mode ldpstp and add peephole for merged zero stores.
20408 Likewise for loads.
20409 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
20411 (aarch64_gen_store_pair): Rename calls to match new patterns.
20412 (aarch64_gen_load_pair): Rename calls to match new patterns.
20413 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
20414 (load_pair<DREG:mode><DREG2:mode>): ... This.
20415 (store_pair<mode>): Rename to...
20416 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
20417 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
20418 New mode iterators.
20419 (V_INT_EQUIV): Handle SImode.
20420 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
20423 2018-05-22 Martin Sebor <msebor@redhat.com>
20426 * calls.c (maybe_warn_nonstring_arg): Use string length to set
20427 or ajust the presumed bound on an operation to avoid unnecessary
20430 2018-05-22 Martin Sebor <msebor@redhat.com>
20432 PR tree-optimization/85826
20433 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
20434 assuming that a DECL necesarily has a constant size.
20436 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
20438 PR middle-end/85862
20439 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
20441 2018-05-22 Richard Biener <rguenther@suse.de>
20443 PR tree-optimization/85834
20444 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
20445 non-constant and non-zero memset arguments.
20447 2018-05-22 Martin Liska <mliska@suse.cz>
20450 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
20452 2018-05-22 Richard Biener <rguenther@suse.de>
20454 PR tree-optimization/85863
20455 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
20456 comparisons when vectype is specified.
20457 (vectorizable_condition): Do not specify vectype for
20458 vect_is_simple_cond when SLP vectorizing.
20460 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
20463 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
20464 define __ibm128 as long double.
20465 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
20466 as a distinct type when IEEE 128-bit support is enabled.
20467 (init_float128_ieee): Fix up conversions between IFmode and IEEE
20468 128-bit types to use the correct functions.
20469 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
20470 convert between 128-bit floating point types that have different
20471 modes but the same representation, instead of using gen_lowpart to
20473 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
20475 (IFKF_reg): New attributes to give the register constraints for
20477 (extend<mode>tf2_internal): New insns to mark an explicit
20478 conversion between 128-bit floating point types that have a
20479 different mode but share the same representation.
20481 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
20483 PR tree-optimization/85814
20484 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
20485 a null return from get_strinfo when unsharing the next
20486 strinfo in the chain.
20488 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
20491 * varasm.c (weak_finish): Clean up weak_decls.
20493 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20495 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
20496 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
20497 UNSPEC_UADALP values.
20498 * config/aarch64/iterators.md (ABAL): New int iterator.
20501 (sur): Add mappings for the above.
20502 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
20504 (aarch64_<sur>abal<mode>_4): Likewise.
20505 (aarch64_<sur>adalp<mode>_3): Likewise.
20506 (<sur>sadv16qi): New define_expand.
20508 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
20510 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
20511 (*movdf_internal): Ditto.
20512 (*rcpsf2_sse): Ditto.
20513 (*rsqrtsf2_sse): Ditto.
20514 (*sqrt<mode>2_sse): Ditto.
20516 2018-05-21 Tamar Christina <tamar.christina@arm.com>
20518 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
20520 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
20521 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
20522 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
20523 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
20525 * config/aarch64/arm_neon.h: Likewise.
20526 * config/aarch64/iterators.md (VQ_I): New.
20528 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
20530 * config.gcc: Add arc/t-multilib-linux to tmake_file for
20532 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
20535 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
20537 * config/nds32/constraints.md (S): New constraint.
20538 * config/nds32/nds32.md (call_internal): Use constraint S.
20539 (call_value_internal): Likewise.
20540 (sibcall_internal): Likewise.
20541 (sibcall_value_internal): Likewise.
20543 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
20544 Chung-Ju Wu <jasonwucj@gmail.com>
20546 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
20547 into consideration.
20549 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
20550 Chung-Ju Wu <jasonwucj@gmail.com>
20552 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
20553 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
20554 (nds32_rtx_costs_impl): Simplify.
20555 (nds32_address_cost_impl): Simplify.
20556 (nds32_init_rtx_costs): New function.
20557 (nds32_rtx_costs_speed_prefer): Likewise.
20558 (nds32_rtx_costs_size_prefer): Likewise.
20559 (nds32_address_cost_speed_prefer): Likewise.
20560 (nds32_address_cost_speed_fwprop): Likewise.
20561 (nds32_address_cost_size_prefer): Likewise.
20562 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
20563 * config/nds32/nds32.c (nds32_option_override): Use
20564 nds32_init_rtx_costs function.
20566 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
20568 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
20569 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
20570 (TARGET_PIPELINE_N8): Likewise.
20571 (TARGET_PIPELINE_N10): Likewise.
20572 (TARGET_PIPELINE_N13): Likewise.
20573 (TARGET_PIPELINE_GRAYWOLF): Likewise.
20575 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
20577 * config/nds32/nds32-fpu.md: Update copyright year.
20579 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
20581 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
20583 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
20585 * config/nds32/nds32.c
20586 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
20587 * config/nds32/nds32.opt (minline-asm-r15): New option.
20589 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
20591 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
20593 * config/nds32/nds32.md (abssi2): New pattern.
20595 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
20597 * config/i386/i386.md (rex64namesuffix): New mode attribute.
20598 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
20599 Merge insn pattern from sse_cvtsi2ss<round_name> and
20600 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
20601 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
20602 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
20603 using SWI48 mode iterator.
20604 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
20605 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
20606 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
20607 pattern from sse_cvttss2si<round_saeonly_name>
20608 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
20609 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
20610 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
20611 using SWI48 mode iterator.
20612 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
20613 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
20614 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
20615 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
20616 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
20617 using SWI48 mode iterator.
20618 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
20619 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
20620 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
20621 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
20622 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
20623 SWI48 mode iterator.
20624 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
20625 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
20626 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
20627 pattern from sse_cvttsd2si<round_saeonly_name>
20628 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
20630 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
20632 * config/nds32/nds32-md-auxiliary.c
20633 (nds32_valid_smw_lwm_base_p): Refine.
20634 (nds32_output_smw_single_word): Refine.
20635 (nds32_output_smw_double_word): New.
20636 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
20638 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
20640 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
20641 (nds32_output_stack_pop): Refine.
20642 (nds32_expand_unaligned_load): Refine.
20643 (nds32_expand_unaligned_store): Refine.
20645 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
20646 Chung-Ju Wu <jasonwucj@gmail.com>
20648 * config/nds32/constants.md: Add TP_REGNUM constant.
20649 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
20650 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
20652 * config/nds32/nds32-doubleword.md: Consider flag_pic.
20653 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
20654 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
20655 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
20656 and PIC code generation.
20657 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
20659 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
20661 * config/nds32/nds32.md: Support TLS and PIC.
20662 * config/nds32/nds32.c: Support TLS and PIC.
20663 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
20664 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
20667 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
20669 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
20670 mode with E_ prefix.
20672 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
20673 Chung-Ju Wu <jasonwucj@gmail.com>
20675 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
20676 * config/nds32/nds32-md-auxiliary.c
20677 (symbolic_reference_mentioned_p): New.
20678 (nds32_legitimize_ict_address): New.
20679 (nds32_expand_ict_move): New.
20680 (nds32_indirect_call_referenced_p): New.
20681 (nds32_symbol_binds_local_p): Delete.
20682 (nds32_long_call_p): Modify.
20683 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
20684 * config/nds32/nds32-protos.h
20685 (symbolic_reference_mentioned_p): Declare.
20686 (nds32_legitimize_ict_address): Declare.
20687 (nds32_expand_ict_move): Declare.
20688 (nds32_indirect_call_referenced_p): Declare.
20689 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
20690 (nds32_relax_group): Use nds32_ict_const_p as condition.
20691 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
20692 (nds32_asm_file_start): Output ict_model directive in asm code.
20693 (nds32_legitimate_address_p): Consider indirect call.
20694 (nds32_print_operand): Consider indirect call.
20695 (nds32_print_operand_address): Consider indirect call.
20696 (nds32_insert_attributes): Handle "indirect_call" attribute.
20697 (TARGET_LEGITIMATE_ADDRESS_P): Define.
20698 (TARGET_LEGITIMATE_CONSTANT_P): Define.
20699 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
20700 (TARGET_DELEGITIMIZE_ADDRESS): Define.
20701 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
20702 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
20703 (TARGET_ICT_MODEL_SMALL): Define.
20704 (TARGET_ICT_MODEL_LARGE): Define.
20705 * config/nds32/nds32.md (movsi): Consider ict model.
20706 (call, call_value): Consider ict model.
20707 (sibcall, sibcall_value): Consider ict model.
20708 * config/nds32/nds32.opt (mict-model): New option.
20709 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
20712 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
20713 Monk Chiang <sh.chiang04@gmail.com>
20714 Jim Wilson <jimw@sifive.com>
20716 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
20717 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
20718 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
20719 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
20720 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
20721 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
20722 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
20723 compute save_libcall_adjustment properly.
20724 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
20725 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
20726 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
20727 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
20728 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
20729 (ABI_SPEC): Handle mabi=ilp32e.
20730 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
20731 (RVE): Add RVE mask.
20732 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
20733 <-march>: Add rv32e as an example.
20735 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
20738 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
20739 (intra_create_variable_infos): Handle C++ constructors.
20741 2018-05-18 Martin Liska <mliska@suse.cz>
20743 * passes.def: Remove a redundant pass.
20745 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
20748 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
20750 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20752 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
20754 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
20756 (armv2, armv2a, armv3, armv3m): Delete architectures.
20757 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
20758 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
20759 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
20761 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
20762 (*mulsidi3adddi): Likewise.
20763 (mulsidi3): Likewise.
20764 (*mulsidi3_nov6): Likewise.
20765 (umulsidi3): Likewise.
20766 (umulsidi3_nov6): Likewise.
20767 (umaddsidi4): Likewise.
20768 (*umulsidi3adddi): Likewise.
20769 (smulsi3_highpart): Likewise.
20770 (*smulsi3_highpart_nov6): Likewise.
20771 (umulsi3_highpart): Likewise.
20772 (*umulsi3_highpart_nov6): Likewise.
20773 * config/arm/arm.h (arm_arch3m): Delete.
20774 * config/arm/arm.c (arm_arch3m): Delete.
20775 (arm_option_override_internal): Update armv3-related comment.
20776 (arm_configure_build_target): Delete use of isa_bit_mode32.
20777 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
20778 (arm_rtx_costs_internal): Delete check of arm_arch3m.
20779 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
20780 (mulsa3): Likewise.
20781 (mulusa3): Likewise.
20782 * config/arm/arm-protos.h (arm_arch3m): Delete.
20783 * config/arm/arm-tables.opt: Regenerate.
20784 * config/arm/arm-tune.md: Likewise.
20785 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
20786 deleted architectures.
20788 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20790 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
20791 (armv5t, armv5te): New features.
20792 (ARMv5, ARMv5e): Delete fgroups.
20793 (ARMv5t, ARMv5te): Adjust for above changes.
20794 (ARMv6m): Likewise.
20795 (armv5, armv5e): Delete arches.
20796 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
20798 (*call_reg_arm): Likewise.
20799 (*call_value_reg_armv5): Likewise.
20800 (*call_value_reg_arm): Likewise.
20801 (*call_symbol): Likewise.
20802 (*call_value_symbol): Likewise.
20803 (*sibcall_insn): Likewise.
20804 (*sibcall_value_insn): Likewise.
20805 (clzsi2): Likewise.
20806 (prefetch): Likewise.
20807 (define_split and define_peephole2 dependent on arm_arch5):
20809 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
20811 (TARGET_ARM_QBIT): Likewise.
20812 (TARGET_DSP_MULTIPLY): Likewise.
20813 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
20814 (arm_arch5, arm_arch5e): Delete.
20815 (arm_arch5t, arm_arch5te): Declare.
20816 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
20817 (arm_arch5t): Declare.
20818 (arm_option_reconfigure_globals): Update for the above.
20819 (arm_options_perform_arch_sanity_checks): Update comment, replace
20820 use of arm_arch5 with arm_arch5t.
20821 (use_return_insn): Likewise.
20822 (arm_emit_call_insn): Likewise.
20823 (output_return_instruction): Likewise.
20824 (arm_final_prescan_insn): Likewise.
20825 (arm_coproc_builtin_available): Likewise.
20826 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
20827 arm_arch5e with arm_arch5t and arm_arch5te.
20828 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
20829 (arm_arch5t, arm_arch5te): Declare.
20830 * config/arm/arm-tables.opt: Regenerate.
20831 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
20832 * config/arm/t-multilib: Likewise.
20833 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
20834 instead of arm_arch5.
20835 (*call_reg_thumb1): Likewise.
20836 (*call_value_reg_thumb1_v5): Likewise.
20837 (*call_value_reg_thumb1): Likewise.
20838 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
20840 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
20842 2018-05-18 Martin Liska <mliska@suse.cz>
20844 PR gcov-profile/84846
20845 * doc/gcov.texi: Document -t option of gcov tool.
20847 2018-05-18 Martin Liska <mliska@suse.cz>
20849 PR gcov-profile/84846
20850 * gcov.c (print_usage): Add new -t option.
20851 (process_args): Handle the option.
20852 (generate_results): Use stdout as output when requested by
20855 2018-05-18 Martin Liska <mliska@suse.cz>
20857 PR gcov-profile/84846
20858 * coverage.c (coverage_init): Write PWD to .gcno file.
20859 * doc/gcov.texi: Document how working directory is printed.
20860 * gcov-dump.c (dump_gcov_file): Print PWD.
20861 * gcov.c (output_intermediate_file): Likewise.
20862 (read_graph_file): Read PWD string.
20863 (output_lines): Print PWD.
20865 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20867 PR middle-end/85817
20868 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
20869 for retval and return false if all args to phi are zero.
20871 2018-05-18 Richard Biener <rguenther@suse.de>
20873 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
20875 (evrp_dom_walker::before_dom_children): Call it.
20877 2018-05-18 Richard Biener <rguenther@suse.de>
20879 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
20880 results when processing array refs with variable index.
20882 2018-05-18 Toon Moene <toon@moene.org>
20884 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
20885 directly after that of -floop-interchange. Indicate that both
20886 options are enabled by default when specifying -O3.
20888 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20890 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
20891 iterator. Delete separate integer-mode vec_set<mode> expander.
20892 (aarch64_simd_vec_setv2di): Delete.
20893 (vec_setv2di): Delete.
20894 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
20895 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
20896 the "w, r" alternative.
20898 2018-05-18 Martin Liska <mliska@suse.cz>
20900 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
20901 * tree-pass.h (make_pass_lower_switch_O0): New function.
20902 * tree-switch-conversion.c (node_has_low_bound): Remove.
20903 (node_has_high_bound): Likewise.
20904 (node_is_bounded): Likewise.
20905 (class pass_lower_switch): Make it a template type and create
20907 (pass_lower_switch::execute): Add template argument.
20908 (make_pass_lower_switch): New function.
20909 (make_pass_lower_switch_O0): New function.
20910 (do_jump_if_equal): Remove.
20911 (emit_case_nodes): Simplify to just handle all 3 cases and leave
20912 all the hard work to tree optimization passes.
20914 2018-05-18 Martin Liska <mliska@suse.cz>
20916 * dbgcnt.c (limit_low): Renamed from limit.
20917 (limit_high): New variable.
20918 (dbg_cnt_is_enabled): Check for upper limit.
20919 (dbg_cnt): Adjust dumping.
20920 (dbg_cnt_set_limit_by_index): Add new argument for high
20922 (dbg_cnt_set_limit_by_name): Likewise.
20923 (dbg_cnt_process_single_pair): Parse new format.
20924 (dbg_cnt_process_opt): Use strtok.
20925 (dbg_cnt_list_all_counters): Remove 'value' and add
20927 * doc/invoke.texi: Document changes.
20929 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
20931 * doc/sourcebuild.texi (scalar_all_fma): Document.
20932 * tree.def (FMA_EXPR): Delete.
20933 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
20934 * internal-fn.c (ternary_direct): New macro.
20935 (expand_ternary_optab_fn): Likewise.
20936 (direct_ternary_optab_supported_p): Likewise.
20937 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
20938 * builtins.c (fold_builtin_fma): Delete.
20939 (fold_builtin_3): Don't call it.
20940 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
20941 * expr.c (expand_expr_real_2): Likewise.
20942 * fold-const.c (operand_equal_p): Likewise.
20943 (fold_ternary_loc): Likewise.
20944 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
20945 * gimple.c (DEFTREECODE): Likewise.
20946 * gimplify.c (gimplify_expr): Likewise.
20947 * optabs-tree.c (optab_for_tree_code): Likewise.
20948 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
20949 * tree-eh.c (operation_could_trap_p): Likewise.
20950 (stmt_could_throw_1_p): Likewise.
20951 * tree-inline.c (estimate_operator_cost): Likewise.
20952 * tree-pretty-print.c (dump_generic_node): Likewise.
20953 (op_code_prio): Likewise.
20954 * tree-ssa-loop-im.c (stmt_cost): Likewise.
20955 * tree-ssa-operands.c (get_expr_operands): Likewise.
20956 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
20957 * fold-const-call.h (fold_fma): Delete.
20958 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
20959 CFN_FNMA and CFN_FNMS.
20960 (fold_fma): Delete.
20961 * genmatch.c (combined_fn): New enum.
20962 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
20963 (commutative_op): New function.
20964 (commutate): Use it. Handle more than 2 operands.
20965 (dt_operand::gen_gimple_expr): Use commutative_op.
20966 (parser::parse_expr): Allow :c to be used with non-binary
20967 operators if the commutative operand is known.
20968 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
20969 CFN_FMS, CFN_FNMA and CFN_FNMS.
20970 (backprop::process_assign_use): Remove FMA_EXPR handling.
20971 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
20972 (gen_hsa_fma): New function.
20973 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
20974 IFN_FNMA and IFN_FNMS.
20975 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
20976 * gimple-fold.h (follow_all_ssa_edges): Declare.
20977 * gimple-fold.c (follow_all_ssa_edges): New function.
20978 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
20979 gimple_build interface and use follow_all_ssa_edges to fold the result.
20980 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
20981 instead of checking for optabs directly.
20982 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
20983 rather than FMA_EXPRs.
20984 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
20985 call to IFN_FMA instead of an FMA_EXPR.
20987 2018-05-17 Jim Wilson <jimw@sifive.com>
20989 * expr.c (do_tablejump): When converting index to Pmode, if we have a
20990 sign extended promoted subreg, and the range does not have the sign bit
20991 set, then do a sign extend.
20993 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
20994 test, check for sign extended subreg and/or constant operands, and
20995 do a sign extend in that case.
20997 2018-05-17 Steve Ellcey <sellcey@cavium.com>
20999 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
21000 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
21002 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
21003 Change logics_shift_reg to logics_shift_imm.
21004 (thunderx2t99_fp_loadpair_basic): Delete.
21005 (thunderx2t99_fp_storepair_basic): Delete.
21006 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
21007 (thunderx2t99_asimd_polynomial): Delete.
21008 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
21009 and neon_fp_mul_d_scalar_q.
21010 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
21011 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
21012 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
21013 (thunderx2t99_asimd_lut): Add missing tbl types.
21014 (thunderx2t99_asimd_ext): Delete.
21015 (thunderx2t99_asimd_load1_1_mult): Delete.
21016 (thunderx2t99_asimd_load1_2_mult): Delete.
21017 (thunderx2t99_asimd_load1_ldp): New.
21018 (thunderx2t99_asimd_load1): New.
21019 (thunderx2t99_asimd_load2): Add missing *load2* types.
21020 (thunderx2t99_asimd_load3): New.
21021 (thunderx2t99_asimd_load4): New.
21022 (thunderx2t99_asimd_store1_1_mult): Delete.
21023 (thunderx2t99_asimd_store1_2_mult): Delete.
21024 (thunderx2t99_asimd_store2_mult): Delete.
21025 (thunderx2t99_asimd_store2_onelane): Delete.
21026 (thunderx2t99_asimd_store_stp): New.
21027 (thunderx2t99_asimd_store1): New.
21028 (thunderx2t99_asimd_store2): New.
21029 (thunderx2t99_asimd_store3): New.
21030 (thunderx2t99_asimd_store4): New.
21032 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
21034 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
21035 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
21037 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
21038 Segher Boessenkool <segher@kernel.crashing.org>
21041 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
21044 2018-05-17 Richard Biener <rguenther@suse.de>
21046 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
21047 for pruning loop and prune defs feeding only already visited PHIs.
21049 2018-05-17 Richard Biener <rguenther@suse.de>
21051 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
21053 2018-05-17 Bin Cheng <bin.cheng@arm.com>
21054 Richard Biener <rguenther@suse.de>
21056 PR tree-optimization/85793
21057 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
21058 for VMAT_ELEMENTWISE.
21060 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
21062 * internal-fn.h (lookup_internal_fn): Declare
21063 * internal-fn.c (lookup_internal_fn): New function.
21064 * gimple.c (gimple_build_call_from_tree): Handle calls to
21065 internal functions.
21066 * gimple-pretty-print.c (dump_gimple_call): Print "." before
21067 internal function names.
21068 * tree-pretty-print.c (dump_generic_node): Likewise.
21069 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
21071 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
21073 * gimple-fold.h (gimple_build): Make the function forms take
21074 combined_fn rather than built_in_function.
21075 (gimple_simplify): Likewise.
21076 * gimple-match-head.c (gimple_simplify): Likewise.
21077 * gimple-fold.c (gimple_build): Likewise.
21078 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
21079 rather than gimple_build_call_internal.
21080 (get_initial_defs_for_reduction): Likewise.
21081 (vect_create_epilog_for_reduction): Likewise.
21082 (vectorizable_live_operation): Likewise.
21084 2018-05-17 Martin Liska <mliska@suse.cz>
21086 * gimple-ssa-sprintf.c (format_directive): Do not use
21087 space in between 'G_' and '('.
21089 2018-05-17 Jakub Jelinek <jakub@redhat.com>
21092 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
21093 even if the mask is not all ones.
21096 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
21098 (ix86_gimple_fold_builtin): Likewise.
21101 * config/i386/i386.c: Include tree-vector-builder.h.
21102 (ix86_vector_shift_count): New function.
21103 (ix86_fold_builtin): Fold shift builtins by scalar count.
21104 (ix86_gimple_fold_builtin): Likewise.
21106 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
21107 _mm512_setzero): New intrinsics.
21109 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
21110 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21112 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
21113 code generation for cases where splatting a value is not useful.
21114 * simplify-rtx.c (simplify_ternary_operation): Simplify
21115 vec_merge across a vec_duplicate and a paradoxical subreg forming
21116 a vector mode to a vec_concat.
21118 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
21120 * config.gcc: Support "goldmont-plus".
21121 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
21123 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
21124 PROCESSOR_GOLDMONT_PLUS.
21125 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
21126 (processor_target_table): Add "goldmont-plus".
21127 (PTA_GOLDMONT_PLUS): Define.
21128 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
21129 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
21130 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
21131 (fold_builtin_cpu): Add "goldmont-plus".
21132 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
21133 (ix86_option_override_internal): Add "goldmont-plus".
21134 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
21135 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
21136 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
21137 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
21139 2018-05-17 Richard Biener <rguenther@suse.de>
21141 PR tree-optimization/85757
21142 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
21143 remove defs that only feed that PHI from further processing.
21145 2018-05-16 Jim Wilson <jimw@sifive.com>
21147 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
21149 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
21151 2018-05-16 Mark Wielaard <mark@klomp.org>
21153 * dwarf2out.c (count_index_strings): New function.
21154 (output_indirect_strings): Call count_index_strings and generate
21155 header for dwarf_version >= 5.
21157 2018-05-16 Mark Wielaard <mark@klomp.org>
21159 * dwarf2out.c (dwarf_FORM): New function.
21160 (set_indirect_string): Use dwarf_FORM.
21161 (reset_indirect_string): Likewise.
21162 (size_of_die): Likewise.
21163 (value_format): Likewise.
21164 (output_die): Likewise.
21165 (add_skeleton_AT_string): Likewise.
21166 (output_macinfo_op): Likewise.
21167 (index_string): Likewise.
21168 (output_index_string_offset): Likewise.
21169 (output_index_string): Likewise.
21170 (count_index_strings): Likewise.
21172 2018-05-16 Carl Love <cel@us.ibm.com>
21174 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
21175 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
21177 2018-05-16 Martin Jambor <mjambor@suse.cz>
21179 * ipa-prop.c (ipa_free_all_edge_args): Remove.
21180 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
21182 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
21184 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
21185 (fnma<mode>4): Likewise.
21186 (fms<mode>4): Likewise.
21187 (fnms<mode>4): Likewise.
21188 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
21189 (aarch64_fnma<mode>4): Likewise.
21190 (aarch64_fms<mode>4): Likewise.
21191 (aarch64_fnms<mode>4): Likewise.
21192 (aarch64_fnmadd<mode>4): Likewise.
21194 2018-05-16 Jason Merrill <jason@redhat.com>
21196 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
21198 2018-05-16 Richard Biener <rguenther@suse.de>
21200 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
21201 (dump_stmt_cost): Declare.
21202 (add_stmt_cost): Dump cost we add.
21203 (add_stmt_costs): New function.
21204 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
21205 No longer exported.
21206 (vect_analyze_stmt): Adjust prototype.
21207 (vectorizable_condition): Likewise.
21208 (vectorizable_live_operation): Likewise.
21209 (vectorizable_reduction): Likewise.
21210 (vectorizable_induction): Likewise.
21211 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
21212 cost vector to pass to vectorizable_ and record afterwards.
21213 (vect_model_reduction_cost): Take cost vector argument and adjust.
21214 (vect_model_induction_cost): Likewise.
21215 (vectorizable_reduction): Likewise.
21216 (vectorizable_induction): Likewise.
21217 (vectorizable_live_operation): Likewise.
21218 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
21219 SLP_TREE_NUMBER_OF_VEC_STMTS.
21220 (vect_analyze_slp_cost_1): Remove.
21221 (vect_analyze_slp_cost): Likewise.
21222 (vect_slp_analyze_node_operations): Take visited args and
21223 a target cost vector. Avoid processing already visited stmt sets.
21224 (vect_slp_analyze_operations): Use a local cost vector to gather
21225 costs and register those of non-discarded instances.
21226 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
21227 (vect_schedule_slp_instance): Remove copying of
21228 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
21230 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
21231 adding cost. Record cost entry location.
21232 (vect_prologue_cost_for_slp_op): Function to compute cost of
21233 a constant or invariant generated for SLP vect in the prologue,
21234 split out from vect_analyze_slp_cost_1.
21235 (vect_model_simple_cost): Make static. Adjust for SLP costing.
21236 (vect_model_promotion_demotion_cost): Likewise.
21237 (vect_model_store_cost): Likewise, make static.
21238 (vect_model_load_cost): Likewise.
21239 (vectorizable_bswap): Add cost vector arg and adjust.
21240 (vectorizable_call): Likewise.
21241 (vectorizable_simd_clone_call): Likewise.
21242 (vectorizable_conversion): Likewise.
21243 (vectorizable_assignment): Likewise.
21244 (vectorizable_shift): Likewise.
21245 (vectorizable_operation): Likewise.
21246 (vectorizable_store): Likewise.
21247 (vectorizable_load): Likewise.
21248 (vectorizable_condition): Likewise.
21249 (vectorizable_comparison): Likewise.
21250 (can_vectorize_live_stmts): Likewise.
21251 (vect_analyze_stmt): Likewise.
21252 (vect_transform_stmt): Adjust calls to vectorizable_*.
21253 * tree-vectorizer.c: Include gimple-pretty-print.h.
21254 (dump_stmt_cost): New function.
21256 2018-05-16 Richard Biener <rguenther@suse.de>
21258 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
21259 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
21260 * tree-ssa-dse.c: Include tree-ssa-loop.h.
21261 (check_name): New callback.
21262 (dse_classify_store): Track cycles via a visited bitmap of PHI
21263 defs and simplify handling of in-loop and across loop dead stores
21264 and properly fail for loop-variant refs. Handle byte-tracking with
21265 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
21268 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
21270 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
21271 (vect_get_mask_type_for_stmt): Likewise.
21272 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
21274 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
21275 to determine the statement's vector type and the vector type that
21276 should be used for calculating nunits. Deal with cases in which
21277 the type has to be deferred.
21278 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
21279 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
21280 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
21281 (vect_determine_vf_for_stmt): New functions, split out from...
21282 (vect_determine_vectorization_factor): ...here.
21283 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
21284 (vect_get_mask_type_for_stmt): New functions, split out from
21285 vect_determine_vectorization_factor.
21287 2018-05-16 Richard Biener <rguenther@suse.de>
21289 * tree-cfg.c (verify_gimple_assign_ternary): Properly
21290 verify the [VEC_]COND_EXPR embedded comparison.
21292 2018-05-15 Martin Sebor <msebor@redhat.com>
21294 PR tree-optimization/85753
21295 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
21296 RECORD_TYPE in addition to ARRAY_TYPE.
21298 2018-05-15 Martin Sebor <msebor@redhat.com>
21300 PR middle-end/85643
21301 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
21303 2018-05-15 Richard Biener <rguenther@suse.de>
21305 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
21306 add by_clobber_p one. Change algorithm to collect all defs
21307 representing uses we need to walk and try reducing them to
21308 a single one before failing.
21309 (dse_dom_walker::dse_optimize_stmt): Adjust.
21311 2018-05-13 Mark Wielaard <mark@klomp.org>
21313 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
21314 (size_of_loc_descr): Likewise.
21315 (output_loc_operands): Likewise.
21316 (output_loc_operands_raw): Likewise.
21317 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
21318 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
21319 (hash_loc_operands): Likewise.
21320 (compare_loc_operands): Likewise.
21322 2018-05-14 Mark Wielaard <mark@klomp.org>
21324 * dwarf2out.c (count_index_addrs): New function.
21325 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
21327 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21329 PR tree-optimization/83648
21330 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
21331 return value as malloc candidate.
21333 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21336 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
21337 param as true in call to suggest_attribute.
21339 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
21341 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
21342 -mreadonly-in-sdata.
21344 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21346 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
21348 (aarch64_crypto_aesd_fused): Likewise.
21350 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
21352 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
21353 (movsi_aarch64): Likewise.
21354 (load_pairsi): Likewise.
21355 (load_pairdi): Likewise.
21356 (store_pairsi): Likewise.
21357 (store_pairdi): Likewise.
21358 (load_pairsf): Likewise.
21359 (load_pairdf): Likewise.
21360 (store_pairsf): Likewise.
21361 (store_pairdf): Likewise.
21362 (zero_extend): Likewise.
21363 (trunc): Swap alternatives.
21364 (fcvt_target): Add '?' to prefer w over r.
21366 2018-05-14 Jakub Jelinek <jakub@redhat.com>
21369 * config/i386/i386.md: Disallow non-commutative arithmetics in
21370 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
21371 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
21372 in the peephole2 before it.
21374 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
21376 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
21377 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
21378 (ix86_handle_option): Handle -mcldemote.
21379 * config.gcc: New header.
21380 * config/i386/cldemoteintrin.h: New file.
21381 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
21382 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
21384 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
21385 OPTION_MASK_ISA_CLDEMOTE.
21386 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
21387 (ix86_valid_target_attribute_inner_p): Ditto.
21388 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
21389 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
21390 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
21391 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
21392 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
21394 * config/i386/i386.opt: Add -mcldemote.
21395 * config/i386/x86intrin.h: New header.
21396 * doc/invoke.texi: Add -mcldemote.
21398 2018-05-14 Richard Biener <rguenther@suse.de>
21400 * doc/match-and-simplify.texi: Adjust :s documentation.
21402 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
21404 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
21405 intended memcpy size.
21406 (REORDER_45): Likewise.
21408 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
21410 * sort.cc: New file.
21411 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
21412 * vec.c (qsort_chk): Use gcc_qsort.
21413 * Makefile.in (OBJS-libcommon): Add sort.o.
21414 (build/sort.o): New target. Use it...
21415 (BUILD_RTL): ... here, and...
21416 (build/gencfn-macros): ... here, and...
21417 (build/genmatch): ... here.
21419 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
21420 Chung-Ju Wu <jasonwucj@gmail.com>
21422 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
21423 * config/nds32/nds32-graywolf.md: New file.
21424 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
21425 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
21427 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
21428 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
21429 * config/nds32/nds32.md (pipeline_model): Add graywolf.
21430 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
21431 * config/nds32/pipelines.md: Include n15 settings.
21433 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
21434 Chung-Ju Wu <jasonwucj@gmail.com>
21436 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
21437 * config/nds32/nds32-n13.md: New file.
21438 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
21439 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
21441 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
21442 * config/nds32/nds32.md (pipeline_model): Add n13.
21443 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
21444 * config/nds32/pipelines.md: Include n13 settings.
21446 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
21447 Chung-Ju Wu <jasonwucj@gmail.com>
21449 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
21450 * config/nds32/nds32-n10.md: New file.
21451 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
21452 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
21454 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
21455 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
21456 * config/nds32/nds32.md (pipeline_model): Add n10.
21457 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
21458 * config/nds32/pipelines.md: Include n10 settings.
21460 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
21461 Kito Cheng <kito.cheng@gmail.com>
21462 Chung-Ju Wu <jasonwucj@gmail.com>
21464 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
21465 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
21466 Add enum values for DSP extension instructions.
21467 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
21469 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
21470 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
21471 New code iterators.
21472 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
21473 * config/nds32/nds32-dspext.md: New file for DSP implementation.
21474 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
21475 * config/nds32/nds32-intrinsic.md: Likewise.
21476 * config/nds32/nds32_intrinsic.h: Likewise.
21477 * config/nds32/nds32-md-auxiliary.c: Likewise.
21478 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
21479 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
21480 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
21481 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
21482 * config/nds32/nds32-protos.h: New declarations for DSP extension.
21483 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
21484 TYPE_DMAC in switch statement.
21485 * config/nds32/nds32.c: New checking and implementation for DSP
21486 extension instructions.
21487 * config/nds32/nds32.h: Likewise.
21488 * config/nds32/nds32.md: Likewise.
21489 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
21490 * config/nds32/predicates.md: Implement new predicates for DSP
21493 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
21495 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
21496 Reformat alternatives and attributes so it is easier to identify
21497 which constraints/attributes go with which instruction.
21498 (mov<mode>_hardfloat32, FMOVE64): Likewise.
21499 (mov<mode>_softfloat32, FMOVE64): Likewise.
21500 (mov<mode>_hardfloat64, FMOVE64): Likewise.
21501 (mov<mode>_softfloat64, FMOVE64): Likewise.
21503 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
21505 * doc/extend.texi (PowerPC Built-in Functions): Rename this
21507 (Basic PowerPC Built-in Functions): The new name of the
21508 subsection previously known as "PowerPC Built-in Functions".
21509 (Basic PowerPC Built-in Functions Available on all Configurations):
21511 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
21512 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
21513 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
21514 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
21516 2018-05-11 Martin Jambor <mjambor@suse.cz>
21519 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
21522 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
21525 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
21527 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
21529 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
21530 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
21531 (ix86_handle_option): Handle -mwaitpkg.
21532 * config.gcc: New header.
21533 * config/i386/cpuid.h (bit_WAITPKG): New bit.
21534 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
21535 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
21537 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
21538 OPTION_MASK_ISA_WAITPKG.
21539 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
21540 (ix86_option_override_internal): Add PTA_WAITPKG.
21541 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
21542 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
21543 IX86_BUILTIN_TPAUSE.
21544 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
21545 __builtin_ia32_umwait and __builtin_ia32_tpause.
21546 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
21547 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
21548 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
21549 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
21550 UNSPECV_TPAUSE): New.
21551 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
21552 * config/i386/i386.opt: Add -mwaitpkg.
21553 * config/i386/waitpkgintrin.h: New file.
21554 * config/i386/x86intrin.h: New header.
21555 * doc/invoke.texi: Add -mwaitpkg.
21557 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
21560 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
21562 (cortex-m0): Use armv6s-m isa.
21563 (cortex-m0plus): Likewise.
21564 (cortex-m1): Likewise.
21565 (cortex-m0.small-multiply): Likewise.
21566 (cortex-m0plus.small-multiply): Likewise.
21567 (cortex-m1.small-multiply): Likewise.
21569 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
21570 Jakub Jelinek <jakub@redhat.com>
21572 PR tree-optimization/85692
21573 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
21574 source permute as well.
21576 2018-05-11 Martin Liska <mliska@suse.cz>
21579 * doc/extend.texi: Document LLVM style format for no_sanitize
21582 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
21584 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
21585 mode_supports_vsx_dform_quad to mode_supports_dq_form.
21586 (mode_supports_vsx_dform_quad): Likewise.
21587 (mode_supports_vmx_dform): Move these functions to be next to the
21588 other mode_supports functions.
21589 (mode_supports_dq_form): Likewise.
21590 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
21591 mode_supports_dq_form.
21592 (reg_offset_addressing_ok_p): Likewise.
21593 (offsettable_ok_by_alignment): Likewise.
21594 (rs6000_legitimate_offset_address_p): Likewise.
21595 (legitimate_lo_sum_address_p): Likewise.
21596 (rs6000_legitimize_address): Likewise.
21597 (rs6000_legitimize_reload_address): Likewise.
21598 (rs6000_secondary_reload_inner): Likewise.
21599 (rs6000_preferred_reload_class): Likewise.
21600 (rs6000_output_move_128bit): Likewise.
21602 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
21604 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
21605 Generate SImode target register for null target.
21606 <case IX86_BUILTIN_XGETBV>: Ditto.
21607 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
21608 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
21610 2018-05-10 Carl Love <cel@us.ibm.com>
21612 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
21613 dcbtt and dcbtstt if operands[2] is 0.
21615 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
21618 * config/i386/sse.md (usadv64qi): New expander.
21620 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
21622 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
21623 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
21624 -maltivec=be support.
21625 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
21626 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
21627 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
21628 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
21629 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
21630 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
21631 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
21632 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
21633 altivec_vsumsws): Adjust.
21634 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
21635 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
21636 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
21638 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
21639 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
21640 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
21641 (altivec_lve<VI_char>x): Delete expand.
21642 (*altivec_lve<VI_char>x_internal): Rename to...
21643 (altivec_lve<VI_char>x): ... this.
21644 (altivec_lvxl_<mode>): Delete expand.
21645 (*altivec_lvxl_<mode>_internal): Rename to ...
21646 (altivec_lvxl_<mode>): ... this.
21647 (altivec_stvxl_<mode>): Delete expand.
21648 (*altivec_stvxl_<mode>_internal): Rename to ...
21649 (altivec_stvxl_<mode>): ... this.
21650 (altivec_stve<VI_char>x): Delete expand.
21651 (*altivec_stve<VI_char>x_internal): Rename to ...
21652 (altivec_stve<VI_char>x): ... this.
21653 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
21654 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
21655 reduc_plus_scal_<mode>): Adjust.
21656 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
21658 (rs6000_cpu_cpp_builtins): Adjust.
21659 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
21660 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
21661 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
21662 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
21663 -maltivec=be support.
21664 (rs6000_split_vec_extract_var): Adjust.
21665 (rs6000_split_v4si_init): Adjust.
21666 (swap_selector_for_mode): Delete.
21667 (altivec_expand_lvx_be, altivec_expand_stvx_be,
21668 altivec_expand_stvex_be): Delete.
21669 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
21670 -maltivec=be support.
21671 (rs6000_gimple_fold_builtin): Ditto.
21672 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
21674 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
21675 (TARGET_DIRECT_MOVE_64BIT): Adjust.
21676 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
21677 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
21678 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
21679 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
21680 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
21681 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
21682 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
21683 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
21684 anonymous split): Adjust.
21685 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
21686 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
21688 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
21690 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
21691 when --with-gxx-include-dir is also specified.
21692 * configure: Regenerate.
21694 2018-05-09 Jim Wilson <jimw@sifive.com>
21697 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
21698 * config/riscv/t-withmultilib: New.
21699 * config/riscv/withmultilib.h: New.
21700 * doc/install.texi: Document RISC-V --with-multilib-list support.
21702 2018-05-09 Richard Biener <rguenther@suse.de>
21704 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
21706 (vect_bb_vectorization_profitable_p): Adjust. Compute
21707 actual scalar cost using the cost vector and the add_stmt_cost
21710 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
21712 PR rtl-optimization/85645
21713 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
21714 in the REG_CFA_REGISTER note for LR, don't leave it empty.
21716 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
21718 PR rtl-optimization/85645
21719 * shrink-wrap.c (spread_components): Return a boolean saying if
21720 anything was changed.
21721 (try_shrink_wrapping_separate): Iterate spread_components until
21722 nothing changes anymore.
21724 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
21726 PR rtl-optimization/85645
21727 * regrename.c (build_def_use): Also kill the chains that include the
21728 destination of a REG_CFA_REGISTER note.
21730 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
21732 PR rtl-optimization/85645
21733 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
21734 insn that has a REG_CFA_REGISTER note.
21736 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
21738 * cfgexpand.c (expand_clobber): New function.
21739 (expand_gimple_stmt_1): Use it.
21740 * tree-vect-stmts.c (vect_clobber_variable): New function,
21742 (vectorizable_simd_clone_call): ...here.
21743 (vectorizable_store): Emit a clobber either side of an
21744 IFN_STORE_LANES sequence.
21745 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
21747 2018-05-09 Tom de Vries <tom@codesourcery.com>
21750 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
21751 (define_insn "trap_if_false"): Add exit after trap.
21753 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
21755 PR rtl-optimization/85638
21756 * bb-reorder.c: Include common/common-target.h.
21757 (create_forwarder_block): New function extracted from...
21758 (fix_up_crossing_landing_pad): ...here. Rename into...
21759 (dw2_fix_up_crossing_landing_pad): ...this.
21760 (sjlj_fix_up_crossing_landing_pad): New function.
21761 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
21762 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
21763 from both partitions and exit the loop after one iteration.
21765 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
21768 * doc/extend.texi (PowerPC Built-in Functions): Rename this
21770 (Basic PowerPC Built-in Functions): The new name of the
21771 subsection previously known as "PowerPC Built-in Functions".
21772 (Basic PowerPC Built-in Functions Available on all Configurations):
21774 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
21776 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
21778 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
21780 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
21783 2018-05-08 Jim Wilson <jimw@sifive.com>
21785 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
21786 (LD_EMUL_SUFFIX): New.
21787 (LINK_SPEC): Use it.
21789 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
21791 * doc/extend.texi (PowerPC Built-in Functions): Rename this
21793 (Basic PowerPC Built-in Functions): The new name of the
21794 subsection previously known as "PowerPC Built-in Functions".
21795 (Basic PowerPC Built-in Functions Available on all Configurations):
21797 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
21799 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
21801 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
21803 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
21806 2018-05-08 Jakub Jelinek <jakub@redhat.com>
21809 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
21810 after cmpelim optimization.
21812 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
21814 * config.gcc: Support "goldmont".
21815 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
21816 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
21817 PROCESSOR_GOLDMONT.
21818 * config/i386/i386.c (m_GOLDMONT): Define.
21819 (processor_target_table): Add "goldmont".
21820 (PTA_GOLDMONT): Define.
21821 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
21822 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
21823 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
21824 (fold_builtin_cpu): Add "goldmont".
21825 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
21826 (ix86_option_override_internal): Add "goldmont".
21827 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
21828 (processor_type): Add PROCESSOR_GOLDMONT.
21829 * config/i386/i386.md: Add CPU "glm".
21830 * config/i386/glm.md: New file.
21831 * config/i386/x86-tune.def: Add m_GOLDMONT.
21832 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
21834 2018-05-08 Jakub Jelinek <jakub@redhat.com>
21837 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
21839 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
21840 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
21841 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
21844 * config/i386/i386.c (ix86_fold_builtin): Handle
21845 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
21848 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
21849 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
21851 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
21854 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
21855 (check_arch): Likewise.
21856 (check_fpu): Return the result rather than printing it.
21857 (end arch): Fix operator precedence.
21858 (end cpu): Likewise.
21859 (END): Print the result from check_fpu.
21861 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
21862 Alan Hayward <alan.hayward@arm.com>
21863 David Sherwood <david.sherwood@arm.com>
21865 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
21866 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
21867 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
21868 (*fcmuo<mode>_and): New patterns.
21870 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
21872 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
21873 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
21874 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
21875 (cmp_op, sve_imm_con): New code attributes.
21876 (SVE_COND_INT_CMP, imm_con): Delete.
21877 (cmp_op): Remove above unspecs from int attribute.
21878 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
21880 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
21881 comparison-specific unspecs.
21882 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
21883 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
21884 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
21885 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
21886 (*vec_fcm<cmp_op><mode>): Rename to...
21887 (*fcm<cmp_op><mode>): ...this and adjust likewise.
21888 (*vec_fcmuo<mode>): Rename to...
21889 (*fcmuo<mode>): ...this and adjust likewise.
21890 (*pred_fcm<cmp_op><mode>): New pattern.
21891 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
21892 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
21894 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
21896 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
21897 (aarch64_emit_sve_predicated_cond): New function.
21898 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
21899 (aarch64_emit_unspec_cond_or): Replace with...
21900 (aarch64_emit_sve_or_conds): ...this new function. Use
21901 aarch64_emit_sve_ptrue_op for the individual comparisons and
21902 aarch64_emit_binop to OR them together.
21903 (aarch64_emit_inverted_unspec_cond): Replace with...
21904 (aarch64_emit_sve_inverted_cond): ...this new function. Use
21905 aarch64_emit_sve_ptrue_op for the comparison and
21906 aarch64_emit_unop to invert the result.
21907 (aarch64_expand_sve_vec_cmp_float): Update after the above
21908 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
21910 2018-05-07 Nathan Sidwell <nathan@acm.org>
21912 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
21913 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
21914 (Backwards Compatibility): Likewise.
21916 2018-05-07 Luis Machado <luis.machado@linaro.org>
21920 2018-05-07 Luis Machado <luis.machado@linaro.org>
21922 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
21923 <prefetch_dynamic_strides>: New const bool field.
21924 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
21925 prefetch_dynamic_strides.
21926 (exynosm1_prefetch_tune): Likewise.
21927 (thunderxt88_prefetch_tune): Likewise.
21928 (thunderx_prefetch_tune): Likewise.
21929 (thunderx2t99_prefetch_tune): Likewise.
21930 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
21932 (aarch64_override_options_internal): Update to set
21933 PARAM_PREFETCH_DYNAMIC_STRIDES.
21934 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
21935 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
21936 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
21937 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
21938 prefetch-dynamic-strides setting.
21940 2018-05-07 Luis Machado <luis.machado@linaro.org>
21942 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
21943 <minimum_stride>: New const int field.
21944 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
21945 minimum_stride field.
21946 (exynosm1_prefetch_tune): Likewise.
21947 (thunderxt88_prefetch_tune): Likewise.
21948 (thunderx_prefetch_tune): Likewise.
21949 (thunderx2t99_prefetch_tune): Likewise.
21950 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
21951 (aarch64_override_options_internal): Update to set
21952 PARAM_PREFETCH_MINIMUM_STRIDE.
21953 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
21954 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
21955 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
21956 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
21957 stride is constant and is below the minimum stride threshold.
21959 2018-05-07 Luis Machado <luis.machado@linaro.org>
21961 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
21964 2018-05-07 Luis Machado <luis.machado@linaro.org>
21966 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
21967 <prefetch_dynamic_strides>: New const bool field.
21968 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
21969 prefetch_dynamic_strides.
21970 (exynosm1_prefetch_tune): Likewise.
21971 (thunderxt88_prefetch_tune): Likewise.
21972 (thunderx_prefetch_tune): Likewise.
21973 (thunderx2t99_prefetch_tune): Likewise.
21974 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
21976 (aarch64_override_options_internal): Update to set
21977 PARAM_PREFETCH_DYNAMIC_STRIDES.
21978 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
21979 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
21980 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
21981 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
21982 prefetch-dynamic-strides setting.
21984 2018-05-07 Luis Machado <luis.machado@linaro.org>
21986 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
21987 <minimum_stride>: New const int field.
21988 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
21989 minimum_stride field.
21990 (exynosm1_prefetch_tune): Likewise.
21991 (thunderxt88_prefetch_tune): Likewise.
21992 (thunderx_prefetch_tune): Likewise.
21993 (thunderx2t99_prefetch_tune): Likewise.
21994 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
21995 (aarch64_override_options_internal): Update to set
21996 PARAM_PREFETCH_MINIMUM_STRIDE.
21997 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
21998 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
21999 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
22000 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
22001 stride is constant and is below the minimum stride threshold.
22003 2018-05-06 Jakub Jelinek <jakub@redhat.com>
22006 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
22007 the type is addressable. Don't force op into register if it has
22010 2018-05-05 Roland McGrath <mcgrathr@google.com>
22013 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
22014 any section for which we don't know a specific type it should have,
22015 regardless of name. Previously this was done only for the exact
22016 names ".init_array", ".fini_array", and ".preinit_array".
22017 (default_elf_asm_named_section): Add comment about
22018 relationship with default_section_type_flags and SECTION_NOTYPE.
22019 (get_section): Don't consider it a type conflict if one side has
22020 SECTION_NOTYPE and the other doesn't, as long as neither has the
22021 SECTION_BSS et al used in the default_section_type_flags logic.
22023 2018-05-05 Tom de Vries <tom@codesourcery.com>
22026 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
22027 (workaround_barsyncs): New function.
22028 (nvptx_reorg): Use workaround_barsyncs.
22029 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
22030 (define_expand "nvptx_membar_cta"): New define_expand.
22031 (define_insn "*nvptx_membar_cta"): New insn.
22033 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
22035 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
22036 To improve optimization opportunities.
22037 * builtin-types.def: The new needed builtin types for the above.
22039 2018-05-04 Richard Biener <rguenther@suse.de>
22041 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
22042 * gimple-ssa-store-merging.c
22043 (imm_store_chain_info::output_merged_store): Remove redundant create,
22044 release split_store vector contents on failure.
22045 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
22046 scalar stmt vector on cache hit.
22048 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
22050 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
22052 * config.gcc (powerpc-xilinx-eabi*): Remove.
22053 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
22055 (fusion_addis_mem_combo_load): Ditto.
22056 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
22058 (rs6000_cpu_cpp_builtins): Ditto.
22059 * config/rs6000/rs6000-linux.c
22060 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
22061 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
22062 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
22064 (rs6000_setup_reg_addr_masks): Ditto.
22065 (rs6000_init_hard_regno_mode_ok): Ditto.
22066 (rs6000_option_override_internal): Ditto.
22067 (legitimate_lo_sum_address_p): Ditto.
22068 (rs6000_legitimize_address): Ditto.
22069 (rs6000_legitimize_reload_address): Ditto.
22070 (rs6000_legitimate_address_p): Ditto.
22071 (abi_v4_pass_in_fpr): Ditto.
22072 (setup_incoming_varargs): Ditto.
22073 (rs6000_gimplify_va_arg): Ditto.
22074 (rs6000_split_multireg_move): Ditto.
22075 (rs6000_savres_strategy): Ditto.
22076 (rs6000_emit_prologue_components): Ditto.
22077 (rs6000_emit_epilogue_components): Ditto.
22078 (rs6000_emit_prologue): Ditto.
22079 (rs6000_emit_epilogue): Ditto.
22080 (rs6000_elf_file_end): Ditto.
22081 (rs6000_function_value): Ditto.
22082 (rs6000_libcall_value): Ditto.
22083 * config/rs6000/rs6000.h: Ditto.
22084 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
22085 (TARGET_MINMAX): ... this. New.
22086 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
22087 * config/rs6000/rs6000.md: Remove Xilinx FP support.
22088 (*movsi_internal1_single): Delete.
22089 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
22090 mfpu=, mxilinx-fpu): Delete.
22091 * config/rs6000/singlefp.h: Delete.
22092 * config/rs6000/sysv4.h: Remove Xilinx FP support.
22093 * config/rs6000/t-rs6000: Ditto.
22094 * config/rs6000/t-xilinx: Delete.
22095 * config/rs6000/titan.md: Adjust for fp_type removal.
22096 * config/rs6000/vsx.md: Remove Xilinx FP support.
22097 (VStype_simple): Delete.
22098 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
22099 * config/rs6000/xfpu.h: Delete.
22100 * config/rs6000/xfpu.md: Delete.
22101 * config/rs6000/xilinx.h: Delete.
22102 * config/rs6000/xilinx.opt: Delete.
22103 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
22104 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
22106 2018-05-04 Tom de Vries <tom@codesourcery.com>
22109 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
22112 2018-05-04 Richard Biener <rguenther@suse.de>
22114 PR middle-end/85627
22115 * tree-complex.c (update_complex_assignment): We are always in SSA form.
22116 (expand_complex_div_wide): Likewise.
22117 (expand_complex_operations_1): Likewise.
22118 (expand_complex_libcall): Preserve EH info of the original stmt.
22119 (tree_lower_complex): Handle removed blocks.
22120 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
22121 on complex multiplication and division libcall builtins.
22123 2018-05-04 Richard Biener <rguenther@suse.de>
22125 PR middle-end/85574
22126 * fold-const.c (negate_expr_p): Restrict negation of operand
22127 zero of a division to when we know that can happen without
22129 (fold_negate_expr_1): Likewise.
22131 2018-05-04 Jakub Jelinek <jakub@redhat.com>
22134 * real.h (real_nextafter): Declare.
22135 * real.c (real_nextafter): New function.
22136 * fold-const-call.c (fold_const_nextafter): New function.
22137 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
22138 CASE_CFN_NEXTTOWARD.
22139 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
22140 even when arg1_mode is different from arg0_mode.
22142 2018-05-03 Nathan Sidwell <nathan@acm.org>
22144 * doc/extend.texi (Deprecated Features): Remove
22145 -ffriend-injection.
22146 (Backwards Compatibility): Likewise.
22147 * doc/invoke.texi (C++ Language Options): Likewise.
22148 (C++ Dialect Options): Likewise.
22150 2018-05-03 Jakub Jelinek <jakub@redhat.com>
22153 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
22154 _mm512_mask_mullox_epi64): New intrinsics.
22156 2018-05-03 Tom de Vries <tom@codesourcery.com>
22159 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
22160 dump files): Add offload-tree.
22162 2018-05-03 Richard Biener <rguenther@suse.de>
22164 PR tree-optimization/85615
22165 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
22166 to loops not nested in BBs loop father to avoid creating multi-entry
22169 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22171 PR tree-optimization/70291
22172 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
22173 arguments. Change return type to tree. Emit libcall as a new
22174 statement rather than replacing existing one when inplace_p is true.
22175 (expand_complex_multiplication_components): New function.
22176 (expand_complex_multiplication): Expand floating-point complex
22177 multiplication using the above.
22178 (expand_complex_division): Rename inner_type parameter to type.
22179 Update expand_complex_libcall call-site.
22180 (expand_complex_operations_1): Update expand_complex_multiplication
22181 and expand_complex_division call-sites.
22183 2018-05-02 Jakub Jelinek <jakub@redhat.com>
22186 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
22187 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
22188 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
22189 the highest significant bit of the shift count mask is clear. In
22190 check whether and[sq]i3 is needed verify that all significant bits
22191 of the shift count other than the highest are set.
22193 2018-05-02 Tom de Vries <tom@codesourcery.com>
22196 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
22197 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
22198 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
22199 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
22200 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
22201 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
22202 __builtin_goacc_parlevel_size.
22204 2018-05-02 Richard Biener <rguenther@suse.de>
22206 PR tree-optimization/85597
22207 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
22208 do not use split vect_get_vec_defs call but call vect_get_slp_defs
22211 2018-05-02 Tom de Vries <tom@codesourcery.com>
22214 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
22215 dump files): Add ltrans-tree.
22217 2018-05-02 Tom de Vries <tom@codesourcery.com>
22220 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
22221 dump files): Add wpa-ipa.
22223 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
22225 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
22226 powerpc*-*-linux*paired* target.
22227 * config/rs6000/750cl.h: Delete.
22228 * config/rs6000/paired.h: Delete.
22229 * config/rs6000/paired.md: Delete.
22230 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
22232 * config/rs6000/rs6000-builtin.def: Remove paired float support.
22233 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
22234 comment. Remove paired float support.
22235 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
22236 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
22238 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
22239 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
22241 * config/rs6000/rs6000.c: Remove paired float support.
22242 (paired_expand_vector_init, paired_expand_vector_move,
22243 paired_emit_vector_compare, paired_emit_vector_cond_expr,
22244 (paired_expand_lv_builtin, paired_expand_stv_builtin,
22245 paired_expand_builtin, paired_expand_predicate_builtin,
22246 paired_init_builtins): Delete.
22247 * config/rs6000/rs6000.h: Remove paired float support.
22248 * config/rs6000/rs6000.md: Remove paired float support.
22249 (move_from_CR_ov_bit): Delete.
22250 * config/rs6000/rs6000.opt (mpaired): Delete.
22251 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
22252 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
22254 2018-05-02 Richard Biener <rguenther@suse.de>
22256 PR middle-end/85567
22257 * gimplify.c (gimplify_save_expr): When in SSA form allow
22258 SAVE_EXPRs to compute to SSA vars.
22260 2018-05-02 Jakub Jelinek <jakub@redhat.com>
22263 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
22264 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
22265 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
22266 clobber operands[2], instead use a new pseudo. Formatting fixes.
22268 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
22270 PR tree-optimization/85586
22271 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
22272 exit early for statements in the same group if the accesses are
22275 2018-05-02 Tom de Vries <tom@codesourcery.com>
22278 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
22281 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
22283 PR tree-optimization/85143
22284 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
22286 2018-05-01 Tom de Vries <tom@codesourcery.com>
22289 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
22290 not found" error message.
22292 2018-05-01 Tom de Vries <tom@codesourcery.com>
22295 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
22296 * vec.c (test_ordered_remove_if): New function.
22297 (vec_c_tests): Call test_ordered_remove_if.
22298 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
22299 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
22300 * tree-vect-patterns.c (vect_pattern_recog_1): Use
22301 VEC_ORDERED_REMOVE_IF.
22303 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22305 PR tree-optimization/82665
22306 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
22307 pointer subtraction where arguments come from a memchr call.
22309 2018-05-01 Jakub Jelinek <jakub@redhat.com>
22311 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
22312 --push-state --as-needed and --pop-state instead of --as-needed and
22313 --no-as-needed if ld supports it.
22314 * configure: Regenerated.
22317 * doc/install.texi2html: Replace _002d with - and _002a with * in
22318 generated html files using sed.
22320 2018-04-30 David Malcolm <dmalcolm@redhat.com>
22323 * gcc-rich-location.c (blank_line_before_p): New function.
22324 (use_new_line): New function.
22325 (gcc_rich_location::add_fixit_insert_formatted): New function.
22326 * gcc-rich-location.h
22327 (gcc_rich_location::add_fixit_insert_formatted): New function.
22329 2018-04-30 David Malcolm <dmalcolm@redhat.com>
22331 * selftest.c (assert_streq): Rename "expected" and "actual" to
22332 "val1" and "val2". Extend NULL-handling to cover both inputs
22333 symmetrically, while still requiring both to be non-NULL for a pass.
22334 * selftest.h (assert_streq): Rename "expected" and "actual" to
22336 (ASSERT_EQ): Likewise.
22337 (ASSERT_EQ_AT): Likewise.
22338 (ASSERT_KNOWN_EQ): Likewise.
22339 (ASSERT_KNOWN_EQ_AT): Likewise.
22340 (ASSERT_NE): Likewise.
22341 (ASSERT_MAYBE_NE): Likewise.
22342 (ASSERT_MAYBE_NE_AT): Likewise.
22343 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
22344 the assertion to pass.
22345 (ASSERT_STREQ_AT): Likewise.
22347 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
22349 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
22350 interaction with -pie.
22352 2018-04-30 David Malcolm <dmalcolm@redhat.com>
22354 * selftest.h: Fix alphabetization of per-source-file selftest
22357 2018-04-30 Jason Merrill <jason@redhat.com>
22359 PR c++/61982 - dead stores to destroyed objects.
22360 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
22363 2018-04-30 Jason Merrill <jason@redhat.com>
22365 * tree.c (build_clobber): New.
22366 * tree.h: Declare it.
22367 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
22369 2018-04-30 David Malcolm <dmalcolm@redhat.com>
22371 * diagnostic-show-locus.c (layout::layout): Update for
22372 location_get_source_line returning a char_span.
22373 (struct char_span): Move to input.h.
22374 (struct correction): Update for fields in char_span becoming
22376 (struct source_line): Update for location_get_source_line
22377 returning a char_span.
22378 (layout::print_line): Likewise.
22379 * edit-context.c (edited_file::print_content): Likewise.
22380 (edited_file::print_diff_hunk): Likewise.
22381 (edited_file::print_run_of_changed_lines): Likewise.
22382 (edited_file::get_num_lines): Likewise.
22383 (edited_line::edited_line): Likewise.
22384 * final.c (asm_show_source): Likewise.
22385 * input.c (location_get_source_line): Convert return type
22386 from const char * to char_span, losing the final "line_len"
22388 (dump_location_info): Update for the above.
22389 (get_substring_ranges_for_loc): Likewise. Use a char_span
22390 when handling the literal within the line.
22391 (test_reading_source_line): Update for location_get_source_line
22392 returning a char_span.
22393 * input.h (class char_span): Move here from
22394 diagnostic-show-locus.c, converting from a struct to a class.
22395 Make data members private.
22396 (char_span::operator bool): New.
22397 (char_span::length): New.
22398 (char_span::get_buffer): New.
22399 (char_span::operator[]): New.
22400 (char_span::subspan): Make const.
22401 (char_span::xstrdup): New.
22402 (location_get_source_line): Convert return type from const char *
22403 to char_span, losing the final "line_size" param.
22405 2018-04-30 Jan Hubicka <jh@suse.cz>
22407 * lto-wrapper.c (ltrans_priorities): New static var.
22408 (cmp_priority): New.
22409 (run_gcc): Read priorities and if doing parallel build order
22410 the Makefile by them.
22412 2018-04-30 David Malcolm <dmalcolm@redhat.com>
22414 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
22416 2018-04-30 Richard Biener <rguenther@suse.de>
22418 * tree-cfg.c (verify_address): Remove base argument, add
22419 flag whether to check TREE_ADDRESSABLE and do that.
22420 (verify_expr): Remove.
22421 (verify_types_in_gimple_reference): Add pieces from verify_expr.
22422 (verify_gimple_assign_single): Likewise.
22423 (verify_gimple_switch): Likewise.
22424 (verify_expr_location_1): Dereference tp once. Add (disabled)
22425 piece from verify_expr.
22426 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
22428 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
22430 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
22432 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
22434 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
22435 (small_data_pattern): Likewise.
22436 (arc_rewrite_small_data): Likewise.
22437 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
22438 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
22439 (get_symbol_alignment): New function.
22440 (legitimate_small_data_address_p): Likewise.
22441 (legitimate_scaled_address): Update, call
22442 legitimate_small_data_address_p.
22443 (output_sdata): New static variable.
22444 (arc_print_operand): Update how we handle small data operands.
22445 (arc_print_operand_address): Likewise.
22446 (arc_legitimate_address_p): Update, use
22447 legitimate_small_data_address_p.
22448 (arc_rewrite_small_data_p): Remove.
22449 (arc_rewrite_small_data_1): Likewise.
22450 (arc_rewrite_small_data): Likewise.
22451 (small_data_pattern): Likewise.
22452 (compact_sda_memory_operand): Update to use
22453 legitimate_small_data_address_p and get_symbol_alignment.
22454 (prepare_move_operands): Don't rewite sdata pattern.
22455 (prepare_extend_operands): Remove.
22456 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
22458 (zero_extendqisi2): Likewise.
22459 (zero_extendhisi2): Likewise.
22460 (extendqihi2): Likewise.
22461 (extendqisi2): Likewise.
22462 (extendhisi2): Likewise.
22463 (addsi3): Likewise.
22464 (subsi3): Likewise.
22465 (andsi3): Likewise.
22466 * config/arc/constraints.md (Usd): Change it to memory constraint.
22468 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
22470 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
22471 as source of std instructions.
22472 * config/arc/arc.md (movsi_insn): Update pattern predicate to
22473 allow 6-bit constants as source for store instructions.
22474 (movdi_insn): Update instruction pattern to allow 6-bit constants
22475 as source for store instructions.
22477 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
22479 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
22481 2018-04-30 Nathan Sidwell <nathan@acm.org>
22482 Sandra Loosemore <sandra@codesourcery.com>
22484 * dumpfile.c (dump_open): Allow '-' for stdout.
22485 * doc/invoke.texi (Developer Options): Document dump filename
22486 determination early. Document stdin/stdout selection.
22488 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
22490 Microblaze Target: PIC data text relative
22492 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
22493 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
22495 * config/microblaze/microblaze.h (microblaze_constant_address_p):
22496 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
22497 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
22498 New addressing mode for data-text relative position indepenedent code.
22499 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
22500 'ADDRESS_SYMBOLIC_TXT_REL'.
22501 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
22502 (microblaze_legitimate_pic_operand): Exclude function calls from
22503 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
22504 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
22506 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
22507 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
22508 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
22509 for 'address + offset'.
22510 (microblaze_expand_prologue): Add new function prologue call for
22512 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
22513 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
22514 table in case of TARGET_PIC_DATA_TEXT_REL.
22515 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
22516 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
22517 Add new macros 'UNSPEC_TEXT',
22518 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
22519 + exclude function calls from 'UNSPEC_PLT' in case of data text
22521 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
22522 new target hook for generating address diff vector tables in case of
22524 * doc/tm.texi : Regenerate.
22525 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
22526 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
22527 of addr diff vector generation.
22528 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
22529 target hook definition.
22530 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
22531 Add default function for generate_pic_addr_diff_vec -> flag_pic.
22532 * doc/invoke.texi (Add new pic option): Add new microblaze pic
22533 option for data text relative.
22535 2018-04-30 Richard Biener <rguenther@suse.de>
22537 * tree-chrec.h (evolution_function_is_constant_p): Remove
22539 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
22541 2018-04-30 Richard Biener <rguenther@suse.de>
22544 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
22546 2018-04-30 Richard Biener <rguenther@suse.de>
22548 PR tree-optimization/28364
22549 PR tree-optimization/85275
22550 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
22551 copying first exit test.
22553 2018-04-28 Mark Wielaard <mark@klomp.org>
22555 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
22556 dwarf_version >= 5.
22557 (dwarf_AT): Handle DW_AT_addr_base.
22558 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
22560 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
22563 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
22564 (*ashl<dwi>3_doubleword_mask_1): Ditto.
22565 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
22566 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
22568 2018-04-28 Richard Biener <rguenther@suse.de>
22570 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
22571 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
22572 to reflect use. Only add interesting stmts.
22574 2018-04-27 Martin Jambor <mjambor@suse.cz>
22577 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
22578 the jump function allows for passing through aggregate values.
22580 2018-04-27 David Malcolm <dmalcolm@redhat.com>
22582 * input.h (in_system_header_at): Convert from macro to inline
22584 (from_macro_expansion_at): Likewise.
22585 (from_macro_definition_at): Likewise.
22587 2018-04-27 Jeff Law <law@redhat.com>
22589 * config.gcc: Mark tile* targets as deprecated/obsolete.
22591 2018-04-27 Richard Biener <rguenther@suse.de>
22593 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
22596 2018-04-27 Richard Biener <rguenther@suse.de>
22598 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
22600 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
22602 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
22603 with Yd constraint. Set "preferred_for_speed" attribute from
22604 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
22605 with Yd constraint.
22606 (*movdi_internal): Ditto.
22607 (movti_interunit splitters): Remove
22608 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
22609 (movdi_interunit splitters): Ditto.
22610 * config/i386/constraints.md (Ye): Remove.
22611 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
22613 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22616 * config/aarch64/constraints.md (Usg): Limit to 31.
22617 (Usj): Limit to 63.
22619 2018-04-27 Jakub Jelinek <jakub@redhat.com>
22621 PR tree-optimization/85529
22622 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
22623 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
22624 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
22625 zero extension or masking of the MSB bit.
22626 (optimize_range_tests): Add FIRST_BB argument, pass it through
22627 to optimize_range_tests_var_bound.
22628 (maybe_optimize_range_tests, reassociate_bb): Adjust
22629 optimize_range_tests callers.
22631 2018-04-26 Richard Biener <rguenther@suse.de>
22632 Jakub Jelinek <jakub@redhat.com>
22634 * cgraph.h (symbol_table): Just declare debug method here.
22635 * symtab.c (symbol_table::debug): Define.
22637 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
22639 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
22641 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
22643 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
22644 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
22645 (*movdi_internal): Substitute Yi and Yj constraint with x
22646 and Ym and Yn constraint with y constraint. Update "isa"
22647 attribute and set "preferred_for_speed" attribute from
22648 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
22649 (*movsi_internal): Ditto.
22650 (*movdf_internal): Ditto.
22651 (*movsf_internal): Ditto.
22652 (*zero_extendsidi2): Ditto.
22653 * config/i386/sse.md (vec_set<mode>_0): Ditto.
22654 (sse2_loadld): Ditto.
22655 (*vec_extract<ssevecmodelower>_0): Ditto.
22656 (*vec_extractv4si_0_zext_sse4): Ditto.
22657 (vec_concatv2di): Ditto.
22658 (*vec_dup<mode>): Ditto.
22659 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
22660 * config/i386/constraints.md (Yi): Remove.
22665 2018-04-26 Nathan Sidwell <nathan@acm.org>
22667 * dumpfile.c (dump_open): New.
22668 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
22669 (dump_finish): Detect stdio/stderr by value not name.
22671 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
22673 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
22675 2018-04-26 Tom de Vries <tom@codesourcery.com>
22678 * config/nvptx/nvptx.c (verify_neutering_jumps)
22679 (verify_neutering_labels): New function
22680 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
22682 2018-04-26 Tom de Vries <tom@codesourcery.com>
22685 * config/nvptx/nvptx.c (needs_neutering_p): New function.
22686 (nvptx_single): Use needs_neutering_p to skip over insns that do not
22689 2018-04-26 Richard Biener <rguenther@suse.de>
22690 Tom de Vries <tom@codesourcery.com>
22693 * lto-streamer-out.c (output_function): Fixup loops if required to match
22694 discovery done in the reader.
22696 2018-04-26 Richard Biener <rguenther@suse.de>
22698 PR tree-optimization/85116
22699 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
22700 have a loop exit from the single latch predecessor. Remove
22701 case of header with just condition.
22702 (ch_base::copy_headers): Exclude infinite loops from any
22704 (pass_ch::execute): Record exits.
22706 2018-04-26 Richard Biener <rguenther@suse.de>
22708 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
22709 prologue cost vector and pass it to vect_get_load_cost.
22710 (vect_get_peeling_costs_all_drs): Likewise.
22711 (vect_peeling_hash_get_lowest_cost): Likewise.
22712 (vect_enhance_data_refs_alignment): Likewise.
22714 2018-04-26 Richard Biener <rguenther@suse.de>
22716 PR middle-end/85450
22717 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
22718 checking of integer<->pointer conversions.
22719 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
22720 sign-/zero-extending pointer types.
22721 (expand_omp_for_static_chunk): Likewise.
22723 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
22724 Jean Lee <xiaoyur347@gmail.com>
22726 * config/mips/mips.c (mips_asan_shadow_offset): New function.
22727 (TARGET_ASAN_SHADOW_OFFSET): Define.
22728 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
22729 true for -fsanitize=address.
22731 2018-04-25 Mark Wielaard <mark@klomp.org>
22733 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
22736 2018-04-25 Jakub Jelinek <jakub@redhat.com>
22738 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
22739 than "alu", remove explicit "memory" and "imm_disp" attributes.
22740 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
22742 PR middle-end/85414
22743 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
22744 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
22745 gen_lowpart_no_emit.
22747 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
22750 * config/i386/i386.c (ix86_expand_builtin): Change memory
22751 operand to XI, extend p0 to Pmode.
22752 * config/i386/i386.md: Change unspec volatile and operand
22753 1 mode to XI, change operand 0 mode to P.
22755 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
22757 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
22758 GET_MODE_MASK before any checking.
22759 (nds32_can_use_bset_p): Likewise.
22760 (nds32_can_use_btgl_p): Likewise.
22762 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
22764 * config/nds32/nds32-doubleword.md: New define_split pattern for
22765 illegal register number.
22767 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
22769 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
22771 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
22773 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
22775 2018-04-25 Richard Biener <rguenther@suse.de>
22777 * lto-streamer.h (LTO_major_version): Bump to 8.
22779 2018-04-25 Jakub Jelinek <jakub@redhat.com>
22781 * BASE-VER: Set to 9.0.0.
22783 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
22785 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
22786 in __abskf2 and __powikf2.
22788 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22791 * config/aarch64/constraints.md (Usg, Usj): New constraints.
22792 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
22793 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
22794 Use the above on operand 2. Reindent.
22795 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
22797 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
22800 * common/config/i386/i386-common.c (ix86_handle_option): Don't
22802 * config/i386/i386.opt (mcet): Removed.
22803 * doc/install.texi: Remove -mcet documentation.
22804 * doc/invoke.texi: Likewise.
22806 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
22809 * doc/install.texi: Remove -mcet from bootstrap-cet.
22811 2018-04-24 Jakub Jelinek <jakub@redhat.com>
22814 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
22815 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
22819 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
22820 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
22821 containing a CONST_VECTOR.
22823 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
22825 * doc/install.texi: Update newlib dependency for nvptx.
22827 2018-04-24 Jakub Jelinek <jakub@redhat.com>
22830 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
22831 instead of INTVAL when shifting x left.
22833 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
22835 PR tree-optimization/85478
22836 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
22837 vect_grouped_store_supported for single element vectors.
22839 2018-04-24 Richard Biener <rguenther@suse.de>
22842 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
22843 load cost increase to the case of non-constant step.
22845 2018-04-24 Jakub Jelinek <jakub@redhat.com>
22848 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
22849 destination if any_malformed_asm.
22851 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
22853 PR middle-end/85496
22854 * expr.c (store_field): In the bitfield case, if the value comes from
22855 a function call and is returned in registers by means of a PARALLEL,
22856 do not change the mode of the temporary unless BLKmode and VOIDmode.
22858 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
22860 PR rtl-optimization/85423
22861 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
22862 dependencies to debug insns when the previous insn is non-debug.
22864 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
22866 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
22867 enums into a single definition.
22868 (fls): Fix predicates and printing.
22871 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
22873 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
22874 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
22875 and short u6 immediate.
22876 (check_if_valid_sleep_operand): Remove.
22877 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
22879 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
22881 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
22882 flag_always_save_lp condition.
22883 * config/nds32/nds32.opt (malways-save-lp): New option.
22885 2018-04-22 Shiva Chen <shiva0217@gmail.com>
22887 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
22888 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
22889 * config/nds32/nds32.h
22890 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
22891 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
22893 2018-04-22 Shiva Chen <shiva0217@gmail.com>
22895 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
22896 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
22898 2018-04-22 Shiva Chen <shiva0217@gmail.com>
22899 Chung-Ju Wu <jasonwucj@gmail.com>
22901 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
22903 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
22904 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
22906 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
22908 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
22910 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
22912 * config/nds32/nds32-protos.h (nds32_data_alignment,
22913 nds32_local_alignment): Declare.
22914 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
22915 nds32_local_alignment): New functions.
22916 (TARGET_CONSTANT_ALIGNMENT): Define.
22917 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
22919 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
22921 * config/nds32/nds32.c
22922 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
22923 (TARGET_MODES_TIEABLE_P): Likewise.
22925 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
22927 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
22928 level Ofast and Og.
22930 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
22931 Chung-Ju Wu <jasonwucj@gmail.com>
22933 * config/nds32/constants.md (unspec_volatile_element): Add enum values
22934 for unaligned access.
22935 * config/nds32/nds32-intrinsic.c: Implementation of expanding
22937 * config/nds32/nds32-intrinsic.md: Likewise.
22938 * config/nds32/nds32_intrinsic.h: Likewise.
22939 * config/nds32/nds32.h (nds32_builtins): Likewise.
22940 * config/nds32/nds32.opt (munaligned-access): New option.
22941 * config/nds32/nds32.c (nds32_asm_file_start): Display
22942 flag_unaligned_access status.
22944 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
22946 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
22947 -mno-relax is present.
22948 * config/riscv/linux.h (LINK_SPEC): Ditto.
22950 2018-04-20 Martin Sebor <msebor@redhat.com>
22953 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
22955 (gimple_fold_builtin_stxcpy_chk): Same.
22956 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
22958 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
22961 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
22962 __powikf2 when long double is IEEE 128-bit.
22964 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
22966 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
22967 step to make sure stack always aligned.
22969 2018-04-20 Carl Love <cel@us.ibm.com>
22972 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
22973 size check for arg0.
22975 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
22976 Tom de Vries <tom@codesourcery.com>
22979 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
22980 Emit insns for calls too.
22981 (nvptx_find_par): Always look for worker-level predecessor insn.
22982 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
22984 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
22985 (nvptx_process_pars): Propagate frames for calls.
22987 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
22990 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
22992 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
22993 (ix86_handle_option): Don't handle OPT_mibt.
22994 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
22996 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
22998 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
22999 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
23000 (ix86_target_macros): Define __CET__ with flag_cf_protection
23001 for -fcf-protection.
23002 * config/i386/i386.c (isa2_opts): Remove -mibt.
23003 * config/i386/i386.h (TARGET_IBT): Removed.
23004 (TARGET_IBT_P): Likewise.
23005 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
23006 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
23007 * config/i386/i386.opt (mcet): Update help message.
23008 (mshstk): Likewise.
23010 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
23011 -mcet as an alias for -mshstk.
23013 2018-04-20 Richard Biener <rguenther@suse.de>
23015 PR middle-end/85475
23016 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
23017 complexity by forcing a single use of the multiply operand.
23019 2018-04-20 Martin Jambor <mjambor@suse.cz>
23022 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
23023 recursion dependency to only apply to non-clones.
23025 2018-04-20 Martin Jambor <mjambor@suse.cz>
23028 * ipa-cp.c (create_specialized_node): Check that clones of
23029 self-recursive edges exist during IPA-CP.
23031 2018-04-19 Toon Moene <toon@moene.org>
23033 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
23036 2018-04-19 Jakub Jelinek <jakub@redhat.com>
23038 PR tree-optimization/85467
23039 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
23040 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
23041 VECTOR_CST element to type.
23043 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
23046 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
23047 * config/i386/i386.md (builtin_setjmp_setup): Removed.
23048 (builtin_longjmp): Likewise.
23049 (save_stack_nonlocal): New pattern.
23050 (restore_stack_nonlocal): Likewise.
23052 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
23055 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
23056 Replace ASM_OUTPUT_LABEL with fprintf.
23058 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
23061 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
23062 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
23063 * config/i386/i386-c.c (ix86_target_macros_internal): Also
23064 define __IBT__ and __SHSTK__ for -fcf-protection.
23065 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
23067 (ix86_trampoline_init): Likewise.
23068 (x86_output_mi_thunk): Likewise.
23069 (ix86_notrack_prefixed_insn_p): Likewise.
23070 (ix86_option_override_internal): Don't disallow -fcf-protection.
23071 * config/i386/i386.md (rdssp<mode>): Also enable for
23073 (incssp<mode>): Likewise.
23074 (nop_endbr): Likewise.
23075 * config/i386/i386.opt (mcet): Change help message to built-in
23078 (mshstk): Likewise.
23079 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
23080 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
23081 enable CET built-in functions.
23083 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
23085 * common/config/i386/i386-common.c
23086 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
23087 OPTION_MASK_ISA_MOVDIRI_UNSET,
23088 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
23089 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
23090 * config.gcc (movdirintrin.h): New header.
23091 * config/i386/cpuid.h (bit_MOVDIRI,
23092 bit_MOVDIR64B): New bits.
23093 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
23095 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
23096 (VOID, PVOID, PCVOID)): New function types.
23097 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
23098 __builtin_ia32_directstoreu_u64,
23099 __builtin_ia32_movdir64b): New builtins.
23100 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
23101 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
23103 (ix86_valid_target_attribute_inner_p): Ditto.
23104 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
23105 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
23106 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
23107 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
23108 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
23109 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
23110 (movdiri<mode>, movdir64b_<mode>): New.
23111 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
23112 * config/i386/immintrin.h: Include movdirintrin.h.
23113 * config/i386/movdirintrin.h: New file.
23114 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
23116 2018-04-19 Richard Biener <rguenther@suse.de>
23118 PR middle-end/85455
23119 * cfg.c (clear_bb_flags): When loop state says we have
23120 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
23122 2018-04-19 Richard Biener <rguenther@suse.de>
23124 PR tree-optimization/84737
23125 * tree-vect-data-refs.c (vect_copy_ref_info): New function
23126 copying restrict info.
23127 (vect_setup_realignment): Use it.
23128 * tree-vectorizer.h (vect_copy_ref_info): Declare.
23129 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
23130 the first DR to all generated stores.
23131 (vectorizable_load): Likewise for loads.
23133 2018-04-19 Jakub Jelinek <jakub@redhat.com>
23135 PR tree-optimization/85446
23136 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
23137 the integral and pointer types to have the same precision.
23139 * doc/install.texi: Document --disable-cet being the default and
23142 2018-04-18 Martin Liska <mliska@suse.cz>
23144 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
23147 2018-04-18 Martin Liska <mliska@suse.cz>
23150 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
23153 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
23154 arguments if they are comparable.
23156 2018-04-18 Martin Liska <mliska@suse.cz>
23159 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
23162 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
23165 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
23168 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
23169 ENDBR after calling __morestack.
23171 2018-04-18 David Malcolm <dmalcolm@redhat.com>
23174 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
23175 by using gcc_base_ver to generate a gcc_driver_version, and use
23176 it when generating GCC_DRIVER_NAME.
23177 * configure: Regenerate.
23179 2018-04-18 Jakub Jelinek <jakub@redhat.com>
23182 * config.gcc: Obsolete powerpc*-*-*spe*.
23184 2018-04-17 Jakub Jelinek <jakub@redhat.com>
23187 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
23188 (stabstr_D): Change type of unum from unsigned int to
23189 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
23192 2018-04-17 Jim Wilson <jimw@sifive.com>
23195 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
23196 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
23197 Set arg_pointer_offset after using pretend_args_size.
23199 2018-04-17 Jakub Jelinek <jakub@redhat.com>
23201 PR rtl-optimization/85431
23202 * dse.c (record_store): Ignore zero width stores.
23205 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
23206 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
23207 __builtin_stack_restore rather than after it.
23208 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
23209 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
23210 argument instead of virtual_dynamic_stack_rtx.
23212 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
23214 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
23216 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
23217 Add note to error message to explain internal mapping of overloaded
23218 built-in function name to non-overloaded built-in function name.
23219 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
23222 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
23225 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
23226 where the inputs overlap with the output.
23228 2018-04-17 Jakub Jelinek <jakub@redhat.com>
23230 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
23231 (=v, v) alternative and explicit "memory" attribute.
23232 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
23233 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
23235 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
23236 "sselog1" type instead of "sselog".
23237 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
23238 "sselog". Remove explicit "memory" attribute.
23239 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
23240 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
23242 (vec_extract_hi_v32hi): Merge all alternatives into one, use
23243 "sselog1" type instead of "sselog". Remove explicit "memory"
23245 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
23246 use "sselog1" type instead of "sselog". Remove explicit "memory"
23248 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
23249 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
23251 (vec_extract_hi_v64qi): Merge all alternatives into one, use
23252 "sselog1" type instead of "sselog". Remove explicit "memory"
23254 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
23255 use "sselog1" type instead of "sselog". Remove explicit "memory"
23259 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
23261 PR middle-end/85414
23262 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
23265 2018-04-17 Martin Jambor <mjambor@suse.cz>
23268 * ipa-cp.c (create_specialized_node): Call
23269 expand_all_artificial_thunks if necessary.
23271 2018-04-17 Martin Liska <mliska@suse.cz>
23274 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
23275 in message, remote space in between '_G' and '('.
23277 2018-04-17 Jakub Jelinek <jakub@redhat.com>
23280 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
23281 avx512f_vmcmp<mode>3<round_saeonly_name>,
23282 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
23283 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
23284 avx512f_rndscale<mode><round_saeonly_name>,
23285 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
23286 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
23287 Use %<iptr>2 instead of %2 for -masm=intel.
23288 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
23289 avx512f_vcvttss2usi<round_saeonly_name>,
23290 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
23292 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
23293 avx512f_vcvttsd2usi<round_saeonly_name>,
23294 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
23295 Use %q1 instead of %1 for -masm=intel.
23296 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
23297 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
23298 of %3 for -masm=intel.
23299 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
23301 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
23303 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
23305 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
23306 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
23308 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
23310 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
23311 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
23312 %g1 and one with %0 and %1.
23313 (avx512er_vmrcp28<mode><round_saeonly_name>,
23314 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
23315 %1 for -masm=intel.
23316 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
23317 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
23318 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
23319 of %0 and %{%4%} for -masm=intel.
23320 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
23321 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
23322 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
23323 order of %0 and %{%5%}%{z%} for -masm=intel.
23325 2018-04-17 Jan Hubicka <jh@suse.cz>
23328 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
23330 2018-04-17 Martin Liska <mliska@suse.cz>
23333 * multiple_target.c (create_dispatcher_calls): Set apostrophes
23334 for target_clone error message. Make default implementation
23335 clone to be a local declaration.
23336 (separate_attrs): Add new argument and check for an empty
23338 (expand_target_clones): Handle it.
23339 (ipa_target_clone): Make redirection just for target_clones
23342 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
23343 Tom de Vries <tom@codesourcery.com>
23345 PR middle-end/84955
23346 * omp-expand.c (expand_oacc_for): Add dummy false branch for
23347 tiled basic blocks without omp continue statements.
23349 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
23352 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
23353 vec_extract expression as having side effects to make sure it gets
23356 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
23359 * config/i386/i386.c (get_builtin_code_for_version): Check
23362 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
23365 * config.gcc: Support "skylake".
23366 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
23368 * config/i386/i386.c (m_SKYLAKE): Define.
23369 (processor_target_table): Add "skylake".
23370 (ix86_option_override_internal): Add "skylake".
23371 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
23372 PROCESSOR_CANNONLAKE.
23373 (get_builtin_code_for_version): Fix priority for
23374 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
23375 PROCESSOR_SKYLAKE-AVX512.
23376 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
23377 (processor_type): Add PROCESSOR_SKYLAKE.
23379 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
23380 Jason Merrill <jason@redhat.com>
23383 * convert.c (convert_to_integer_1): Use direct recursion for
23384 enumeral types and types with a precision less than the number
23385 of bits in their mode.
23387 2018-04-16 Julia Koval <julia.koval@intel.com>
23390 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
23391 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
23393 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
23396 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
23397 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
23398 and -mno-direct-move.
23400 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
23403 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
23404 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
23405 Ensure negative shifts result in {0}.
23407 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
23409 PR rtl-optimization/79916
23410 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
23411 regs (if any) to define how to gnerate SD moves when LRA is in
23414 2018-04-13 Jakub Jelinek <jakub@redhat.com>
23416 PR rtl-optimization/85393
23417 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
23418 * except.c (expand_dw2_landing_pad_for_region): Make static.
23419 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
23420 a label and unconditional jump to old_bb, rather than
23421 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
23424 PR rtl-optimization/85376
23425 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
23426 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
23427 instead of a specific value.
23429 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
23430 Bin Cheng <bin.cheng@arm.com>
23432 PR tree-optimization/82965
23433 PR tree-optimization/83991
23434 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
23435 by_profile_only parameter.
23436 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
23437 information if the loop was predicted to iterate too many times.
23438 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
23440 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
23443 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
23446 2018-04-13 Martin Liska <mliska@suse.cz>
23447 Jakub Jelinek <jakub@redhat.com>
23449 PR middle-end/81657
23450 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
23451 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
23452 * builtins.c (expand_builtin_memory_copy_args): Use
23453 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
23454 handle dest_addr == pc_rtx.
23456 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
23459 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
23460 asked to not generate direct moves.
23461 (fix_trunc<mode>si2_stfiwx): Similar.
23462 (fix_trunc<mode>si2_internal): Similar.
23464 2018-04-12 Jakub Jelinek <jakub@redhat.com>
23467 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
23468 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
23469 lookup if dest in some wider mode is known to be const0_rtx and
23470 if so, record permanent equivalence for it to be ZERO_EXTEND of
23471 the narrower mode destination.
23473 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
23475 * lto-streamer-out.c (output_function): Revert 259346.
23476 * omp-expand.c (expand_oacc_for): Likewise.
23478 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
23480 PR rtl-optimization/85354
23481 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
23482 * sel-sched.c (sel_global_init): ... here.
23484 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
23487 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
23488 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
23489 mode for PE-COFF targets.
23490 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
23491 (i386_pe_asm_lto_end): Likewise.
23492 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
23493 (TARGET_ASM_LTO_END): Likewise.
23494 * config/i386/winnt.c (saved_debug_info_level): New static variable.
23495 (i386_pe_asm_lto_start): New function.
23496 (i386_pe_asm_lto_end): Likewise.
23498 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
23499 Richard Biener <rguenther@suse.de>
23501 PR middle-end/84955
23502 * lto-streamer-out.c (output_function): Fix CFG loop state before
23504 * omp-expand.c (expand_oacc_for): Handle calls to internal
23505 functions like regular functions.
23507 2018-04-12 Richard Biener <rguenther@suse.de>
23510 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
23511 for the early LTO debug to properly generate references to it
23512 during DIE emission. Do not re-use that for the skeleton for
23514 (dwarf2out_early_finish): Likewise.
23516 2018-04-12 Jakub Jelinek <jakub@redhat.com>
23519 * config/i386/sse.md
23520 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
23521 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
23522 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
23523 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
23524 and output is a reg, avoid creating invalid lowpart subreg, but
23525 instead split into a 512-bit move. Don't split if not AVX512VL,
23526 input is xmm16+ reg and output is a mem.
23527 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
23528 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
23529 xmm16+ reg and output is a mem.
23531 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23533 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
23534 also for flag_dwarf2_cfi_asm.
23536 2018-04-12 Jakub Jelinek <jakub@redhat.com>
23538 PR rtl-optimization/85342
23539 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
23540 a bool scalar var inside of the loop instead. Don't try to update
23541 recog_data.operand after failed apply_change_group.
23543 2018-04-12 Tom de Vries <tom@codesourcery.com>
23546 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
23547 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
23548 array with flexible array member as array without given dimension.
23549 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
23550 argument for undefined param to true.
23552 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
23555 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
23556 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
23557 from PowerPC section.
23558 * config/rs6000/sysv4.opt (mcall-): Improve help text.
23559 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
23560 help text that is too long.
23561 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
23562 help text that is too long.
23563 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
23564 help text that is too long.
23566 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
23568 * config/alpha/alpha.md (stack_probe_internal): Rename
23569 from "probe_stack". Update all callers.
23571 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
23573 PR rtl-optimization/84566
23574 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
23575 sched_macro_fuse_insns.
23577 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
23580 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
23581 (compute_block_dependences): ... from here.
23583 2018-04-11 Jakub Jelinek <jakub@redhat.com>
23585 PR tree-optimization/85331
23586 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
23587 from int to HOST_WIDE_INT.
23589 2018-04-11 Martin Jambor <mjambor@suse.cz>
23592 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
23593 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
23594 not the same as the source val.
23595 (cgraph_edge_brings_value_p): New parameter.
23596 (gather_edges_for_value): Pass destination value to
23597 cgraph_edge_brings_value_p.
23598 (perhaps_add_new_callers): Likewise.
23599 (get_info_about_necessary_edges): Likewise and exclude values brought
23600 only by self-recursive edges.
23601 (create_specialized_node): Redirect only clones of self-calling edges.
23602 (+self_recursive_pass_through_p): New function.
23603 (find_more_scalar_values_for_callers_subset): Use it.
23604 (find_aggregate_values_for_callers_subset): Likewise.
23605 (known_aggs_to_agg_replacement_list): Removed.
23606 (decide_whether_version_node): Re-calculate known constants for all
23607 remaining context clones.
23609 2018-04-11 Richard Biener <rguenther@suse.de>
23612 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
23613 from early DWARF output.
23614 (dwarf2out_early_finish): Output line info unconditionally into
23615 early DWARF and add reference to it.
23617 2018-04-11 Jakub Jelinek <jakub@redhat.com>
23620 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
23621 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
23622 other than V2DFmode using iptr mode attribute.
23623 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
23625 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
23627 PR rtl-optimization/84659
23628 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
23630 2018-04-11 Jakub Jelinek <jakub@redhat.com>
23633 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
23635 (output_loc_list): Pass address of a dummy size variable even in the
23636 locview handling loop.
23637 (index_location_lists): Add comment on why skip_loc_list_entry can't
23640 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
23643 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
23646 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
23649 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
23650 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
23651 and -mstring-compare-inline-limit.
23653 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
23656 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
23657 for stack clash protection in a register whenever we need it to be in
23660 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
23662 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
23663 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
23665 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
23668 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
23670 (mlong-double-): Ditto.
23671 * config/rs6000/sysv4.opt (msdata=): Ditto.
23672 (mtls-size=): Ditto.
23674 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
23676 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
23677 erroneous entries for
23678 "vector int vec_ldl (int, long int *)", and
23679 "vector unsigned int vec_ldl (int, unsigned long int *)".
23680 Add comments and entries for
23681 "vector bool char vec_ldl (int, bool char *)",
23682 "vector bool short vec_ldl (int, bool short *)",
23683 "vector bool int vec_ldl (int, bool int *)",
23684 "vector bool long long vec_ldl (int, bool long long *)",
23685 "vector pixel vec_ldl (int, pixel *)",
23686 "vector long long vec_ldl (int, long long *)",
23687 "vector unsigned long long vec_ldl (int, unsigned long long *)".
23688 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
23689 type tree bool_long_long_type_node and correct definition of
23690 bool_V2DI_type_node to make reference to this new type tree.
23691 (rs6000_mangle_type): Replace erroneous reference to
23692 bool_long_type_node with bool_long_long_type_node.
23693 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
23694 comments to emphasize sign distinctions for char and int types and
23695 replace RS6000_BTI_bool_long constant with
23696 RS6000_BTI_bool_long_long constant. Also add comment to restrict
23697 use of RS6000_BTI_pixel.
23698 (bool_long_type_node): Remove this macro definition.
23699 (bool_long_long_type_node): New macro definition
23701 2018-04-10 Jakub Jelinek <jakub@redhat.com>
23703 PR rtl-optimization/85300
23704 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
23705 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
23706 simplify_unary_operation fails.
23708 2018-04-10 Martin Liska <mliska@suse.cz>
23710 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
23711 cgraph_edge and ipa_ref.
23713 2018-04-10 Jakub Jelinek <jakub@redhat.com>
23717 * config/i386/sse.md
23718 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
23719 computation of the VEC_MERGE selector from mask.
23720 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
23721 Fix decoding of the VEC_MERGE selector into mask.
23723 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
23725 PR tree-optimization/85286
23726 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
23728 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
23730 * final.c (final_1): Set insn_last_address as well as
23731 insn_current_address.
23733 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23736 * explow.c (emit_stack_probe): Call validize_mem on memory location
23737 before passing it to gen_probe_stack. Create address operand and
23738 legitimize it for the probe_stack_address case.
23740 2018-04-09 Jan Hubicka <jh@suse.cz>
23743 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
23744 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
23745 * tree.c (free_lang_data_in_type): Fix handling of binfos;
23747 (free_lang_data): Rebuild type inheritance graph.
23749 2018-04-09 Martin Sebor <msebor@redhat.com>
23751 * invoke.texi (-finline-small-functions): Mention other optimization
23753 (-findirect-inlining, -fpartial-inlining): Same.
23754 (-finline-functions-called-once): Same.
23755 (-freorder-blocks-and-partition): Same.
23757 2018-04-09 Jan Hubicka <jh@suse.cz>
23760 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
23761 jumps; choose last target that matches the criteria (i.e.
23762 no partition changes for non-crossing jumps).
23763 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
23764 support for redirecting crossing jumps to non-crossing.
23766 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
23768 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
23769 also for naked functions.
23771 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
23773 * config/arc/arc.md (add_shift): New pattern.
23774 (add_shift2): Likewise.
23775 (sub_shift): Likewise.
23776 (sub_shift_cmp0_noout): Likewise.
23777 (compare_si_ashiftsi): Likewise.
23778 (xbfu_cmp0_noout): New combine pattern.
23779 (xbfu_cmp0"): Likewise.
23780 (movsi_set_cc_insn): Place the predicable variant first.
23781 (commutative_binary_cmp0_noout): Remove clobber.
23782 (commutative_binary_cmp0): New pattern.
23783 (noncommutative_binary_cmp0): Likewise.
23784 (noncommutative_binary_cmp0_noout): Likewise.
23785 (noncommutative_binary_comparison_result_used): Removed.
23786 (rsub_cmp0): New pattern.
23787 (rsub_cmp0_noout): Likewise.
23788 (extzvsi): Changed, keep only meaningful variants.
23789 (SQH, SEZ): New iterators.
23790 (SQH_postfix): New mode attribute.
23791 (SEZ_prefix): New code attribute.
23792 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
23793 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
23794 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
23795 of numerical value.
23796 (noncommutative_operator): Check the availability of barrel
23799 2018-04-09 Richard Biener <rguenther@suse.de>
23801 PR tree-optimization/85284
23802 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
23803 Only use the niter constraining form of simple_iv when the exit
23804 is always executed.
23806 2018-04-09 Tom de Vries <tom@codesourcery.com>
23809 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
23810 (define_expand "*memory_barrier"): New define_expand.
23811 (define_insn "memory_barrier"): New insn.
23813 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
23815 PR rtl-optimization/80463
23816 PR rtl-optimization/83972
23817 PR rtl-optimization/83480
23819 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
23820 correct producer for the insn.
23821 (tidy_control_flow): Fixup seqnos in case of debug insns.
23823 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
23825 PR rtl-optimization/83913
23827 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
23828 different sched-times when merging exprs.
23830 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
23832 PR rtl-optimization/83962
23834 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
23835 tidy_fallthru_edge and tidy_control_flow.
23837 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
23839 PR rtl-optimization/83530
23841 * sel-sched.c (force_next_insn): New global variable.
23842 (remove_insn_for_debug): When force_next_insn is true, also leave only
23843 next insn in the ready list.
23844 (sel_sched_region): When the region wasn't scheduled, make another pass
23845 over it with force_next_insn set to 1.
23847 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
23849 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
23851 * config/nds32/constants.md (unspec_volatile_element): Add enum values
23852 for interrupt control.
23853 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
23854 functions for interrupt control.
23855 * config/nds32/nds32-intrinsic.md: Likewise.
23856 * config/nds32/nds32_intrinsic.h: Likewise.
23857 * config/nds32/nds32.h (nds32_builtins): Likewise.
23859 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
23861 * config/nds32/nds32.c (nds32_init_machine_status,
23862 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
23863 strict_aligned_p field.
23864 (nds32_expand_to_rtl_hook): New function.
23865 (TARGET_EXPAND_TO_RTL_HOOK): Define.
23866 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
23868 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
23869 Chung-Ju Wu <jasonwucj@gmail.com>
23871 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
23872 * config/nds32/nds32-n7.md: New file.
23873 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
23874 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
23876 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
23877 * config/nds32/nds32.md (pipeline_model): Add n7.
23878 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
23879 * config/nds32/pipelines.md: Include n7 settings.
23881 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
23882 Chung-Ju Wu <jasonwucj@gmail.com>
23884 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
23885 * config/nds32/nds32-e8.md: New file.
23886 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
23887 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
23889 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
23890 * config/nds32/nds32.md (pipeline_model): Add e8.
23891 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
23892 * config/nds32/pipelines.md: Include e8 settings.
23894 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
23895 Chung-Ju Wu <jasonwucj@gmail.com>
23897 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
23898 * config/nds32/nds32-n8.md: New file.
23899 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
23900 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
23902 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
23903 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
23904 * config/nds32/nds32.md (pipeline_model): Add n8.
23905 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
23906 * config/nds32/pipelines.md: Include n8 settings.
23908 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
23909 Chung-Ju Wu <jasonwucj@gmail.com>
23911 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
23912 * config/nds32/nds32-n9-2r1w.md: New file.
23913 * config/nds32/nds32-n9-3r2w.md: New file.
23914 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
23915 nds32_register_ports): New or modify for cpu n9.
23916 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
23918 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
23919 * config/nds32/nds32-utils.c: New file.
23920 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
23921 TARGET_MUL_SLOW): Define.
23922 * config/nds32/nds32.md (pipeline_model): New attribute.
23923 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
23924 New options that support cpu n9.
23925 * config/nds32/pipelines.md: Include n9 settings.
23926 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
23928 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
23930 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
23931 information if necessary.
23932 (output_cond_branch_compare_zero): Likewise.
23933 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
23934 (nds32_target_alignment): Refine for alignment.
23935 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
23936 (FUNCTION_BOUNDARY): Modify.
23937 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
23939 * config/nds32/nds32.opt (malways-align, malign-functions): New.
23941 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
23943 * config/nds32/constants.md (unspec_volatile_element): Add values for
23944 TLB operation and data prefetch.
23945 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
23946 functions for TLB operation and data prefetch.
23947 * config/nds32/nds32-intrinsic.md: Likewise.
23948 * config/nds32/nds32_intrinsic.h: Likewise.
23949 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
23950 (nds32_print_operand): Likewise.
23951 * config/nds32/nds32.h (nds32_builtins): Likewise.
23953 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
23954 Andrew Pinski <pinsika@gcc.gnu.org>
23956 PR middle-end/82976
23957 * match.pd: Use constant_boolean_node of correct type instead of
23958 boolean_true_node or boolean_false_node for simplifying
23959 pointer comparisons to zero.
23961 2018-04-07 Jakub Jelinek <jakub@redhat.com>
23963 PR tree-optimization/80021
23964 * tree.c (verify_type_variant): Make error call in verify_variant_match
23965 translatable and remove final full stop.
23967 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
23969 * config/nds32/constants.md (unspec_volatile_element): Add
23970 UNSPEC_VOLATILE_EH_RETURN.
23971 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
23972 nds32_output_stack_pop): Support dwarf exception handling process.
23973 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
23974 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
23975 exception handling process.
23976 (nds32_compute_stack_frame): Likewise.
23977 (nds32_return_addr_rtx): Likewise.
23978 (nds32_initial_elimination_offset): Likewise.
23979 (nds32_expand_prologue): Likewise.
23980 (nds32_expand_epilogue): Likewise.
23981 (nds32_dynamic_chain_address): New function.
23982 * config/nds32/nds32.h (machine_function): Add fields for dwarf
23983 exception handling.
23984 (DYNAMIC_CHAIN_ADDRESS): Define.
23985 (EH_RETURN_DATA_REGNO): Define.
23986 (EH_RETURN_STACKADJ_RTX): Define.
23987 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
23988 patterns for dwarf exception handling.
23990 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
23992 * config/nds32/nds32.h: Clean up obsolete macros.
23994 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
23996 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
23997 Add enum values for particular instructions.
23998 * config/nds32/nds32-intrinsic.c: Implementation of expanding
23999 particular intrinsic functions.
24000 * config/nds32/nds32-intrinsic.md: Likewise.
24001 * config/nds32/nds32_intrinsic.h: Likewise.
24002 * config/nds32/nds32.h (nds32_builtins): Likewise.
24003 * config/nds32/nds32.md (type): Add pbsad and pbsada.
24004 (btst, ave): New patterns for particular instructions.
24006 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
24008 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
24009 Add enum values for atomic load/store and memory sync.
24010 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
24012 * config/nds32/nds32-intrinsic.md: Likewise.
24013 * config/nds32/nds32_intrinsic.h: Likewise.
24014 * config/nds32/nds32.h (nds32_builtins): Likewise.
24016 2018-04-07 Jakub Jelinek <jakub@redhat.com>
24018 PR tree-optimization/85257
24019 * fold-const.c (native_encode_vector): If not all elts could fit
24020 and off is -1, return 0 rather than offset.
24021 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
24022 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
24023 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
24024 adjust buffer in native_interpret_expr call.
24026 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
24028 * config/nds32/constants.md (unspec_volatile_element): Add cache
24029 control enum values.
24030 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
24031 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
24032 * config/nds32/nds32.c (nds32_cctl_names): New.
24033 (nds32_print_operand): Handle cache control register names.
24034 * config/nds32/nds32.h (nds32_builtins): New enum values.
24035 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
24037 * config/nds32/nds32.md (type): Add mmu.
24038 * config/nds32/pipelines.md (simple_insn): Add mmu.
24040 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
24042 * config/nds32/nds32.md (type): Remove call.
24043 * config/nds32/pipelines.md (simple_insn): Likewise.
24045 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
24047 * config/nds32/constants.md (unspec_volatile_element): Add
24048 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
24049 UNSPEC_VOLATILE_FMFCFG.
24050 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
24051 description for fmfcfg and fmfcsr.
24052 (bdesc_1arg): Add fmtcsr.
24053 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
24054 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
24055 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
24056 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
24057 unspec_fmfcfg): New patterns.
24058 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
24059 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
24060 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
24061 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
24062 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
24063 __nds32__fmfcfg): Define.
24065 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
24067 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
24068 intrinsic register names.
24069 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
24070 intrinsic register enum values and macros.
24072 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
24074 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
24075 for load/store addressing form.
24076 (nds32_print_operand_address): Likewise.
24078 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
24081 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
24082 based on LABEL_REF. Remove useless assertion.
24083 (pic_address_needs_scratch): Fix formatting.
24084 (sparc_legitimize_pic_address): Minor tweaks.
24085 (sparc_delegitimize_address): Adjust assertion accordingly.
24086 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
24087 into symbolic_operand.
24088 (movsi_high_pic_label_ref): Likewise.
24089 (movsi_lo_sum_pic_label_ref): Likewise.
24090 (movdi_pic_label_ref): Likewise.
24091 (movdi_high_pic_label_ref): Likewise.
24092 (movdi_lo_sum_pic_label_ref): Likewise.
24094 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
24096 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
24097 custom LIB_SPEC setup.
24099 2018-04-06 Ruslan Bukin <br@bsdpad.com>
24100 Kito Cheng <kito.cheng@gmail.com>
24102 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
24103 * config/riscv/freebsd.h: New.
24105 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
24107 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
24108 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
24111 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
24112 Kito Cheng <kito.cheng@gmail.com>
24114 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
24115 nds32_output_call, nds32_symbol_binds_local_p): New functions.
24116 * config/nds32/nds32-protos.h (nds32_output_call,
24117 nds32_output_return): Declare.
24118 * config/nds32/nds32.md: Refine all the call and return patterns.
24120 2018-04-06 Jakub Jelinek <jakub@redhat.com>
24123 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
24124 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
24126 PR rtl-optimization/84872
24127 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
24128 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
24129 EDGE_CROSSING edge.
24131 2018-04-06 Tamar Christina <tamar.christina@arm.com>
24133 * expr.c (copy_blkmode_to_reg): Revert 254862.
24134 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
24136 2018-04-06 Richard Biener <rguenther@suse.de>
24138 PR middle-end/85244
24139 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
24140 after seeing a component reference with an adjacent field. Treat
24141 refs to arrays at struct end of external decls similar to
24142 refs to unconstrained commons.
24144 2018-04-06 Jakub Jelinek <jakub@redhat.com>
24147 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
24148 look through SAVE_EXPRs with non-side-effects argument. Adjust
24150 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
24153 2018-04-06 Richard Biener <rguenther@suse.de>
24155 PR middle-end/85180
24156 * alias.c (find_base_term): New wrapper around find_base_term
24157 unwinding CSELIB_VAL_PTR changes.
24158 (find_base_term): Do not restore CSELIB_VAL_PTR during the
24161 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24163 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
24165 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
24166 constant definitions.
24167 ("nop"): lr 0,0 -> nopr r0
24168 ("nop_lr0", "nop_lr1"): New insn definitions.
24170 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
24172 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
24173 NDS32_V3PUSH_AVAILABLE_P macro.
24175 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
24176 Chung-Ju Wu <jasonwucj@gmail.com>
24178 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
24179 (nds32*-*-*): Add float and fpu_config into supported_defaults.
24180 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
24181 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
24182 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
24183 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
24184 * config/nds32/constraints.md: New constraints and checking for hard
24185 float configuration.
24186 * config/nds32/iterators.md: New mode iterator and attribute for hard
24187 float configuration.
24188 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
24190 * config/nds32/nds32-fpu.md: New file.
24191 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
24192 deal with hard float code generation.
24193 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
24195 (abi_type, float_reg_number): New enum type.
24196 * config/nds32/nds32-predicates.c: New predicates for hard float.
24197 * config/nds32/nds32-protos.h: Declare functions for hard float.
24198 * config/nds32/nds32.c: Implementation for hard float configuration.
24199 * config/nds32/nds32.h: Definitions for hard float configuration.
24200 * config/nds32/nds32.md: Include hard float machine description and
24201 modify patterns for hard float configuration.
24202 * config/nds32/nds32.opt: New options for hard float configuration.
24203 * config/nds32/predicates.md: New predicates for hard float
24206 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
24208 * common/config/nds32/nds32-common.c
24209 (nds32_option_optimization_table): Enable -mreleax-hint by default.
24211 2018-04-05 Jakub Jelinek <jakub@redhat.com>
24213 PR middle-end/85195
24214 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
24215 CONSTRUCTOR_ELT (ctor, ...)->value.
24217 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
24220 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
24222 2018-04-05 Tom de Vries <tom@codesourcery.com>
24225 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
24228 2018-04-05 Shiva Chen <shiva0217@gmail.com>
24229 Kito Cheng <kito.cheng@gmail.com>
24231 * config/nds32/constraints.md (U33): Fine-tune checking condition.
24232 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
24233 * config/nds32/nds32.h (nds32_16bit_address_type): Add
24234 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
24236 2018-04-05 Shiva Chen <shiva0217@gmail.com>
24237 Kito Cheng <kito.cheng@gmail.com>
24239 * config/nds32/constraints.md (Ufe): New memory constraint.
24240 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
24241 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
24242 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
24244 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
24245 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
24247 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
24249 * config/nds32/nds32.md: Use optimize_size in the condition for
24250 alu-shift instructions.
24252 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
24254 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
24256 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
24258 * config/nds32/nds32.md (negsi2): Refine pattern.
24260 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
24261 Chung-Ju Wu <jasonwucj@gmail.com>
24263 * config/nds32/iterators.md (shift_rotate): New code iterator.
24264 (shift): New code attribute.
24265 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
24266 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
24267 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
24268 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
24269 bit-wise operations.
24270 (andsi3, *andsi3): Ditto.
24271 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
24272 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
24273 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
24274 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
24275 nds32_ior_operand, nds32_xor_operand): New predicates.
24277 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
24279 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
24280 (addsi3, subsi3): ... this.
24282 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
24284 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
24286 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
24288 * config/nds32/nds32.md: Adjust indention.
24290 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
24292 * config/nds32/nds32.md (feature): New attribute.
24294 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
24296 * config/nds32/nds32.md (subtype): New attribute.
24298 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24301 * config/arm/arm-builtins.c (arm_expand_builtin): Change
24302 expansion to perform a bitwise AND of the argument followed by a
24303 boolean negation of the result.
24305 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
24307 PR rtl-optimization/84878
24308 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
24309 the basic block. Assert the use reference is not artificial and that
24310 it has an associated insn.
24312 2018-04-04 Michael Matz <matz@suse.de>
24314 * builtins.c (compute_objsize): Pass correct operand
24315 to array_at_struct_end_p.
24317 2018-04-04 Richard Biener <rguenther@suse.de>
24320 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
24321 from contexts for DINFO_LEVEL_TERSE and below.
24323 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
24325 * config/nds32/nds32-doubleword.md (move_<mode>): Require
24326 resiter_operand condition.
24327 * config/nds32/nds32.md (*move<mode>): Ditto.
24329 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
24330 Monk Chiang <sh.chiang04@gmail.com>
24332 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
24334 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
24336 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
24338 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
24339 Kito Cheng <kito.cheng@gmail.com>
24341 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
24342 nds32_cond_code_str, output_cond_branch,
24343 output_cond_branch_compare_zero, nds32_expand_cbranch,
24344 nds32_expand_cstore, nds32_expand_movcc,
24345 nds32_output_cbranchsi4_equality_zero,
24346 nds32_output_cbranchsi4_equality_reg,
24347 nds32_output_cbranchsi4_equality_reg_or_const_int,
24348 nds32_output_cbranchsi4_greater_less_zero: New functions.
24349 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
24350 nds32_expand_cstore, nds32_expand_movcc,
24351 nds32_output_cbranchsi4_equality_zero,
24352 nds32_output_cbranchsi4_equality_reg,
24353 nds32_output_cbranchsi4_equality_reg_or_const_int,
24354 nds32_output_cbranchsi4_greater_less_zero): Declare.
24355 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
24356 nds32_rimm11s_operand): New predicates.
24357 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
24358 * config/nds32/nds32.md: Rewrite all the branch and conditional move
24361 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
24363 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
24364 * config/nds32/nds32.md: Ditto.
24365 * config/nds32/pipelines.md: Ditto.
24367 2018-04-04 Richard Biener <rguenther@suse.de>
24369 PR tree-optimization/85168
24370 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
24371 propagating abnormals.
24373 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
24375 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
24377 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
24378 Kito Cheng <kito.cheng@gmail.com>
24380 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
24381 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
24382 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
24383 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
24384 * config/nds32/nds32.md (sibcall_internal): New.
24385 (sibcall_register): Remove.
24386 (sibcall_immediate): Remove.
24387 (sibcall_value_internal): New.
24388 (sibcall_value_register): Remove.
24389 (sibcall_value_immediate): Remove.
24390 * config/nds32/predicates.md (nds32_general_register_operand): New.
24391 (nds32_call_address_operand): New.
24393 2018-04-03 Jakub Jelinek <jakub@redhat.com>
24395 PR rtl-optimization/85167
24396 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
24397 bb_defs if *split_p, instead preinitialize it to NULL.
24399 PR tree-optimization/85156
24400 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
24401 evaluating the argument multiple times.
24403 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
24405 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
24407 (_mm_cvtpd_ps): Likewise.
24408 (_mm_cvttpd_epi32): Likewise.
24409 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
24410 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
24411 vector, pixel, and bool following altivec.h include.
24413 2018-04-03 Martin Sebor <msebor@redhat.com>
24415 * doc/extend.texi (Common Function Attributes): Clarify.
24416 (const attribute): Likewise.
24417 (pure attribute): Likewise.
24419 2018-04-03 Jakub Jelinek <jakub@redhat.com>
24422 * config/i386/i386.c (ix86_expand_vector_set): Use
24423 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
24425 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
24427 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
24428 instructions when changing rounding bits to preserve precision bits
24429 in the x87 control word.
24431 2018-04-03 Martin Liska <mliska@suse.cz>
24433 PR tree-optimization/82491
24434 * rtl.h (strip_offset_and_add): Replace += suboffset with
24435 poly_uint64 () + suboffset.
24437 2018-03-29 Martin Liska <mliska@suse.cz>
24438 Martin Jambor <mjambor@suse.cz>
24441 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
24442 param_type is not an integral or pointer type.
24444 2018-04-03 Richard Biener <rguenther@suse.de>
24446 * sese.h (recompute_all_dominators): Remove.
24448 2018-04-02 Martin Sebor <msebor@redhat.com>
24450 * doc/invoke.texi (-Wrestrict): Fix typos.
24452 2018-04-02 Jim Wilson <jimw@sifive.com>
24454 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
24455 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
24456 (<optab>di3, <optab>si3_extend): Likewise.
24457 (<optab>si3_mask, <optab>si3_mask_1): New.
24458 (<optab>di3_mask, <optab>di3_mask_1): New.
24459 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
24460 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
24461 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
24463 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
24465 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
24468 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
24470 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
24471 (nds32_canonicalize_comparison): New function.
24473 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
24474 Kito Cheng <kito.cheng@gmail.com>
24475 Kuan-Lin Chen <kuanlinchentw@gmail.com>
24477 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
24478 * config/nds32/constants.md (unspec_volatile_element): Add
24479 UNSPEC_VOLATILE_RELAX_GROUP.
24480 * config/nds32/nds32-relax-opt.c: New file.
24481 * config/nds32/nds32-predicates.c
24482 (nds32_symbol_load_store_p): New function.
24483 * config/nds32/nds32-protos.h
24484 (nds32_symbol_load_store_p): Declare function.
24485 (make_pass_nds32_relax_opt): Declare new rtl pass function.
24486 * config/nds32/nds32.c
24487 (nds32_register_pass): New function to register pass.
24488 (nds32_register_passes): New function to register passes.
24489 * config/nds32/nds32.md (relax_group): New pattern.
24490 * config/nds32/nds32.opt (mrelax-hint): New option.
24491 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
24493 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
24495 * config/nds32/t-nds32: Modify files dependency.
24497 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
24499 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
24500 (PROFILE_HOOK): Define its implementation.
24502 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
24504 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
24505 type and 32-bit size.
24507 2018-04-01 Jakub Jelinek <jakub@redhat.com>
24509 PR middle-end/85090
24510 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
24511 (V_128_256): New mode iterator.
24512 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
24513 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
24514 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
24516 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
24517 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
24519 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
24522 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
24523 NaN inputs correctly.
24525 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
24528 * config/rs6000/vsx.md (??r): New mode attribute.
24529 (*vsx_mov<mode>_64bit): Use it.
24530 (*vsx_mov<mode>_32bit): Likewise.
24532 2018-03-30 Martin Sebor <msebor@redhat.com>
24534 PR tree-optimization/84818
24535 * builtins.c (check_access): Use warning_n.
24537 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24540 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
24542 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
24545 2018-03-30 Julia Koval <julia.koval@intel.com>
24548 * x86-tune.def (movx, partial_reg_dependency): Enable for
24551 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
24553 PR inline-asm/84985
24554 * lra-constraints.c (process_alt_operands): Move setting
24555 this_alternative_matches below.
24557 2018-03-29 Martin Liska <mliska@suse.cz>
24560 * doc/invoke.texi: Document how LTO works with debug info.
24561 Describe auto-load support of binutils. Mention 'x86-64'
24562 as valid option value of -march option.
24564 2018-03-29 Jakub Jelinek <jakub@redhat.com>
24566 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
24569 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
24570 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
24571 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
24574 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
24577 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
24578 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
24579 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
24580 for RS6000_BTM_POWERPC64.
24581 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
24582 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
24583 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
24586 (DIVDEU): Likewise.
24588 2018-03-28 Carl Love <cel@us.ibm.com>
24591 2017-09-27 Carl Love <cel@us.ibm.com>
24593 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
24594 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
24595 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
24598 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24600 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
24601 instead of __vector bool.
24602 (_mm_max_pu8): Likewise.
24603 (_mm_min_pi16): Likewise.
24605 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
24608 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
24609 (DIVWEUO): Likewise.
24610 (DIVDEO): Likewise.
24611 (DIVDEUO): Likewise.
24612 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
24613 DIVWEUO and DIVDEUO.
24614 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
24615 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
24616 (div_extend): Likewise.
24617 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
24619 (__builtin_divweuo): Likewise.
24620 (__builtin_divdeo): Likewise.
24621 (__builtin_divdeuo): Likewise.
24623 2018-03-28 Jakub Jelinek <jakub@redhat.com>
24626 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
24627 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
24629 PR tree-optimization/82004
24630 * gimple-match-head.c (optimize_pow_to_exp): New function.
24631 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
24632 Don't fold to exp if optimize_pow_to_exp is false.
24634 2018-03-28 Martin Liska <mliska@suse.cz>
24637 * calls.c (initialize_argument_information): Fix trailing space.
24638 * common.opt: Fix typo and provide better explanation for
24639 -fsanitize-coverage option.
24640 * config/i386/i386.opt: Fix typo.
24642 2018-03-28 Jakub Jelinek <jakub@redhat.com>
24643 Martin Liska <mliska@suse.cz>
24646 * gimplify.c (asan_poison_variable): Don't do the check for
24647 gimplify_omp_ctxp here.
24648 (gimplify_decl_expr): Do it here.
24649 (gimplify_target_expr): Likewise.
24651 2018-03-28 Martin Liska <mliska@suse.cz>
24654 * config/i386/i386.c (ix86_function_arg_advance): Do not call
24655 chkp_type_bounds_count if MPX is not enabled.
24657 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
24659 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
24661 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
24664 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
24665 function to create the function decl for complex long double
24666 multiply and divide for -mabi=ieeelongdouble.
24667 (init_float128_ieee): Call it.
24669 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
24672 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
24673 -fcf-protection=branch -mibt.
24674 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
24676 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24679 * config/arm/arm.c (arm_valid_symbolic_address): Handle
24680 arm_word_relocations.
24682 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
24685 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
24686 extern array declarations.
24688 2018-03-27 Richard Biener <rguenther@suse.de>
24690 PR middle-end/84067
24691 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
24692 explicit single_use checks.
24694 2018-03-27 Richard Biener <rguenther@suse.de>
24696 PR tree-optimization/85082
24697 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
24700 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24702 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
24703 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
24704 Turn on fasynchronous-unwind-tables and funwind-tables.
24706 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
24709 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
24710 (*bmi_blsr_<mode>_ccz): Ditto.
24712 2018-03-26 Tom de Vries <tom@codesourcery.com>
24714 PR tree-optimization/85063
24715 * omp-general.c (offloading_function_p): New function. Factor out
24717 * omp-offload.c (pass_omp_target_link::gate): ... here.
24718 * omp-general.h (offloading_function_p): Declare.
24719 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
24720 with attribute omp declare target for offloading functions.
24722 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
24724 PR tree-optimization/84005
24725 * tree-data-ref.h (get_base_for_alignment): Declare.
24726 * tree-data-ref.c (get_base_for_alignment_1): New function.
24727 (get_base_for_alignment): Likewise.
24728 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
24729 get_base_for_alignment to find a suitable base object, instead
24730 of always using drb->base_address.
24732 2018-03-23 Jakub Jelinek <jakub@redhat.com>
24734 PR inline-asm/85022
24735 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
24736 known size by default.
24738 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
24740 PR inline-asm/85030
24741 * lra-constraints.c (process_alt_operands): Don't match BLKmode
24742 and non BLKmode operands.
24744 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24747 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
24748 Clean up attributes.
24750 2018-03-23 Richard Biener <rguenther@suse.de>
24753 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
24754 we are going to emit early debug for LTO.
24756 2018-03-23 Jakub Jelinek <jakub@redhat.com>
24758 PR inline-asm/85034
24759 * function.c (match_asm_constraints_1): Don't optimize if input
24760 doesn't satisfy general_operand predicate for output's mode.
24762 PR inline-asm/85022
24763 * alias.c (write_dependence_p): Don't require for x_canonicalized
24764 non-VOIDmode if x has VOIDmode.
24767 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
24768 just don't try to optimize it rather than assert it never happens.
24770 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
24772 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
24773 macro expansions for definition of ST_INTERNAL_<mode> and
24774 LD_INTERNAL_<mode> builtins.
24775 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
24777 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
24779 (altivec_expand_st_builtin): Likewise.
24780 (altivec_expand_builtin): Remove calls to deleted functions.
24781 (rs6000_address_for_altivec): Delete this function.
24782 * config/rs6000/vector.md: Remove expands for
24783 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
24785 2018-03-22 Sudakshina Das <sudi.das@arm.com>
24788 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
24789 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
24790 re-computing once computed.
24791 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
24792 (arm_init_machine_status): Initialize
24793 machine->static_chain_stack_bytes.
24795 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
24798 * doc/extend.texi: Add four new prototypes for vec_ld.
24799 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
24800 definitions for more logical presentation.
24801 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
24802 entries for V1TI variants of __builtin_altivec_ld builtin.
24803 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
24804 handling of V1TI variant of LVX icode pattern.
24805 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
24806 (rs6000_gimple_fold_builtin): Likewise.
24807 (altivec_init_builtins): Add code to define
24808 __builtin_altivec_lvx_v1ti function.
24810 2018-03-22 Jakub Jelinek <jakub@redhat.com>
24812 PR inline-asm/84941
24813 * function.c (match_asm_constraints_1): Don't do the optimization
24814 if input isn't a REG, SUBREG, MEM or constant.
24816 2018-03-22 Tom de Vries <tom@codesourcery.com>
24818 PR tree-optimization/84956
24819 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
24820 bb_has_abnormal_pred.
24822 2018-03-22 Jakub Jelinek <jakub@redhat.com>
24825 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
24826 DECL_INITIAL (decl) to decl at the end.
24827 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
24828 adjust the comment.
24830 2018-03-21 Joseph Myers <joseph@codesourcery.com>
24832 * doc/extend.texi (__builtin_tgmath): Document when complex
24833 integer types are treated as _Complex _Float64.
24835 2018-03-21 Tom de Vries <tom@codesourcery.com>
24837 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
24839 2018-03-21 Jakub Jelinek <jakub@redhat.com>
24841 PR tree-optimization/84960
24842 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
24843 if it is ENTRY block, move them into single succ of ENTRY in that case.
24845 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
24847 PR tree-optimization/84811
24848 * poly-int.h (poly_span_traits): Remove the T3 parameter and
24849 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
24850 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
24851 (known_subrange_p): Update accordingly. Cast each value involved
24852 in the size comparison, rather than casting the result of the
24855 2018-03-21 Jakub Jelinek <jakub@redhat.com>
24857 PR tree-optimization/84982
24858 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
24859 by flipping the least significant bit rather than all bits from
24860 bitpos to bitpos + bitsize - 1.
24862 2018-03-21 Nathan Sidwell <nathan@acm.org>
24864 * doc/extend.texi (Deprecated Features): Remove mention of
24865 long-deleted deprecations.
24867 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24870 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
24871 * configure: Regenerate.
24873 2018-03-21 Tom de Vries <tom@codesourcery.com>
24875 PR tree-optimization/83126
24876 * tree-parloops.c (num_phis): New function.
24877 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
24879 2018-03-21 Nathan Sidwell <nathan@acm.org>
24881 * doc/extend.texi (Deprecated Features): Update deprecated flags,
24882 mention anon-struct/union members and trailing attributes.
24884 2018-03-21 Bin Cheng <bin.cheng@arm.com>
24886 PR tree-optimization/84969
24887 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
24888 builtin memset partitions if they set different rhs values.
24890 2018-03-21 Jakub Jelinek <jakub@redhat.com>
24892 PR rtl-optimization/84989
24893 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
24894 VEC_DUPLICATE with scalar result mode.
24896 2018-03-21 Martin Liska <mliska@suse.cz>
24899 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
24900 not intended return statement.
24902 2018-03-21 Martin Liska <mliska@suse.cz>
24905 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
24906 (chkp_find_bound_slots_1): Limit number of iterations.
24908 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
24911 * Minor grammar fixes for x86 options.
24913 2018-03-20 Jakub Jelinek <jakub@redhat.com>
24916 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
24917 holding REG_CFA_RESTORE notes, instead turn them into a USE.
24919 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
24922 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
24923 (altivec_lvx_<mode>_1op): Likewise.
24924 (altivec_stvx_<mode>_2op): Likewise.
24925 (altivec_stvx_<mode>_1op): Likewise.
24926 (altivec_lvx_<VM2:mode>): New define_expand.
24927 (altivec_stvx_<VM2:mode>): Likewise.
24928 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
24929 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
24930 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
24931 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
24932 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
24933 (rs6000_gen_lvx): Likewise.
24934 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
24935 (altivec_expand_stv_builtin): Likewise.
24936 (altivec_expand_builtin): Likewise.
24937 * config/rs6000/vector.md: Likewise.
24939 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24942 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
24945 2018-03-20 Richard Biener <rguenther@suse.de>
24948 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
24949 sign-conversions as zero, fall back to standard scalar_stmt
24952 2018-03-20 Martin Liska <mliska@suse.cz>
24955 * predict.c (rebuild_frequencies): Handle case when we have
24956 PROFILE_ABSENT, but flag_guess_branch_prob is false.
24958 2018-03-20 Jakub Jelinek <jakub@redhat.com>
24961 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
24962 flag_section_anchors.
24963 * varasm.c (use_blocks_for_decl_p): Remove hack for
24964 dw2_force_const_mem.
24967 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
24969 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
24970 be created, use lowpart_subreg of operands[0] rather than operands[0]
24972 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
24973 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
24974 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
24975 and n constraint instead of aarch64_shift_imm_di and Usd.
24976 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
24977 (*aarch64_<optab>_reg_minus<mode>3): ... this.
24979 2018-03-20 Sudakshina Das <sudi.das@arm.com>
24982 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
24983 to favor GPR over NEON registers.
24984 (<shift>di3_neon): Likewise.
24986 2018-03-20 Tom de Vries <tom@codesourcery.com>
24989 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
24990 (nvptx_process_pars): Emit bar.sync asap and alap.
24992 2018-03-20 Tom de Vries <tom@codesourcery.com>
24995 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
24996 seen_label if seen_label is already set.
24998 2018-03-20 Jakub Jelinek <jakub@redhat.com>
25001 * config/i386/i386.c (fold_builtin_cpu): For features above 31
25002 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
25003 Use 1U instead of 1. Formatting fixes.
25006 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
25007 instead of TREE_TYPE (s1) for the return value.
25009 2018-03-19 Jakub Jelinek <jakub@redhat.com>
25011 PR tree-optimization/84946
25012 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
25013 bitsize + bitsize in poly_uint64 rather than poly_int64.
25016 * dwarf2asm.c: Include fold-const.c.
25017 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
25018 of decl rather than decl itself.
25020 PR rtl-optimization/84643
25021 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
25023 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
25026 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
25027 calling assemble_variable.
25029 2018-03-19 Sudakshina Das <sudi.das@arm.com>
25032 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
25033 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
25035 2018-03-19 Jim Wilson <jimw@sifive.com>
25038 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
25039 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
25040 (riscv_first_stack_step): Likewise.
25041 (riscv_option_override): Use STACK_BOUNDARY instead of
25042 MIN_STACK_BOUNDARY.
25043 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
25044 MIN_STACK_BOUNDARY.
25045 (BIGGEST_ALIGNMENT): Set to 128.
25046 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
25047 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
25050 2018-03-19 Richard Biener <rguenther@suse.de>
25052 PR tree-optimization/84933
25053 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
25054 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
25056 2018-03-19 Richard Biener <rguenther@suse.de>
25058 PR tree-optimization/84859
25059 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
25060 (cond_if_else_store_replacement): Perform sinking operation on
25061 single-store BBs regardless of MAX_STORES_TO_SINK setting.
25062 Generalize what a BB with a single eligible store is.
25064 2018-03-19 Richard Biener <rguenther@suse.de>
25066 PR tree-optimization/84929
25067 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
25068 chrec_is_positive against non-chrec arg.
25070 2018-03-19 Tamar Christina <tamar.christina@arm.com>
25073 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
25075 2018-03-18 Martin Liska <mliska@suse.cz>
25077 PR rtl-optimization/84635
25078 * regrename.c (build_def_use): Use matches_mode only when
25081 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
25083 PR tree-optimization/84913
25084 * tree-vect-loop.c (vectorizable_reduction): Don't try to
25085 vectorize chains of COND_EXPRs.
25087 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
25089 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
25091 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
25093 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
25095 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
25097 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
25099 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
25100 Kito Cheng <kito.cheng@gmail.com>
25102 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
25103 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
25104 (nds32_adjust_reg_alloc_order): New function.
25105 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
25107 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
25109 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
25110 nds32_print_operand, nds32_print_operand_address): Use
25111 HOST_WIDE_INT_PRINT_DEC instead.
25113 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
25115 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
25117 2018-03-17 Jakub Jelinek <jakub@redhat.com>
25120 * config/i386/i386.c (initial_ix86_tune_features,
25121 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
25122 unsigned long long.
25123 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
25124 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
25125 rather than 1u << ix86_tune. Formatting fix.
25126 (ix86_option_override_internal): Change ix86_arch_mask from
25127 unsigned int to unsigned HOST_WIDE_INT, initialize to
25128 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
25129 (ix86_function_specific_restore): Likewise.
25131 2018-03-16 Jakub Jelinek <jakub@redhat.com>
25134 * postreload.c (reload_combine_recognize_pattern): Perform
25135 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
25136 truncate_int_for_mode the result for the destination's mode.
25139 * hsa-gen.c (mem_type_for_type): Fix comment typo.
25140 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
25142 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
25145 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
25148 * lra-assigns.c (lra_split_hard_reg_for): Don't use
25149 regno_allocno_class_array and sorted_pseudos.
25150 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
25151 insns where regno is used.
25153 2018-03-16 Martin Liska <mliska@suse.cz>
25156 * multiple_target.c (create_dispatcher_calls): Redirect
25157 reference in the symbol table.
25159 2018-03-16 Martin Liska <mliska@suse.cz>
25162 * multiple_target.c (create_dispatcher_calls): Redirect also
25165 2018-03-16 Jakub Jelinek <jakub@redhat.com>
25169 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
25170 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
25171 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
25173 2018-03-16 Julia Koval <julia.koval@intel.com>
25175 * doc/invoke.texi (Skylake Server): Add CLWB.
25176 Cannonlake): Remove CLWB.
25178 2018-03-16 Jakub Jelinek <jakub@redhat.com>
25180 PR tree-optimization/84841
25181 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
25183 (FLOAT_ONE_CONST_TYPE): Define.
25184 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
25185 (sort_by_operand_rank): Put entries with higher constant_type last
25186 rather than first to match comments.
25188 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
25190 * config/nios2/nios2.md (movsi_internal): Fix thinko in
25193 2018-03-15 Jakub Jelinek <jakub@redhat.com>
25196 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
25197 check and use address of target always.
25199 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
25202 * config/i386/i386.c (indirect_thunk_needed): Update comments.
25203 (indirect_thunk_bnd_needed): Likewise.
25204 (indirect_thunks_used): Likewise.
25205 (indirect_thunks_bnd_used): Likewise.
25206 (indirect_return_needed): New.
25207 (indirect_return_bnd_needed): Likewise.
25208 (output_indirect_thunk_function): Add a bool argument for
25210 (output_indirect_thunk_function): Don't generate alias for
25211 function return thunk.
25212 (ix86_code_end): Call output_indirect_thunk_function to generate
25213 function return thunks.
25214 (ix86_output_function_return): Set indirect_return_bnd_needed
25215 and indirect_return_needed instead of indirect_thunk_bnd_needed
25216 and indirect_thunk_needed.
25218 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
25220 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
25221 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
25222 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
25224 2018-03-15 David Malcolm <dmalcolm@redhat.com>
25225 Paul Hua <paul.hua.gm@gmail.com>
25228 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
25230 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
25232 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
25233 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
25234 resp. SFmode cases.
25236 2018-03-15 Tamar Christina <tamar.christina@arm.com>
25239 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
25240 instead of GET_MODE_SIZE when comparing Units.
25242 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
25245 * varasm.c (hash_section): Return an unchangeble hash value
25246 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
25247 Return !aarch64_can_use_per_function_literal_pools_p ().
25249 2018-03-15 Jakub Jelinek <jakub@redhat.com>
25252 * optabs.c (emit_conditional_move): Pass address of cmode's copy
25253 rather than address of cmode as last argument to prepare_cmp_insn.
25255 2018-03-15 Julia Koval <julia.koval@intel.com>
25257 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
25258 F_AVX512VNNI, F_AVX512BITALG): New.
25260 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
25263 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
25264 insn for floating-point loads and stores.
25266 2018-03-14 Carl Love <cel@us.ibm.com>
25268 * config/rs6000/rs6000-c.c: Add macro definitions for
25269 ALTIVEC_BUILTIN_VEC_PERMXOR.
25270 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
25271 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
25272 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
25273 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
25275 * config/doc/extend.texi: Add prototypes for vec_permxor.
25277 2018-03-14 David Malcolm <dmalcolm@redhat.com>
25280 * diagnostic-show-locus.c (class layout_point): Convert m_line
25281 from int to linenum_type.
25282 (line_span::comparator): Use linenum "compare" function when
25283 comparing line numbers.
25284 (test_line_span): New function.
25285 (layout_range::contains_point): Convert param "row" from int to
25287 (layout_range::intersects_line_p): Likewise.
25288 (layout::will_show_line_p): Likewise.
25289 (layout::print_source_line): Likewise.
25290 (layout::should_print_annotation_line_p): Likewise.
25291 (layout::print_annotation_line): Likewise.
25292 (layout::print_leading_fixits): Likewise.
25293 (layout::annotation_line_showed_range_p): Likewise.
25294 (struct line_corrections): Likewise for field m_row.
25295 (line_corrections::line_corrections): Likewise for param "row".
25296 (layout::print_trailing_fixits): Likewise.
25297 (layout::get_state_at_point): Likewise.
25298 (layout::get_x_bound_for_row): Likewise.
25299 (layout::print_line): Likewise.
25300 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
25301 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
25302 * input.c (selftest::test_linenum_comparisons): New function.
25303 (selftest::input_c_tests): Call it.
25304 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
25305 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
25306 * selftest.h (ASSERT_GT): New macro.
25307 (ASSERT_GT_AT): New macro.
25308 (ASSERT_LT): New macro.
25309 (ASSERT_LT_AT): New macro.
25311 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
25313 PR rtl-optimization/84780
25314 * combine.c (distribute_links): Don't make a link based on pc_rtx.
25316 2018-03-14 Martin Liska <mliska@suse.cz>
25318 * tree.c (record_node_allocation_statistics): Use
25319 get_stats_node_kind.
25320 (get_stats_node_kind): New function extracted from
25321 record_node_allocation_statistics.
25322 (free_node): Use get_stats_node_kind.
25324 2018-03-14 Richard Biener <rguenther@suse.de>
25326 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
25327 that the value-set of ANTIC_IN doesn't grow.
25330 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
25332 (BB_VISITED_WITH_VISITED_SUCCS): New define.
25333 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
25335 2018-03-14 Julia Koval <julia.koval@intel.com>
25337 * config.gcc (icelake-client, icelake-server): New.
25339 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
25340 (initial_ix86_arch_features): Ditto.
25341 (PTA_SKYLAKE): Add SGX.
25342 (PTA_ICELAKE): Remove.
25343 (PTA_ICELAKE_CLIENT): New.
25344 (PTA_ICELAKE_SERVER): New.
25345 (ix86_option_override_internal): Split up icelake on icelake client and
25347 (get_builtin_code_for_version): Ditto.
25348 (fold_builtin_cpu): Ditto.
25349 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
25350 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
25351 * config/i386/i386.h (processor_type): Ditto.
25352 * doc/invoke.texi: Ditto.
25354 2018-03-14 Jakub Jelinek <jakub@redhat.com>
25357 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
25358 INTEGER_CST offset, add it together with bitpos / 8 and
25359 sign extend based on POINTER_SIZE.
25363 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25366 * config/i386/constraints.md (Yc): New register constraint.
25367 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
25368 Use Yc constraint for alternative 2 of operand 0. Remove
25369 preferred_for_speed attribute.
25371 2018-03-14 Richard Biener <rguenther@suse.de>
25373 PR tree-optimization/84830
25374 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
25375 with the old one to avoid oscillations.
25377 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
25380 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
25382 (assign_by_spills): Return a flag of reload assignment failure.
25383 Do not process the reload assignment failures. Do not spill other
25384 reload pseudos if they has the same reg class. Update n if
25386 (lra_assign): Add a return arg. Set up from the result of
25387 assign_by_spills call.
25388 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
25389 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
25390 usage_insns if it is not NULL.
25391 (spill_hard_reg_in_range): New function.
25392 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
25393 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
25394 function prototypes.
25395 (lra_assign): Change prototype.
25396 * lra.c (lra): Add code to deal with fails by splitting hard reg
25399 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
25401 * config/riscv/riscv.opt (mrelax): New option.
25402 * config/riscv/riscv.c (riscv_file_start): Emit ".option
25403 "norelax" when riscv_mrelax is disabled.
25404 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
25406 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25409 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
25410 reassociation for int modes.
25412 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
25414 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
25415 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
25417 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
25418 * config/aarch64/aarch64-sve.md
25419 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
25420 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
25421 (*extend<mode><Vwide>2): Rename to...
25422 (aarch64_sve_extend<mode><Vwide>2): ...this.
25423 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
25424 renaming the old pattern to...
25425 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
25427 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
25428 define_expand, renaming the old pattern to...
25429 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
25430 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
25431 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
25432 account when deciding which SVE instruction the optab should use.
25433 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
25435 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
25437 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
25438 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
25439 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
25440 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
25441 (tlsdesc_small_<mode>): Turn a define_expand and use
25442 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
25443 (tlsdesc_small_advsimd_<mode>): ...this.
25444 (tlsdesc_small_sve_<mode>): New pattern.
25446 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
25448 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
25449 (UNSPEC_UMUL_HIGHPART): New constants.
25450 (MUL_HIGHPART): New int iteraor.
25451 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
25452 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
25454 (*<su>mul<mode>3_highpart): New define_insn.
25456 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
25459 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
25462 2018-03-13 Martin Liska <mliska@suse.cz>
25465 * (sem_item_optimizer::sem_item_optimizer): Initialize new
25467 (sem_item_optimizer::~sem_item_optimizer): Release it.
25468 (sem_item_optimizer::merge_classes): Register variable aliases.
25469 (sem_item_optimizer::fixup_pt_set): New function.
25470 (sem_item_optimizer::fixup_points_to_sets): Likewise.
25471 * ipa-icf.h: Declare new variables and functions.
25473 2018-03-13 Jakub Jelinek <jakub@redhat.com>
25475 PR middle-end/84834
25476 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
25477 integer_pow2p@2 and test integer_pow2p in condition.
25478 (A < 0 ? C : 0): Similarly for @1.
25480 PR middle-end/84831
25481 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
25482 characters starting at p contain '\0' character, don't look beyond
25486 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
25487 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
25490 * reg-stack.c (change_stack): Change update_end var from int to
25491 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
25492 also call set_block_for_insn on the newly added insns and rescan.
25495 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
25496 on the last operand.
25499 * tree.c (stabilize_reference_1): Return save_expr (e) for
25500 STATEMENT_LIST even if it doesn't have side-effects.
25502 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
25504 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
25506 2018-03-12 Renlin Li <renlin.li@arm.com>
25508 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
25509 aarch64_output_scalar_simd_mov_immediate.
25511 2018-03-12 Martin Sebor <msebor@redhat.com>
25513 PR tree-optimization/83456
25514 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
25515 for perfectly overlapping calls to memcpy.
25516 (gimple_fold_builtin_memory_chk): Same.
25517 (gimple_fold_builtin_strcpy): Handle no-warning.
25518 (gimple_fold_builtin_stxcpy_chk): Same.
25519 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
25521 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
25523 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
25524 parameter. Use it for SFmode.
25525 (rs6000_function_arg_advance_1): Adjust.
25526 (rs6000_function_arg): Adjust.
25527 (rs6000_gimplify_va_arg): Pass false for that new parameter.
25529 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
25531 PR rtl-optimization/84169
25532 PR rtl-optimization/84780
25533 * combine.c (can_combine_p): Check for a 2-insn combination whether
25534 the destination register is used between the two insns, too.
25536 2018-03-12 Richard Biener <rguenther@suse.de>
25538 PR tree-optimization/84803
25539 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
25540 for refs DR analysis didn't process.
25542 2018-03-12 Richard Biener <rguenther@suse.de>
25544 PR tree-optimization/84777
25545 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
25546 force-vectorize loops ignore whether we are optimizing for size.
25548 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
25550 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
25551 (TARGET_MD_ASM_ADJUST): Define.
25553 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
25554 Kito Cheng <kito.cheng@gmail.com>
25555 Chung-Ju Wu <jasonwucj@gmail.com>
25557 * config/nds32/nds32.c (nds32_compute_stack_frame,
25558 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
25559 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
25560 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
25561 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
25562 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
25563 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
25564 * config/nds32/nds32.md (prologue, epilogue): Use macro
25565 NDS32_V3PUSH_AVAILABLE_P to do checking.
25567 2018-03-11 Jakub Jelinek <jakub@redhat.com>
25570 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
25571 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
25572 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
25573 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
25574 addition of most attributes on !orig_type_die or the attribute not
25575 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
25577 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
25578 Chung-Ju Wu <jasonwucj@gmail.com>
25580 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
25581 __NDS32_VH__ macro.
25582 * config/nds32/nds32.opt (mvh): New option.
25584 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
25585 Chung-Ju Wu <jasonwucj@gmail.com>
25587 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
25589 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
25590 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
25593 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
25594 Chung-Ju Wu <jasonwucj@gmail.com>
25596 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
25598 * config/nds32/nds32-multiple.md (strlensi): New pattern.
25599 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
25601 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
25602 Kito Cheng <kito.cheng@gmail.com>
25603 Chung-Ju Wu <jasonwucj@gmail.com>
25605 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
25606 UNSPEC_FFMISM and UNSPEC_FLMISM.
25607 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
25608 for ffb, ffmism and flmism.
25609 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
25610 (unspec_ffmism): Ditto.
25611 (unspec_flmism): Ditto.
25612 (nds32_expand_builtin_impl): Check if string extension is available.
25613 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
25614 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
25616 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
25619 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
25622 * lra-assigns.c (assign_by_spills): Return a flag of reload
25623 assignment failure. Do not process the reload assignment
25624 failures. Do not spill other reload pseudos if they has the same
25626 (lra_assign): Add a return arg. Set up from the result of
25627 assign_by_spills call.
25628 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
25629 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
25630 usage_insns if it is not NULL.
25631 (spill_hard_reg_in_range): New function.
25632 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
25633 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
25634 function prototypes.
25635 (lra_assign): Change prototype.
25636 * lra.c (lra): Add code to deal with fails by splitting hard reg
25639 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
25642 * config/i386/i386.opt: Replace Enforcment with Enforcement.
25644 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
25647 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
25648 (dw_val_node): Add val_symbolic_view.
25649 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
25650 (symview_upper_bound): New.
25651 (new_line_info_table): Initialize symviews_since_reset.
25652 (dwarf2out_source_line): Count symviews_since_reset and set
25653 symview_upper_bound.
25654 (dw_val_equal_p): Handle symview.
25655 (add_AT_symview): New.
25656 (print_dw_val): Handle symview.
25657 (attr_checksum, attr_checksum_ordered): Likewise.
25658 (same_dw_val_p, size_of_die): Likewise.
25659 (value_format, output_die): Likewise.
25660 (add_high_low_attributes): Use add_AT_symview for entry_view.
25661 (dwarf2out_finish): Reset symview_upper_bound, clear
25664 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
25667 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
25668 Add strict argument and use it.
25669 (rs6000_split_multireg_move): Update for new strict argument.
25670 (mem_operand_gpr): Disallow all non-offsettable addresses.
25671 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
25673 2018-03-09 Jakub Jelinek <jakub@redhat.com>
25676 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
25677 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
25678 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
25681 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
25682 decl, use remap_type if we want to use the type.
25684 2018-03-09 Martin Sebor <msebor@redhat.com>
25686 PR tree-optimization/84526
25687 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
25689 (builtin_access::generic_overlap): Be prepared to handle non-array
25692 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
25694 PR rtl-optimization/84682
25695 * lra-constraints.c (process_address_1): Check is_address flag
25696 for address constraints.
25697 (process_alt_operands): Likewise.
25698 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
25699 preprocess_constraints.
25700 * recog.h (preprocess_constraints): Add oploc parameter.
25702 * recog.c (preprocess_constraints): Test address_operand for
25703 CT_ADDRESS constraints.
25705 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
25708 * lra-assigns.c (assign_by_spills): Return a flag of reload
25709 assignment failure. Do not process the reload assignment
25710 failures. Do not spill other reload pseudos if they has the same
25712 (lra_assign): Add a return arg. Set up from the result of
25713 assign_by_spills call.
25714 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
25715 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
25716 usage_insns if it is not NULL.
25717 (spill_hard_reg_in_range): New function.
25718 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
25719 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
25720 function prototypes.
25721 (lra_assign): Change prototype.
25722 * lra.c (lra): Add code to deal with fails by splitting hard reg
25725 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25728 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
25729 Accept complain bool parameter. Only emit errors if it is true.
25730 (arm_parse_cpu_option_name): Likewise.
25731 (arm_target_thumb_only): Adjust callers of the above.
25732 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
25733 prototype to take a default true bool parameter.
25734 (arm_parse_arch_option_name): Likewise.
25736 2018-03-09 David Malcolm <dmalcolm@redhat.com>
25737 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
25741 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
25742 * configure: Regenerate.
25743 * configure.ac ("linker --version-script option"): New.
25744 ("linker soname option"): New.
25746 2018-03-09 Richard Biener <rguenther@suse.de>
25748 PR tree-optimization/84775
25749 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
25750 immediate uses of predicate stmts and mark them modified.
25753 PR tree-optimization/84178
25754 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
25756 (version_loop_for_if_conversion): Delay update_ssa call.
25757 (tree_if_conversion): Delay update_ssa until after predicate
25760 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
25763 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
25764 when the function accesses prior frames.
25766 2018-03-08 Jakub Jelinek <jakub@redhat.com>
25769 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
25770 gen_llsym, otherwise call maybe_gen_llsym.
25772 PR inline-asm/84742
25773 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
25774 has ',' character inside of it.
25776 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25779 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
25780 as clobbering CC_REGNUM.
25782 2018-03-08 Richard Biener <rguenther@suse.de>
25784 PR middle-end/84552
25785 * tree-scalar-evolution.c: Include tree-into-ssa.h.
25786 (follow_copies_to_constant): Do not follow SSA names registered
25789 2018-03-08 Richard Biener <rguenther@suse.de>
25791 PR tree-optimization/84178
25792 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
25794 (version_loop_for_if_conversion): Delay update_ssa call.
25795 (tree_if_conversion): Delay update_ssa until after predicate
25798 2018-03-08 David Malcolm <dmalcolm@redhat.com>
25800 PR tree-optimization/84178
25801 * tree-if-conv.c (release_bb_predicate): Remove the
25802 the assertion that the stmts have NULL use_ops.
25803 Discard the statements, asserting that they haven't
25804 yet been added to a BB.
25806 2018-03-08 Richard Biener <rguenther@suse.de>
25808 PR tree-optimization/84746
25809 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
25810 (phi_translate): Pass in destination ANTIC_OUT set.
25811 (phi_translate_1): Likewise. For a simplified result lookup
25812 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
25813 (phi_translate_set): Adjust.
25814 (do_pre_regular_insertion): Likewise.
25815 (do_pre_partial_partial_insertion): Likewise.
25817 2018-03-08 Martin Liska <mliska@suse.cz>
25819 PR gcov-profile/84735
25820 * doc/gcov.texi: Document usage of profile files.
25821 * gcov-io.h: Document changes in the format.
25823 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
25827 * dwarf2out.c (struct dw_line_info_table): Update comments for
25829 (FORCE_RESET_NEXT_VIEW): New.
25830 (FORCE_RESETTING_VIEW_P): New.
25831 (RESETTING_VIEW_P): Check for -1 too.
25832 (ZERO_VIEW_P): Likewise.
25833 (new_line_info_table): Force-reset next view.
25834 (dwarf2out_begin_function): Likewise.
25835 (dwarf2out_source_line): Simplify zero_view_p initialization.
25836 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
25837 view directly. Omit view when omitting .loc at line 0.
25839 2018-03-08 Jakub Jelinek <jakub@redhat.com>
25841 PR tree-optimization/84740
25842 * tree-switch-conversion.c (process_switch): Call build_constructors
25843 only if info.phi_count is non-zero.
25845 PR tree-optimization/84739
25846 * tree-tailcall.c (find_tail_calls): Check call arguments against
25847 DECL_ARGUMENTS (current_function_decl) rather than
25848 DECL_ARGUMENTS (func) when checking for tail recursion.
25850 2018-03-07 Jakub Jelinek <jakub@redhat.com>
25852 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
25853 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
25854 Volker Reichelt's entry and add entries for people that perform
25855 GCC fuzzy testing and report numerous bugs.
25857 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
25860 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
25861 readonly data in sdata, if that is disabled.
25862 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
25863 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
25864 -mreadonly-in-sdata option.
25866 2018-03-07 Martin Sebor <msebor@redhat.com>
25868 PR tree-optimization/84468
25869 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
25870 basic block when looking for nul assignment.
25872 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
25875 * except.h (output_function_exception_table): Adjust prototype.
25876 * except.c (output_function_exception_table): Remove FNNAME parameter
25877 and add SECTION parameter. Ouput one part of the table at a time.
25878 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
25879 the first part of the exception table and emit unwind directives.
25880 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
25881 (i386_pe_seh_cold_init): Likewise.
25882 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
25883 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
25884 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
25885 (ix86_output_call_insn): Emit a nop in one more case for SEH.
25886 * config/i386/winnt.c: Include except.h.
25887 (struct seh_frame_state): Add reg_offset, after_prologue and
25888 in_cold_section fields.
25889 (i386_pe_seh_end_prologue): Set seh->after_prologue.
25890 (i386_pe_seh_cold_init): New function.
25891 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
25892 to seh->in_cold_section.
25893 (seh_emit_push): Record the offset of the push.
25894 (seh_emit_save): Record the offet of the save.
25895 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
25896 Test seh->after_prologue to disregard the epilogue.
25897 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
25898 (i386_pe_end_cold_function): New function.
25900 2018-03-07 Jakub Jelinek <jakub@redhat.com>
25903 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
25904 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
25907 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
25909 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
25910 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
25912 PR middle-end/84723
25913 * multiple_target.c: Include tree-inline.h and intl.h.
25914 (expand_target_clones): Diagnose and fail if node->definition and
25915 !tree_versionable_function_p (node->decl).
25917 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
25919 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
25921 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
25922 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
25923 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
25925 2018-03-06 Jakub Jelinek <jakub@redhat.com>
25928 * combine.c (try_combine): Use reg_or_subregno instead of handling
25929 just paradoxical SUBREGs and REGs.
25931 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
25933 * config/arc/arc.c (arc_finalize_pic): Remove function.
25934 (arc_must_save_register): We use single base PIC register, remove
25935 checks to save/restore the PIC register.
25936 (arc_expand_prologue): Likewise.
25937 * config/arc/arc-protos.h (arc_set_default_type_attributes):
25939 (arc_verify_short): Likewise.
25940 (arc_attr_type): Likewise.
25941 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
25942 (walk_stores): Likewise.
25943 (arc_address_cost): Make it static.
25944 (arc_verify_short): Likewise.
25945 (branch_dest): Likewise.
25946 (arc_attr_type): Likewise.
25947 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
25948 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
25949 (arc_final_prescan_insn): Remove inserting the nops due to
25950 hardware hazards. It is done in reorg step.
25951 (insn_length_variant_t): Remove.
25952 (insn_length_parameters_t): Likewise.
25953 (arc_insn_length_parameters): Likewise.
25954 (arc_get_insn_variants): Likewise.
25955 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
25957 2018-03-06 Jakub Jelinek <jakub@redhat.com>
25959 PR inline-asm/84683
25960 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
25963 PR tree-optimization/84687
25964 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
25966 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
25968 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25970 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
25971 Rename to ppc_speculation_barrier.
25972 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
25973 __builtin_ppc_speculation_barrier.
25975 2018-03-05 Jakub Jelinek <jakub@redhat.com>
25978 * combine.c (combine_simplify_rtx): Don't try to simplify if
25979 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
25982 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
25984 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
25985 to 32 bytes when compiling for POWER9.
25987 2018-03-05 Jakub Jelinek <jakub@redhat.com>
25990 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
25991 regparm >= 3 with no arg reg available also for calls with
25992 flag_force_indirect_call. Pass decl to ix86_function_regparm.
25995 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
25997 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
25999 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
26002 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
26004 2018-03-05 Richard Biener <rguenther@suse.de>
26006 PR tree-optimization/84486
26007 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
26008 When inserting a __builtin_assume_aligned call set the LHS
26009 SSA name alignment info accordingly.
26011 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
26013 PR tree-optimization/84114
26014 * config/aarch64/aarch64.c (aarch64_reassociation_width)
26015 Avoid reassociation of FLOAT_MODE addition.
26017 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
26019 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
26020 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
26021 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
26022 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
26023 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
26024 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
26025 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
26027 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
26028 __builtin_ia32_wbinvd): New builtins.
26029 (SPECIAL_ARGS2): New.
26030 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
26031 (SPECIAL_ARGS2): New.
26032 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
26033 (ix86_valid_target_attribute_inner_p): Ditto.
26034 (ix86_init_mmx_sse_builtins): Add special_args2.
26035 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
26036 TARGET_WBNOINVD_P): New.
26037 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
26038 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
26039 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
26040 * config/i386/immintrin.h (_wbinvd): New intrinsic.
26041 * config/i386/pconfigintrin.h: New file.
26042 * config/i386/wbnoinvdintrin.h: Ditto.
26043 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
26045 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
26047 2018-03-05 Richard Biener <rguenther@suse.de>
26049 PR tree-optimization/84670
26050 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
26052 (BB_VISITED_WITH_VISITED_SUCCS): New define.
26053 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
26054 (compute_antic_aux): Only assert the number of values in ANTIC_IN
26055 doesn't grow if all successors (recursively) were visited at least
26058 2018-03-05 Richard Biener <rguenther@suse.de>
26060 PR tree-optimization/84650
26061 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
26062 if executed in the loop pipeline.
26064 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
26066 * doc/configfiles.texi (Configuration Files): Move info about
26067 conditionalizing $target-protos.h to...
26068 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
26069 differs from $target-protos.h.
26071 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
26072 Chung-Ju Wu <jasonwucj@gmail.com>
26074 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
26075 * config/nds32/nds32-multiple.md (setmemsi): Define.
26076 * config/nds32/nds32-memory-manipulation.c
26077 (nds32_gen_dup_4_byte_to_word_value): New.
26078 (emit_setmem_word_loop): New.
26079 (emit_setmem_byte_loop): New.
26080 (nds32_expand_setmem_loop): New.
26081 (nds32_expand_setmem_loop_v3m): New.
26082 (nds32_expand_setmem_unroll): New.
26083 (nds32_expand_setmem): New.
26085 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
26086 Chung-Ju Wu <jasonwucj@gmail.com>
26088 * config/nds32/nds32-memory-manipulation.c
26089 (nds32_emit_load_store): New.
26090 (nds32_emit_post_inc_load_store): New.
26091 (nds32_emit_mem_move): New.
26092 (nds32_emit_mem_move_block): New.
26093 (nds32_expand_movmemsi_loop_unknown_size): New.
26094 (nds32_expand_movmemsi_loop_known_size): New.
26095 (nds32_expand_movmemsi_loop): New.
26096 (nds32_expand_movmemsi_unroll): New.
26097 (nds32_expand_movmemqi): Rename ...
26098 (nds32_expand_movmemsi): ... to this.
26099 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
26100 (movmemsi): ... to this.
26101 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
26102 (nds32_expand_movmemsi): ... to this.
26104 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
26105 Monk Chiang <sh.chiang04@gmail.com>
26106 Chung-Ju Wu <jasonwucj@gmail.com>
26108 * config/nds32/nds32-protos.h
26109 (nds32_expand_load_multiple): New arguments.
26110 (nds32_expand_store_multiple): Ditto.
26111 (nds32_valid_multiple_load_store): Rename ...
26112 (nds32_valid_multiple_load_store_p): ... to this.
26113 * config/nds32/nds32-memory-manipulation.c
26114 (nds32_expand_load_multiple): Refine implementation.
26115 (nds32_expand_store_multiple): Ditto.
26116 * config/nds32/nds32-multiple.md
26117 (load_multiple): Update nds32_expand_load_multiple interface.
26118 (store_multiple): Update nds32_expand_store_multiple interface.
26119 * config/nds32/nds32-predicates.c
26120 (nds32_valid_multiple_load_store): Rename ...
26121 (nds32_valid_multiple_load_store_p): ... to this and refine
26123 * config/nds32/predicates.md
26124 (nds32_load_multiple_and_update_address_operation): New predicate.
26125 (nds32_store_multiple_and_update_address_operation): New predicate.
26127 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
26128 Chung-Ju Wu <jasonwucj@gmail.com>
26130 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
26131 (combo): New attribute.
26132 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
26134 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
26136 * config/nds32/nds32.opt: Change -mcmodel= default value.
26138 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
26139 Monk Chiang <sh.chiang04@gmail.com>
26140 Chung-Ju Wu <jasonwucj@gmail.com>
26142 * config/nds32/constants.md (unspec_element): New enum.
26143 * config/nds32/constraints.md (Umw): New constraint.
26144 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
26145 * config/nds32/nds32-intrinsic.md: Likewise.
26146 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
26147 (nds32_valid_smw_lwm_base_p): New.
26148 (nds32_output_smw_single_word): New.
26149 (nds32_output_lmw_single_word): New.
26150 (nds32_expand_unaligned_load): New.
26151 (nds32_expand_unaligned_store): New.
26152 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
26153 (nds32_output_smw_single_word): Declare.
26154 (nds32_output_lmw_single_word): Declare.
26155 (nds32_expand_unaligned_load): Declare.
26156 (nds32_expand_unaligned_store): Declare.
26157 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
26158 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
26159 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
26160 NDS32_BUILTIN_UASTORE_DW.
26161 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
26164 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
26165 Kito Cheng <kito.cheng@gmail.com>
26166 Chung-Ju Wu <jasonwucj@gmail.com>
26168 * config/nds32/nds32-intrinsic.c
26169 (nds32_expand_builtin_null_ftype_reg): Delete.
26170 (nds32_expand_builtin_reg_ftype_imm): Ditto.
26171 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
26172 (nds32_read_argument): New.
26173 (nds32_legitimize_target): Ditto.
26174 (nds32_legitimize_argument): Ditto.
26175 (nds32_check_constant_argument): Ditto.
26176 (nds32_expand_unop_builtin): Ditto.
26177 (nds32_expand_unopimm_builtin): Ditto.
26178 (nds32_expand_binop_builtin): Ditto.
26179 (nds32_builtin_decl_impl): Ditto.
26180 (builtin_description): Ditto.
26181 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
26182 (nds32_init_builtins_impl): Ditto.
26183 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
26184 (nds32_builtin_decl): New.
26185 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
26186 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
26188 2018-03-02 Jeff Law <law@redhat.com>
26190 * reorg.c (stop_search_p): Handle DEBUG_INSN.
26191 (redundant_insn, fill_simple_delay_slots): Likewise.
26192 (fill_slots_from_thread): Likewise.
26193 * resource.c (mark_referenced_resources): Likewise.
26194 (mark_set_resources, find_dead_or_set_registers): Likewise.
26196 2018-03-02 Jakub Jelinek <jakub@redhat.com>
26198 * substring-locations.h (format_warning_va): Formatting fix for
26199 ATTRIBUTE_GCC_DIAG.
26200 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
26202 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
26203 * substring-locations.c: Include intl.h.
26204 (format_warning_va): Turned into small wrapper around
26205 format_warning_n_va, renamed to ...
26206 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
26207 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
26209 (format_warning_at_substring_n): New function.
26210 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
26211 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
26212 format_warning_at_substring with just a shorter name instead of
26213 const function pointer.
26214 (fmtwarn_n): New function.
26215 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
26216 appropriate, get rid of all the fmtstr temporaries, move conditionals
26217 with G_() wrapped string literals directly into fmtwarn arguments,
26218 cast dir.len to (int), formatting fixes.
26220 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
26222 * doc/invoke.texi: Remove "Cilk Plus" references.
26224 2018-03-02 Jakub Jelinek <jakub@redhat.com>
26225 Richard Biener <rguenther@suse.de>
26228 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
26229 for error or warning attributes if CALL_FROM_THUNK_P is set.
26232 2018-03-02 Jakub Jelinek <jakub@redhat.com>
26235 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
26236 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
26239 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
26240 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
26242 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
26243 instead of -1U in last predictors element's probability member.
26245 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
26248 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
26249 arguments if they are comparable.
26251 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
26253 PR tree-optimization/84634
26254 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
26255 masks and masked_loop_p with a single loop_masks, making sure it's
26256 null for bb vectorization.
26258 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
26260 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
26261 (vect_analyze_data_ref_access): Use loop->safe_len rather than
26262 loop->force_vectorize to check whether there is no alias.
26264 2018-03-02 Jakub Jelinek <jakub@redhat.com>
26267 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
26269 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
26271 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
26272 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
26273 instead of a loop around prev_real_insn.
26274 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
26277 PR inline-asm/84625
26278 * config/i386/i386.c (ix86_print_operand): Use conditional
26279 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
26282 2018-03-02 Richard Biener <rguenther@suse.de>
26284 PR tree-optimization/84427
26285 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
26286 (bitmap_set_subtract_values): Rewrite to handle multiple
26289 (prune_clobbered_mems): Likewise.
26290 (phi_translate): Take edge instead of pred/phiblock.
26291 (phi_translate_1): Likewise.
26292 (phi_translate_set): Likewise. Insert all translated
26293 exprs for a value into the set, keeping possibly multiple
26294 expressions per value.
26295 (compute_antic_aux): Adjust for phi_translate changes.
26296 When intersecting union the expressions and prune those
26297 not in the final value set, keeping possibly multiple
26298 expressions per value. Do not use value-insertion
26299 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
26300 all expressions. Add verification that the value-sets
26301 only shrink during iteration.
26302 (compute_partial_antic_aux): Adjust for the phi_translate changes.
26303 (do_pre_regular_insertion): Likewise.
26304 (do_pre_partial_partial_insertion): Likewise.
26306 2018-03-02 Richard Biener <rguenther@suse.de>
26309 * config/darwin.c (saved_debug_info_level): New static global.
26310 (darwin_asm_lto_start): Disable debug info generation for LTO out.
26311 (darwin_asm_lto_end): Restore debug info generation settings.
26313 2018-03-01 Martin Liska <mliska@suse.cz>
26316 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
26317 volatile arguments.
26319 2018-03-01 Richard Biener <rguenther@suse.de>
26322 * dwarf2out.c (gen_variable_die): Properly handle late VLA
26323 type annotation with LTO when debug was disabled at compile-time.
26325 2018-03-01 Matthew Fortune <mfortune@gmail.com>
26327 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
26329 (mips_final_postscan_insn): Likewise.
26331 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
26333 PR rtl-optimization/84528
26334 * alias.c (init_alias_target): Add commentary.
26335 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
26336 a unique base value if the frame pointer is not eliminated
26337 to the stack pointer.
26339 2018-03-01 Tom de Vries <tom@codesourcery.com>
26341 PR rtl-optimization/83327
26342 * lra-int.h (hard_regs_spilled_into): Declare.
26343 * lra.c (hard_regs_spilled_into): Define.
26344 (init_reg_info): Init hard_regs_spilled_into.
26345 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
26346 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
26347 (process_bb_lives): Handle hard_regs_spilled_into.
26348 (lra_create_live_ranges_1): Before doing liveness propagation, clear
26349 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
26351 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
26353 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
26354 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
26355 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
26356 * config/rs6000/aix72.h: New file.
26358 2018-02-28 Jakub Jelinek <jakub@redhat.com>
26360 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
26361 instead of warning_at with conditional singular and plural messages
26365 * stor-layout.c (update_alignment_for_field): For
26366 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
26367 && !DECL_PACKED (field), do the alignment update, just use
26368 only desired_align instead of MAX (type_align, desired_align)
26370 (place_field): Don't do known_align < desired_align handling
26371 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
26372 is non-NULL, instead do it after rli->prev_field handling and
26373 only if not within a bitfield word. For DECL_PACKED (field)
26374 use type_align of BITS_PER_UNIT.
26376 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
26378 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
26379 superfluous parentheses and trailing spaces.
26381 2018-02-28 Richard Biener <rguenther@suse.de>
26383 PR tree-optimization/84584
26384 * graphite-scop-detection.c (scop_detection::add_scop): Discard
26385 SCoPs with fake exit edge.
26387 2018-02-28 Martin Liska <mliska@suse.cz>
26390 * timevar.c (timer::print): Fix format to properly print 100%
26393 2018-02-28 Richard Biener <rguenther@suse.de>
26395 PR middle-end/84607
26396 * genmatch.c (capture_info::walk_match): Do not mark
26397 captured expressions without operands as expr_p given
26398 they act more like predicates and should be subject to
26399 "lost tail" side-effect preserving.
26401 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
26403 PR rtl-optimization/81611
26404 * auto-inc-dec.c (attempt_change): Move dead note from
26405 mem_insn if it's the next use of regno
26406 (find_address): Take address use of reg holding
26407 non-incremented value. Add parm to limit search to the named
26409 (merge_in_block): Attempt to use a mem insn that is the next
26410 use of the original regno.
26412 2018-02-27 Martin Sebor <msebor@redhat.com>
26415 * doc/invoke.texi (-Wmissing-attributes): New option.
26416 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
26418 2018-02-27 Martin Sebor <msebor@redhat.com>
26420 PR translation/84207
26421 * diagnostic-core.h (warning_n, error_n, inform_n): Change
26422 n argument to unsigned HOST_WIDE_INT.
26423 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
26424 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
26425 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
26426 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
26428 2018-02-27 Richard Biener <rguenther@suse.de>
26430 PR tree-optimization/84512
26431 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
26432 Do not use the estimate returned from record_stmt_cost for
26433 the scalar iteration cost but sum properly using add_stmt_cost.
26435 2018-02-27 Richard Biener <rguenther@suse.de>
26437 PR tree-optimization/84466
26438 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
26439 Adjust last change to less strictly validate use operands.
26441 2018-02-27 Martin Liska <mliska@suse.cz>
26443 PR gcov-profile/84548
26444 * gcov.c (process_file): Allow partial overlap and consider it
26445 also as group functions.
26446 (output_lines): Properly calculate range of lines for a group.
26448 2018-02-27 Martin Liska <mliska@suse.cz>
26450 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
26451 'ggc' suffixes. Change first column width.
26452 (timer::print): Fix formatting of the column.
26454 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
26456 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
26457 preserve inline entry blocks for the sake of debug inline
26458 entry point markers alone.
26459 (remove_unused_locals): Suggest in comments a better place to
26460 force the preservation of inline entry blocks that are
26461 otherwise unused, but do not preserve them.
26463 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
26465 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
26467 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
26470 * config/i386/constraints.md (Bs): Replace
26471 ix86_indirect_branch_register with
26472 TARGET_INDIRECT_BRANCH_REGISTER.
26474 * config/i386/i386.md (indirect_jump): Likewise.
26475 (tablejump): Likewise.
26476 (*sibcall_memory): Likewise.
26477 (*sibcall_value_memory): Likewise.
26478 Peepholes of indirect call and jump via memory: Likewise.
26479 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
26480 (*sibcall_value_GOT_32): Likewise.
26481 * config/i386/predicates.md (indirect_branch_operand): Likewise.
26482 (GOT_memory_operand): Likewise.
26483 (call_insn_operand): Likewise.
26484 (sibcall_insn_operand): Likewise.
26485 (GOT32_symbol_operand): Likewise.
26486 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
26488 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
26490 PR rtl-optimization/83496
26491 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
26492 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
26493 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
26494 redundant insn, if any.
26495 (relax_delay_slots): Likewise.
26496 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
26498 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
26500 PR tree-optimization/83965
26501 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
26502 that grouped statements are part of a reduction chain. Return
26503 true if the statement is not marked as a reduction itself but
26504 is part of a group.
26505 (vect_recog_dot_prod_pattern): Don't check whether the statement
26506 is part of a group here.
26507 (vect_recog_sad_pattern): Likewise.
26508 (vect_recog_widen_sum_pattern): Likewise.
26510 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
26513 * final.c (rest_of_clean_state): Also look for calls inside sequences.
26515 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
26518 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
26520 (ix86_output_indirect_function_return): New prototype.
26521 (ix86_split_simple_return_pop_internal): Likewise.
26522 * config/i386/i386.c (indirect_return_via_cx): New.
26523 (indirect_return_via_cx_bnd): Likewise.
26524 (indirect_thunk_name): Handle return va CX_REG.
26525 (output_indirect_thunk_function): Create alias for
26526 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
26527 (ix86_output_indirect_jmp): Remove the bool argument.
26528 (ix86_output_indirect_function_return): New function.
26529 (ix86_split_simple_return_pop_internal): Likewise.
26530 * config/i386/i386.md (*indirect_jump): Don't pass false
26531 to ix86_output_indirect_jmp.
26532 (*tablejump_1): Likewise.
26533 (simple_return_pop_internal): Change it to define_insn_and_split.
26534 Call ix86_split_simple_return_pop_internal to split it for
26535 -mfunction-return=.
26536 (simple_return_indirect_internal): Call
26537 ix86_output_indirect_function_return instead of
26538 ix86_output_indirect_jmp.
26540 2018-02-26 Jakub Jelinek <jakub@redhat.com>
26543 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
26544 memset and value initialization afterwards.
26546 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
26548 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
26550 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26553 * common/config/aarch64/aarch64-common.c
26554 (aarch_option_optimization_table[]): Switch
26555 off fomit-frame-pointer
26557 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
26558 Chung-Ju Wu <jasonwucj@gmail.com>
26560 * config/nds32/nds32-multiple.md (load_multiple): Disallow
26562 (store_multiple): Ditto.
26564 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
26566 * config.gcc: Add --with-cpu support for nds32 target.
26567 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
26568 * config/nds32/nds32.opt: Add -mcpu= option.
26570 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
26572 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
26573 isel=yes): Warn for these deprecated options.
26575 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
26577 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
26578 ISA_2_5_MASKS_EMBEDDED.
26580 2018-02-23 Jakub Jelinek <jakub@redhat.com>
26582 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
26583 p->max as pointers rather than using iterative_hash_expr.
26585 2018-02-23 Carl Love <cel@us.ibm.com>
26587 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
26588 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
26590 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
26591 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
26592 P8V_BUILTIN_VEC_VUNSIGNED2.
26594 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
26597 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
26598 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
26600 * lra-constraints.c (curr_insn_transform): Set up
26601 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
26603 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
26604 * lra-eliminations.c (spill_pseudos): Ditto.
26605 (process_insn_for_elimination): Ditto.
26606 * lra-lives.c (reg_early_clobber_p): Use the new macros.
26607 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
26608 LRA_NON_CLOBBERED_ALT.
26610 2018-02-22 Martin Sebor <msebor@redhat.com>
26612 PR tree-optimization/84480
26613 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
26614 to maybe_diag_stxncpy_trunc. Call it.
26615 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
26616 from gimple_fold_builtin_strcpy. Print inlining stack.
26617 (handle_builtin_stxncpy): Print inlining stack.
26618 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
26620 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
26623 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
26624 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
26625 and -fcheck-pointer-bounds are used together.
26626 (indirect_thunk_prefix): New enum.
26627 (indirect_thunk_need_prefix): New function.
26628 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
26629 "_nt" instead of "_bnd" for NOTRACK prefix.
26630 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
26631 (output_indirect_thunk_function): Likewise.
26633 (ix86_code_end): Update output_indirect_thunk_function calls.
26634 (ix86_output_indirect_branch_via_reg): Replace
26635 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
26636 (ix86_output_indirect_branch_via_push): Likewise.
26637 (ix86_output_function_return): Likewise.
26638 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
26639 incompatible with -fcf-protection=branch and
26640 -fcheck-pointer-bounds.
26642 2018-02-22 Steve Ellcey <sellcey@cavium.com>
26645 * config/aarch64/aarch64.c (aarch64_print_address_internal):
26646 Change gcc_assert call to output_operand_lossage.
26648 2018-02-22 Steve Ellcey <sellcey@cavium.com>
26650 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
26652 2018-02-22 DJ Delorie <dj@redhat.com>
26653 Sebastian Perta <sebastian.perta@renesas.com>
26654 Oleg Endo <olegendo@gcc.gnu.org>
26656 * config/rx/rx.c (rx_rtx_costs): New function.
26657 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
26659 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
26661 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
26663 2018-02-22 Martin Liska <mliska@suse.cz>
26666 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
26667 Add "native" as a possible value.
26669 2018-02-22 Martin Liska <mliska@suse.cz>
26672 * config/i386/i386.c (ix86_option_override_internal):
26673 Add "native" as a possible value for -march and -mtune.
26675 2018-02-22 Jakub Jelinek <jakub@redhat.com>
26678 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
26679 to all type variants.
26681 PR tree-optimization/84503
26682 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
26683 width as info->bitpos + info->bitsize - start.
26684 (merged_store_group::merge_overlapping): Simplify width computation.
26685 (check_no_overlap): New function.
26686 (imm_store_chain_info::try_coalesce_bswap): Compute expected
26687 start + width and last_order of the group, fail if check_no_overlap
26689 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
26690 to group if check_no_overlap fails.
26692 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
26694 * config/rs6000/altivec.md: Delete contraint arguments to
26695 define_expand, define_split, and define_peephole2, and in
26696 define_insn_and_split if always unused.
26697 * config/rs6000/darwin.md: Ditto.
26698 * config/rs6000/dfp.md: Ditto.
26699 * config/rs6000/rs6000.md: Ditto.
26700 * config/rs6000/sync.md: Ditto.
26701 * config/rs6000/vector.md: Ditto.
26702 * config/rs6000/vsx.md: Ditto.
26704 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
26706 * config/rs6000/altivec.md: Write output control strings as braced
26707 blocks instead of double-quoted strings.
26708 * config/rs6000/darwin.md: Ditto.
26709 * config/rs6000/rs6000.md: Ditto.
26710 * config/rs6000/vector.md: Ditto.
26711 * config/rs6000/vsx.md: Ditto.
26713 2018-02-21 Jason Merrill <jason@redhat.com>
26715 PR c++/84314 - ICE with templates and fastcall attribute.
26716 * attribs.c (build_type_attribute_qual_variant): Remove assert.
26718 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
26720 * ipa-cp.c (determine_versionability): Fix comment typos.
26722 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
26725 * ipa-cp.c (determine_versionability): Do not version functions caling
26728 2018-02-21 Martin Liska <mliska@suse.cz>
26731 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
26732 Add "native" as a possible value.
26733 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
26734 the macro when native cpu detection is available.
26736 2018-02-21 Martin Liska <mliska@suse.cz>
26739 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
26740 Add "native" as a possible value.
26741 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
26742 when native cpu detection is available.
26744 2018-02-21 Jakub Jelinek <jakub@redhat.com>
26745 Martin Sebor <msebor@redhat.com>
26747 PR tree-optimization/84478
26748 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
26750 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
26751 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
26752 support which is conservatively correct, for 2 only stay conservative
26753 for maxlen. Formatting and comment capitalization fixes. Add STRICT
26754 argument to the 2 argument get_range_strlen, adjust 6 arg
26755 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
26757 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
26758 (gimple_fold_builtin_strlen): Pass true as last argument to
26761 2018-02-20 Martin Sebor <msebor@redhat.com>
26763 PR middle-end/84095
26764 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
26765 (builtin_memref::set_base_and_offset): Same. Handle inner references.
26766 (builtin_memref::builtin_memref): Factor out parts into
26767 set_base_and_offset and call it.
26769 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
26771 PR middle-end/84406
26772 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
26773 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
26774 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
26775 search at the associated MODE_INT.
26777 2018-02-20 Jeff Law <law@redhat.com>
26779 PR middle-end/82123
26780 PR tree-optimization/81592
26781 PR middle-end/79257
26782 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
26783 for range data rather than using global data.
26784 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
26785 range data rather than using global data.
26786 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
26787 pass it to children as needed.
26788 (struct directive::fmtresult): Similarly.
26789 (struct directive::set_width): Similarly.
26790 (struct directive::set_precision): Similarly.
26791 (format_integer, format_directive, parse_directive): Similarly.
26792 (format_none): Accept unnamed vr_values parameter.
26793 (format_percent, format_floating, format_character): Similarly.
26794 (format_string, format_plain): Similarly.
26795 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
26796 the EVRP range analyzer for range data rather than using global data.
26797 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
26798 gimple-ssa-evrp-analyze.h
26799 (class sprintf_dom_walker): Add after_dom_children member function.
26800 Add evrp_range_analyzer member.
26801 (sprintf_dom_walker::before_dom_children): Call into the EVRP
26802 range analyzer as needed.
26803 (sprintf_dom_walker::after_dom_children): New member function.
26804 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
26806 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
26807 (evrp_range_analyzer::pop_to_marker): Likewise.
26809 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
26811 PR tree-optimization/84419
26812 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
26813 with the required type if its current type is compatible but
26816 2018-02-20 Jakub Jelinek <jakub@redhat.com>
26818 PR middle-end/82004
26819 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
26820 after vectorization.
26822 2018-02-20 Martin Liska <mliska@suse.cz>
26825 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
26826 possible values if we don't have a hint.
26828 2018-02-20 Martin Liska <mliska@suse.cz>
26832 * final.c (shorten_branches): Build align_tab array with one
26834 * opts.c (finish_options): Add alignment option limit check.
26835 (MAX_CODE_ALIGN): Likewise.
26836 (MAX_CODE_ALIGN_VALUE): Likewise.
26837 * doc/invoke.texi: Document maximum allowed option value for
26838 all -falign-* options.
26840 2018-02-19 Jakub Jelinek <jakub@redhat.com>
26843 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
26844 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
26845 * var-tracking.c (emit_note_insn_var_location): Remove all references
26846 to NOTE_INSN_CALL_ARG_LOCATION.
26847 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
26848 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
26849 Use copy_rtx_if_shared.
26850 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
26851 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
26852 (dwarf2out_var_location): Remove handling of
26853 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
26855 * final.c (final_scan_insn): Remove all references to
26856 NOTE_INSN_CALL_ARG_LOCATION.
26857 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
26858 before dumping final insns.
26859 * except.c (emit_note_eh_region_end): Remove all references to
26860 NOTE_INSN_CALL_ARG_LOCATION.
26861 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
26862 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
26863 * config/arc/arc.c (hwloop_optimize): Likewise.
26864 * config/arm/arm.c (create_fix_barrier): Likewise.
26865 * config/s390/s390.c (s390_chunkify_start): Likewise.
26866 * config/sh/sh.c (find_barrier): Likewise.
26867 * config/i386/i386.c (rest_of_insert_endbranch,
26868 ix86_seh_fixup_eh_fallthru): Likewise.
26869 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
26870 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
26871 * config/frv/frv.c (frv_function_prologue): Likewise.
26872 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
26874 (note_outside_basic_block_p): Remove all references to
26875 NOTE_INSN_CALL_ARG_LOCATION.
26876 * gengtype.c (adjust_field_rtx_def): Likewise.
26877 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
26879 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
26880 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
26883 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
26886 PR tree-optimization/84452
26887 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
26888 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
26891 2018-02-19 Martin Liska <mliska@suse.cz>
26894 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
26896 2018-02-19 Martin Liska <mliska@suse.cz>
26897 Richard Sandiford <richard.sandiford@linaro.org>
26899 PR tree-optimization/82491
26900 * gimple-fold.c (get_base_constructor): Make earlier bail out
26903 2018-02-19 Carl Love <cel@us.ibm.com>
26905 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
26906 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
26908 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
26909 P8V_BUILTIN_VEC_NEG.
26911 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
26913 * config/rl78/rl78.md (movdf): New define expand.
26915 2018-02-19 Martin Liska <mliska@suse.cz>
26918 * doc/invoke.texi: Fix typo.
26919 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
26921 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
26923 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
26924 handle rs6000_single_float and rs6000_double_float specially for
26927 2018-02-16 Jeff Law <law@redhat.com>
26929 * config/rx/rx.c (add_pop_cfi_notes): New function.;
26930 (pop_regs): Use it.
26932 2018-02-16 Jakub Jelinek <jakub@redhat.com>
26935 * ipa-inline.c (inline_small_functions): Fix a typo.
26937 2018-02-16 Nathan Sidwell <nathan@acm.org>
26939 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
26941 2018-02-16 Carl Love <cel@us.ibm.com>
26943 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
26944 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
26945 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
26946 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
26947 expansion to P8V_BUILTIN_VEC_FLOAT2.
26949 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
26951 PR rtl-optimization/70023
26952 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
26953 src_regno into account.
26955 2018-02-16 Carl Love <cel@us.ibm.com>
26957 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
26958 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
26959 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
26960 * config/rs6000/rs6000.c: Remove case statements for
26961 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
26962 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
26963 and P9V_BUILTIN_VEC_VINSERT4B.
26964 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
26965 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
26966 * config/rs6000/vsx.md:
26967 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
26970 2018-02-16 Carl Love <cel@us.ibm.com>
26972 * config/rs6000/altivec.h: Add builtin names vec_extract4b
26974 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
26976 * config/rs6000/rs6000-c.c: Add the definitions for
26977 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
26978 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
26979 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
26980 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
26981 definition for insert4b and define insn *insert3b_internal.
26982 * doc/extend.texi: Add documentation for vec_extract4b.
26984 2018-02-16 Nathan Sidwell <nathan@acm.org>
26986 * doc/extend.texi (Backwards Compatibility): Mention friend
26987 injection. Note for-scope is deprecated.
26988 * doc/invoke.texi (-ffriend-injection): Deprecate.
26990 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
26992 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
26993 that moved to I2, also allow destinations that are a paradoxical
26994 subreg (instead of a normal reg).
26996 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
26999 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
27002 2018-02-16 Richard Biener <rguenther@suse.de>
27004 PR tree-optimization/84037
27005 PR tree-optimization/84016
27007 * config/i386/i386.c (ix86_builtin_vectorization_cost):
27008 Adjust vec_construct for the fact we need additional higher latency
27009 128bit inserts for AVX256 and AVX512 vector builds.
27010 (ix86_add_stmt_cost): Scale vector construction cost for
27013 2018-02-16 Richard Biener <rguenther@suse.de>
27015 PR tree-optimization/84417
27016 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
27017 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
27018 (non_rewritable_lvalue_p): Likewise, use poly-ints.
27020 2018-02-16 Martin Liska <mliska@suse.cz>
27023 * internal-fn.def (ASAN_CHECK): Set proper flags.
27024 (ASAN_MARK): Likewise.
27026 2018-02-16 Julia Koval <julia.koval@intel.com>
27028 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
27029 from PTA_CANNONLAKE.
27031 2018-02-16 Jakub Jelinek <jakub@redhat.com>
27034 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
27035 Use ++iter rather than iter++ for std::list iterators.
27036 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
27037 defer deleting them until all nodes in the forest are processed. Do
27038 free even leaf nodes. Change to_process into auto_vec.
27041 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
27042 * vec.h (vec_default_construct): Use memset instead of placement new
27043 if BROKEN_VALUE_INITIALIZATION is defined.
27044 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
27045 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
27048 PR rtl-optimization/83723
27049 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
27050 * lra.c (lra_substitute_pseudo): Likewise. If true, use
27051 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
27053 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
27055 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
27057 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
27059 PR rtl-optimization/81443
27060 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
27061 from inner REGs to paradoxical SUBREGs.
27063 2018-02-16 Richard Biener <rguenther@suse.de>
27065 PR tree-optimization/84399
27066 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
27067 For operands we can analyze at their definition make sure we can
27068 analyze them at each use as well.
27070 2018-02-16 Richard Biener <rguenther@suse.de>
27072 PR tree-optimization/84190
27073 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
27074 volatile accesses if the decl isn't volatile.
27076 2018-02-15 Jason Merrill <jason@redhat.com>
27078 PR c++/84314 - ICE with templates and fastcall attribute.
27079 * attribs.c (build_type_attribute_qual_variant): Don't clobber
27080 TYPE_CANONICAL on an existing type.
27082 2018-02-15 Jakub Jelinek <jakub@redhat.com>
27084 PR tree-optimization/84383
27085 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
27086 dstoff nor call operand_equal_p if dstbase is NULL.
27088 PR tree-optimization/84334
27089 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
27090 also a CONSTANT_CLASS_P, punt.
27092 2018-02-14 Jim Wilson <jimw@sifive.com>
27094 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
27095 first SMALL_OPERAND check. New local min_second_step. Move assert
27096 to where locals are set. Add TARGET_RVC support.
27097 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
27099 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
27101 * doc/invoke.texi: Correct -Wformat-overflow code sample.
27103 2018-02-14 Martin Sebor <msebor@redhat.com>
27105 PR tree-optimization/83698
27106 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
27107 arrays constrain the offset range to their bounds.
27108 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
27109 (builtin_access::overlap): Avoid setting the size of overlap if it's
27111 (maybe_diag_overlap): Also consider arrays when deciding what values
27112 of offsets to include in diagnostics.
27114 2018-02-14 Martin Sebor <msebor@redhat.com>
27117 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
27118 that correspond to the kind of a declaration.
27120 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
27123 * config/pa/pa.md: Load address of PIC label using the linkage table
27124 if the label is nonlocal.
27126 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
27128 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
27129 warning message if user requests -maltivec=be.
27130 * doc/invoke.texi: Document deprecation of -maltivec=be.
27132 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
27135 * config/rs6000/rs6000-c.c: Update definitions for
27136 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
27137 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
27139 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
27142 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
27143 add _get_ssp intrinsics. Remove argument from
27144 __builtin_ia32_rdssp[d|q].
27145 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
27146 * config/i386/i386-builtin.def: Remove argument from
27147 __builtin_ia32_rdssp[d|q].
27148 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
27149 ix86_expand_special_args_builtin for _rdssp[d|q].
27150 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
27151 Clear register before usage.
27152 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
27153 Add documentation for new _get_ssp and _inc_ssp intrinsics.
27155 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
27157 PR tree-optimization/84357
27158 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
27159 operand 1 of an ARRAY_REF too.
27161 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
27164 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
27165 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
27167 (set_of_reg): New struct.
27168 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
27169 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
27170 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
27172 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
27173 Split into bitclr, bitset, bitinvert patterns if appropriate.
27174 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
27175 use rx_fuse_in_memory_bitop.
27176 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
27177 to named insn, correct maximum insn length.
27179 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
27182 * machmode.def: Define a complex mode for PARTIAL_INT.
27183 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
27185 * doc/rtl.texi: Document CSPImode.
27186 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
27188 (msp430_hard_regno_nregs_with_padding): Likewise.
27190 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
27193 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
27195 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
27197 PR rtl-optimization/84169
27198 * combine.c (try_combine): New variable split_i2i3. Set it to true if
27199 we generated a parallel as new i3 and we split that to new i2 and i3
27200 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
27201 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
27202 those to i2, not i1. Partially rewrite this scan code.
27204 2018-02-13 Jakub Jelinek <jakub@redhat.com>
27207 * stor-layout.c (place_field): For variable length fields, adjust
27208 offset_align afterwards not just based on the field's alignment,
27209 but also on the size.
27211 PR middle-end/84309
27212 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
27213 of exps and logs in the use_exp2 case.
27215 2018-02-13 Jeff Law <law@redhat.com>
27217 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
27218 entry for "vector".
27220 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
27223 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
27227 * common.opt (gas-loc-support, gas-locview-support): New.
27228 (ginline-points, ginternal-reset-location-views): New.
27229 * doc/invoke.texi: Document them. Use @itemx where intended.
27230 (gvariable-location-views): Adjust.
27231 * target.def (reset_location_view): New.
27232 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
27233 (TARGET_RESET_LOCATION_VIEW): New.
27234 * doc/tm.texi: Rebuilt.
27235 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
27236 (dwarf2out_default_as_locview_support): New.
27237 (output_asm_line_debug_info): Use option variables.
27238 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
27239 (output_loc_list): Likewise.
27240 (add_high_low_attributes): Check option variables.
27241 Don't output entry view attribute in strict mode.
27242 (gen_inlined_subroutine_die): Check option variables.
27243 (dwarf2out_inline_entry): Likewise.
27244 (init_sections_and_labels): Likewise.
27245 (dwarf2out_early_finish): Likewise.
27246 (maybe_reset_location_view): New, from...
27247 (dwarf2out_var_location): ... here. Call it.
27248 * debug.h (dwarf2out_default_as_loc_support): Declare.
27249 (dwarf2out_default_as_locview_support): Declare.
27250 * hooks.c (hook_int_rtx_insn_0): New.
27251 * hooks.h (hook_int_rtx_insn_0): Declare.
27252 * toplev.c (process_options): Take -gas-loc-support and
27253 -gas-locview-support from dwarf2out. Enable
27254 -gvariable-location-views by default only with locview
27255 assembler support. Enable -ginternal-reset-location-views by
27256 default only if the target defines the corresponding hook.
27257 Enable -ginline-points by default if location views are
27258 enabled; force it disabled if statement frontiers are
27260 * tree-inline.c (expand_call_inline): Check option variables.
27261 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
27263 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
27265 PR tree-optimization/84321
27266 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
27267 handling. Also check whether the anti-range contains any values
27268 that satisfy the mask; switch to a VR_RANGE if not.
27270 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
27273 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
27275 2018-02-13 Martin Jambor <mjambor@suse.cz>
27278 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
27279 of call statements, also set location of a load to a temporary.
27281 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
27283 * config/rl78/rl78.c (add_vector_labels): New function.
27284 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
27285 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
27286 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
27287 which checks that no arguments are passed.
27288 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
27289 * doc/extend.texi: Documentation for the new attribute.
27291 2018-02-13 Andreas Schwab <schwab@suse.de>
27293 * config/riscv/linux.h (CPP_SPEC): Define.
27295 2018-02-13 Jakub Jelinek <jakub@redhat.com>
27298 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
27299 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
27300 OPTION_MASK_ISA_AES as first argument to def_builtin_const
27301 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
27302 instead of OPTION_MASK_ISA_PCLMUL as first argument to
27303 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
27304 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
27305 temporarily for AES and PCLMUL builtins.
27307 PR tree-optimization/84339
27308 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
27309 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
27312 PR middle-end/84309
27313 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
27314 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
27315 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
27317 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
27319 * omp-simd-clone.h: New file.
27320 * omp-simd-clone.c: Include omp-simd-clone.h.
27321 (expand_simd_clones): No longer static.
27322 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
27323 cgraph.h and omp-simd-clone.h.
27324 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
27325 (vect_recog_widen_shift_pattern): Formatting fix.
27326 (vect_pattern_recog_1): Don't check optab for calls.
27329 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
27330 operands[2] into a REG before using gen_lowpart on it.
27332 2018-02-12 Jeff Law <law@redhat.com>
27335 * config/sh/sh.c (find_barrier): Consider a sibling call
27338 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
27339 successfully back substituting a reg.
27341 2018-02-12 Richard Biener <rguenther@suse.de>
27343 PR tree-optimization/84037
27344 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
27345 parameter, move visited init to caller.
27346 (vect_slp_analyze_operations): Separate cost from validity
27347 check, initialize visited once for all instances.
27348 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
27350 * tree-vect-stmts.c (vect_model_simple_cost): Make early
27352 (vect_model_promotion_demotion_cost): Likewise.
27353 (vectorizable_bswap): Guard cost modeling with !slp_node
27354 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
27356 (vectorizable_call): Likewise.
27357 (vectorizable_conversion): Likewise.
27358 (vectorizable_assignment): Likewise.
27359 (vectorizable_shift): Likewise.
27360 (vectorizable_operation): Likewise.
27361 (vectorizable_store): Likewise.
27362 (vectorizable_load): Likewise.
27363 (vectorizable_condition): Likewise.
27364 (vectorizable_comparison): Likewise.
27366 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
27369 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
27370 (ASAN_MARK): Fix fnspec to account for return value, change pointer
27371 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
27373 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
27375 PR middle-end/83665
27376 * params.def (inline-min-speedup): Increase from 8 to 15.
27377 (max-inline-insns-auto): Decrease from 40 to 30.
27378 * ipa-split.c (consider_split): Add some buffer for function to
27379 be considered inlining candidate.
27380 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
27383 2018-02-12 Richard Biener <rguenther@suse.de>
27385 PR tree-optimization/84037
27386 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
27387 matched stmts if we cannot swap the non-matched ones.
27389 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
27391 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
27392 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
27393 _mm_maskz_scalef_round_ss): New intrinsics.
27394 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
27395 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
27396 __builtin_ia32_scalefss_round): Remove.
27397 (__builtin_ia32_scalefsd_mask_round,
27398 __builtin_ia32_scalefss_mask_round): New intrinsics.
27399 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
27400 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
27401 ((match_operand:VF_128 2 "<round_nimm_predicate>"
27402 "<round_constraint>")): Changed to ...
27403 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
27404 "<round_scalar_constraint>")): ... this.
27405 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
27406 %0, %1, %2<round_op3>}"): Changed to ...
27407 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
27408 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
27409 %2<round_scalar_mask_op3>}"): ... this.
27410 * config/i386/subst.md (round_scalar_nimm_predicate): New.
27412 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
27414 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
27415 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
27416 (_mm_maskz_sqrt_round_ss): New intrinsics.
27417 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
27418 (__builtin_ia32_sqrtsd_mask_round)
27419 (__builtin_ia32_sqrtss_mask_round): New builtins.
27420 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
27421 (__builtin_ia32_sqrtss_round): Remove.
27422 (__builtin_ia32_sqrtsd_mask_round)
27423 (__builtin_ia32_sqrtss_mask_round): New builtins.
27424 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
27425 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
27426 ((match_operand:VF_128 1 "vector_operand"
27427 "xBm,<round_constraint>")): Changed to ...
27428 ((match_operand:VF_128 1 "vector_operand"
27429 "xBm,<round_scalar_constraint>")): ... this.
27430 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
27431 %0, %2, %<iptr>1<round_op3>}): Changed to ...
27432 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
27433 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
27434 %<iptr>1<round_scalar_mask_op3>}): ... this.
27435 ((set_attr "prefix" "<round_prefix>")): Changed to ...
27436 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
27438 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
27441 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
27442 Cast vec_cmpeq result to correct type.
27443 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
27444 Cast vec_cmpgt result to correct type.
27446 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
27448 * final.c (final_scan_insn_1): Renamed from...
27449 (final_scan_insn): ... this. New wrapper, to recover
27450 seen from the outermost call in recursive ones.
27451 * config/sparc/sparc.c (output_return): Drop seen from call.
27452 (output_sibcall): Likewise.
27453 * config/visium/visium.c (output_branch): Likewise.
27455 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
27457 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
27460 2018-02-10 Alan Modra <amodra@gmail.com>
27463 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
27464 Specify LR as an input.
27466 2018-02-10 Jakub Jelinek <jakub@redhat.com>
27469 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
27470 remove_member_access_dummy_vars): New functions.
27471 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
27472 lower_omp_1, execute_lower_omp): Use them.
27474 PR rtl-optimization/84308
27475 * shrink-wrap.c (spread_components): Release todo vector.
27477 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
27479 PR rtl-optimization/57193
27480 * ira-color.c (struct allocno_color_data): Add member
27481 conflict_allocno_hard_prefs.
27482 (update_conflict_allocno_hard_prefs): New.
27483 (bucket_allocno_compare_func): Add a preference based on
27484 conflict_allocno_hard_prefs.
27485 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
27486 (color_allocnos): Remove a dead code. Initiate
27487 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
27489 2018-02-09 Jakub Jelinek <jakub@redhat.com>
27492 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
27493 constraint from =wa to wa. Avoid a subreg on the output operand,
27494 instead use a pseudo and subreg it in a move.
27495 (p9_xxbrd_<mode>): Changed to ...
27496 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
27497 (p9_xxbrd_v2df): New expander.
27498 (p9_xxbrw_<mode>): Changed to ...
27499 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
27500 (p9_xxbrw_v4sf): New expander.
27502 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
27504 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
27506 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
27509 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
27510 multiply in 32-bit mode.
27511 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
27512 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
27515 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
27517 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
27518 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
27519 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
27520 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
27522 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
27525 * dwarf2out.c (is_trivial_indirect_ref): New function.
27526 (dwarf2out_late_global_decl): Do not generate a location
27527 attribute for variables that have a non-trivial DECL_VALUE_EXPR
27528 and that are not defined in the current unit.
27530 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
27532 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
27533 instead of a libcall for UNORDERED.
27535 2018-02-09 Tamar Christina <tamar.christina@arm.com>
27538 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
27539 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
27541 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27544 * config/s390/s390.c (s390_set_current_function): Invoke
27545 s390_indirect_branch_settings also if fndecl didn't change.
27547 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
27549 * config/rs6000/rs6000.md (blockage): Set length to zero.
27551 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
27553 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
27555 2018-02-09 Jakub Jelinek <jakub@redhat.com>
27558 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
27559 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
27563 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
27564 PARALLEL incoming that failed vt_get_decl_and_offset check.
27566 PR middle-end/84237
27567 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
27568 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
27570 (get_variable_section): For decls in named .bss* sections pass true as
27571 second argument to bss_initializer_p.
27573 2018-02-09 Marek Polacek <polacek@redhat.com>
27574 Jakub Jelinek <jakub@redhat.com>
27577 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
27578 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
27579 Sync some changes from cxx_fold_indirect_ref.
27581 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
27583 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
27585 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
27586 (BLOCK_INLINE_ENTRY_LABEL): New.
27587 (dwarf2out_var_location): Disregard inline entry markers.
27588 (inline_entry_data): New struct.
27589 (inline_entry_data_hasher): New hashtable type.
27590 (inline_entry_data_hasher::hash): New.
27591 (inline_entry_data_hasher::equal): New.
27592 (inline_entry_data_table): New variable.
27593 (add_high_low_attributes): Add DW_AT_entry_pc and
27594 DW_AT_GNU_entry_view attributes if a pending entry is found
27595 in inline_entry_data_table. Add old entry_pc attribute only
27596 if debug nonbinding markers are disabled.
27597 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
27598 markers are enabled.
27599 (block_within_block_p, dwarf2out_inline_entry): New.
27600 (dwarf2out_finish): Check that no entries remained in
27601 inline_entry_data_table.
27602 * final.c (reemit_insn_block_notes): Handle inline entry notes.
27603 (final_scan_insn, notice_source_line): Likewise.
27604 (rest_of_clean_state): Skip inline entry markers.
27605 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
27607 * gimple.c (gimple_build_debug_inline_entry): New.
27608 * gimple.h (enum gimple_debug_subcode): Add
27609 GIMPLE_DEBUG_INLINE_ENTRY.
27610 (gimple_build_debug_inline_entry): Declare.
27611 (gimple_debug_inline_entry_p): New.
27612 (gimple_debug_nonbind_marker_p): Adjust.
27613 * insn-notes.def (INLINE_ENTRY): New.
27614 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
27615 inline entry marker notes.
27616 (print_insn): Likewise.
27617 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
27618 (INSN_DEBUG_MARKER_KIND): Likewise.
27619 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
27620 * tree-inline.c (expand_call_inline): Build and insert
27621 debug_inline_entry stmt.
27622 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
27623 inline entry blocks early, if nonbind markers are enabled.
27624 (dump_scope_block): Dump fragment info.
27625 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
27626 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
27627 (gimple_build_debug_inline_entry): New.
27628 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
27629 Enable/disable inline entry points too.
27630 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
27631 (DEBUG_INSN): Describe inline entry markers.
27633 * common.opt (gvariable-location-views): New.
27634 (gvariable-location-views=incompat5): New.
27635 * config.in: Rebuilt.
27636 * configure: Rebuilt.
27637 * configure.ac: Test assembler for view support.
27638 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
27639 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
27640 * dwarf2out.c (var_loc_view): New typedef.
27641 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
27642 (dwarf2out_locviews_in_attribute): New.
27643 (dwarf2out_locviews_in_loclist): New.
27644 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
27645 (enum dw_line_info_opcode): Add LI_adv_address.
27646 (struct dw_line_info_table): Add view.
27647 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
27648 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
27649 (zero_view_p): New variable.
27650 (ZERO_VIEW_P): New macro.
27651 (output_asm_line_debug_info): New.
27652 (struct var_loc_node): Add view.
27653 (add_AT_view_list, AT_loc_list): New.
27654 (add_var_loc_to_decl): Add view param. Test it against last.
27655 (new_loc_list): Add view params. Record them.
27656 (AT_loc_list_ptr): Handle loc and view lists.
27657 (view_list_to_loc_list_val_node): New.
27658 (print_dw_val): Handle dw_val_class_view_list.
27659 (size_of_die): Likewise.
27660 (value_format): Likewise.
27661 (loc_list_has_views): New.
27662 (gen_llsym): Set vl_symbol too.
27663 (maybe_gen_llsym, skip_loc_list_entry): New.
27664 (dwarf2out_maybe_output_loclist_view_pair): New.
27665 (output_loc_list): Output view list or entries too.
27666 (output_view_list_offset): New.
27667 (output_die): Handle dw_val_class_view_list.
27668 (output_dwarf_version): New.
27669 (output_compilation_unit_header): Use it.
27670 (output_skeleton_debug_sections): Likewise.
27671 (output_rnglists, output_line_info): Likewise.
27672 (output_pubnames, output_aranges): Update version comments.
27673 (output_one_line_info_table): Output view numbers in asm comments.
27674 (dw_loc_list): Determine current endview, pass it to new_loc_list.
27675 Call maybe_gen_llsym.
27676 (loc_list_from_tree_1): Adjust.
27677 (add_AT_location_description): Create view list attribute if
27678 needed, check it's absent otherwise.
27679 (convert_cfa_to_fb_loc_list): Adjust.
27680 (maybe_emit_file): Call output_asm_line_debug_info for test.
27681 (dwarf2out_var_location): Reset views as needed. Precompute
27682 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
27683 attribute. Set view.
27684 (new_line_info_table): Reset next view.
27685 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
27686 (dwarf2out_source_line): Likewise. Output view resets and labels to
27687 the assembler, or select appropriate line info opcodes.
27688 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
27689 (optimize_string_length): Catch it. Adjust.
27690 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
27691 dw_val_class_view_list, and remove it if no longer needed.
27692 (hash_loc_list): Hash view numbers.
27693 (loc_list_hasher::equal): Compare them.
27694 (optimize_location_lists): Check whether a view list symbol is
27695 needed, and whether the locview attribute is present, and
27696 whether they match. Remove the locview attribute if no longer
27698 (index_location_lists): Call skip_loc_list_entry for test.
27699 (dwarf2out_finish): Call output_asm_line_debug_info for test.
27700 Use output_dwarf_version.
27701 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
27702 (struct dw_val_node): Add val_view_list.
27703 * final.c (SEEN_NEXT_VIEW): New.
27704 (set_next_view_needed): New.
27705 (clear_next_view_needed): New.
27706 (maybe_output_next_view): New.
27707 (final_start_function): Rename to...
27708 (final_start_function_1): ... this. Take pointer to FIRST,
27709 add SEEN parameter. Emit param bindings in the initial view.
27710 (final_start_function): Reintroduce SEEN-less interface.
27711 (final): Rename to...
27712 (final_1): ... this. Take SEEN parameter. Output final pending
27713 next view at the end.
27714 (final): Reintroduce seen-less interface.
27715 (final_scan_insn): Output pending next view before switching
27716 sections or ending a block. Mark the next view as needed when
27717 outputting variable locations. Notify debug backend of section
27718 changes, and of location view changes.
27719 (rest_of_handle_final): Adjust.
27720 * toplev.c (process_options): Autodetect value for debug variable
27721 location views option. Warn on incompat5 without -gdwarf-5.
27722 * doc/invoke.texi (gvariable-location-views): New.
27723 (gvariable-location-views=incompat5): New.
27724 (gno-variable-location-views): New.
27726 2018-02-08 David Malcolm <dmalcolm@redhat.com>
27728 PR tree-optimization/84136
27729 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
27730 that the result of find_edge is non-NULL.
27732 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
27735 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
27736 storing integer register in SImode. Fix cost of 256 and 512
27737 byte aligned SSE register store.
27739 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
27741 * config/i386/i386.c (ix86_multiplication_cost): Fix
27742 multiplication cost for TARGET_AVX512DQ.
27744 2018-02-08 Marek Polacek <polacek@redhat.com>
27746 PR tree-optimization/84238
27747 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
27750 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
27752 PR tree-optimization/84265
27753 * tree-vect-stmts.c (vectorizable_store): Don't treat
27754 VMAT_CONTIGUOUS accesses as grouped.
27755 (vectorizable_load): Likewise.
27757 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
27759 PR tree-optimization/81635
27760 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
27761 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
27762 (test_round_for_mask): New functions.
27763 (wide_int_cc_tests): Call test_round_for_mask.
27764 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
27765 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
27766 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
27767 range returned by get_range_info.
27769 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
27772 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
27773 * symtab.c: Include builtins.h
27774 (symtab_node::output_to_lto_symbol_table_p): Move here
27775 from lto-streamer-out.c:output_symbol_p.
27776 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
27777 (output_symbol_p): Move all logic to symtab.c
27778 (produce_symtab): Update.
27780 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27782 * config/s390/s390-opts.h (enum indirect_branch): Define.
27783 * config/s390/s390-protos.h (s390_return_addr_from_memory)
27784 (s390_indirect_branch_via_thunk)
27785 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
27786 (enum s390_indirect_branch_type): Define.
27787 * config/s390/s390.c (struct s390_frame_layout, struct
27788 machine_function): Remove.
27789 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
27790 (indirect_branch_table_label_no, indirect_branch_table_name):
27792 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
27793 (enum s390_indirect_branch_option): Define.
27794 (s390_return_addr_from_memory): New function.
27795 (s390_handle_string_attribute): New function.
27796 (s390_attribute_table): Add new attribute handler.
27797 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
27798 (s390_indirect_branch_via_thunk): New function.
27799 (s390_indirect_branch_via_inline_thunk): New function.
27800 (s390_function_ok_for_sibcall): When jumping via thunk disallow
27801 sibling call optimization for non z10 compiles.
27802 (s390_emit_call): Force indirect branch target to be a single
27803 register. Add r1 clobber for non-z10 compiles.
27804 (s390_emit_epilogue): Emit return jump via return_use expander.
27805 (s390_reorg): Handle JUMP_INSNs as execute targets.
27806 (s390_option_override_internal): Perform validity checks for the
27807 new command line options.
27808 (s390_indirect_branch_attrvalue): New function.
27809 (s390_indirect_branch_settings): New function.
27810 (s390_set_current_function): Invoke s390_indirect_branch_settings.
27811 (s390_output_indirect_thunk_function): New function.
27812 (s390_code_end): Implement target hook.
27813 (s390_case_values_threshold): Implement target hook.
27814 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
27816 * config/s390/s390.h (struct s390_frame_layout)
27817 (struct machine_function): Move here from s390.c.
27818 (TARGET_INDIRECT_BRANCH_NOBP_RET)
27819 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
27820 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
27821 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
27822 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
27823 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
27824 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
27825 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
27826 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
27827 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
27828 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
27829 (mnemonic attribute): Add values which aren't recognized
27831 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
27832 pattern for branch conversion. Fix mnemonic attribute.
27833 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
27834 indirect branch via thunk if requested.
27835 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
27836 ("*indirect_jump"): Disable for branch conversion using out of
27838 ("indirect_jump_via_thunk<mode>_z10")
27839 ("indirect_jump_via_thunk<mode>")
27840 ("indirect_jump_via_inlinethunk<mode>_z10")
27841 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
27842 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
27843 ("casesi_jump_via_inlinethunk<mode>_z10")
27844 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
27845 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
27846 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
27847 ("*indirect2_jump"): Disable for branch conversion.
27848 ("casesi_jump"): Turn into expander and expand patterns for branch
27850 ("return_use"): New expander.
27851 ("*return"): Emit return via thunk and rename it to ...
27852 ("*return<mode>"): ... this one.
27853 * config/s390/s390.opt: Add new options and and enum for the
27856 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
27858 * lra-constraints.c (match_reload): Unconditionally use
27859 gen_lowpart_SUBREG, rather than selecting between that
27860 and equivalent gen_rtx_SUBREG code.
27862 2018-02-08 Richard Biener <rguenther@suse.de>
27864 PR tree-optimization/84233
27865 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
27866 changed flag instead of boguously re-using phi_inserted.
27868 2018-02-08 Martin Jambor <mjambor@suse.cz>
27870 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
27871 static local variables.
27873 2018-02-08 Richard Biener <rguenther@suse.de>
27875 PR tree-optimization/84278
27876 * tree-vect-stmts.c (vectorizable_store): When looking for
27877 smaller vector types to perform grouped strided loads/stores
27878 make sure the mode is supported by the target.
27879 (vectorizable_load): Likewise.
27881 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
27883 * config/aarch64/aarch64.c (aarch64_components_for_bb):
27884 Increase LDP/STP opportunities by adding adjacent callee-saves.
27886 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
27888 PR rtl-optimization/84068
27889 PR rtl-optimization/83459
27890 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
27892 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
27894 PR tree-optimization/84224
27895 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
27896 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
27897 non-zero arguments.
27899 2018-02-07 Iain Sandoe <iain@codesourcery.com>
27902 * config/rs6000/altivec.md (*restore_world): Remove LR use.
27903 * config/rs6000/predicates.md (restore_world_operation): Adjust op
27904 count, remove one USE.
27906 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
27908 * doc/install.texi (Configuration): Document the
27909 --with-long-double-format={ibm,ieee} PowerPC configuration
27913 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
27914 Convert from define_expand to be define_insn_and_split. Rework
27915 float/double/_Float128 conversions to QI/HI/SImode to work with
27916 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
27917 conversions to QI/HImode types did a store and then a load to
27918 truncate the value. For conversions to VSX registers, don't split
27919 the insn, instead emit the code directly. Use the code iterator
27920 any_fix to combine signed and unsigned conversions.
27921 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
27922 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
27923 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
27924 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
27925 (fix_<mode>di2_hw): Likewise.
27926 (fixuns_<mode>di2_hw): Likewise.
27927 (fix_<mode>si2_hw): Likewise.
27928 (fixuns_<mode>si2_hw): Likewise.
27929 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
27930 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
27931 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
27932 fix<uns>_trunc<SFDF:mode>si2_p8.
27933 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
27934 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
27935 (fix<uns>_<mode>_mem): Likewise.
27936 (fctiw<u>z_<mode>_mem): Likewise.
27937 (fix<uns>_<mode>_mem): Likewise.
27938 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
27939 the register allocator from doing a direct move to the GPRs to do
27940 a store, and instead use the ISA 3.0 store byte/half-word from
27941 vector register instruction. For IEEE 128-bit floating point,
27942 also optimize stores of 32-bit ints.
27943 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
27945 2018-02-07 Alan Hayward <alan.hayward@arm.com>
27947 * genextract.c (push_pathstr_operand): New function to support
27949 (walk_rtx): Call push_pathstr_operand.
27950 (print_path): Support [a-zA-Z].
27952 2018-02-07 Richard Biener <rguenther@suse.de>
27954 PR tree-optimization/84037
27955 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
27956 (cse_and_gimplify_to_preheader): Declare.
27957 (vect_get_place_in_interleaving_chain): Likewise.
27958 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
27960 (_loop_vec_info::~_loop_vec_info): Delete it.
27961 (cse_and_gimplify_to_preheader): New function.
27962 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
27963 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
27964 (vectorizable_load): Likewise. For grouped stores always base
27965 the IV on the first element.
27966 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
27967 condition before gimplifying.
27969 2018-02-07 Jakub Jelinek <jakub@redhat.com>
27971 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
27972 *DIV_EXPR and *MOD_EXPR.
27974 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
27977 * config/i386/i386.c (ix86_option_override_internal): Mask out
27978 the CF_SET bit when checking -fcf-protection.
27980 2018-02-07 Tom de Vries <tom@codesourcery.com>
27983 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
27986 2018-02-07 Richard Biener <rguenther@suse.de>
27988 PR tree-optimization/84204
27989 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
27992 PR tree-optimization/84205
27993 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
27994 special-case isl_ast_op_zdiv_r.
27996 PR tree-optimization/84223
27997 * graphite-scop-detection.c (gather_bbs::before_dom_children):
27998 Only add conditions from within the region.
27999 (gather_bbs::after_dom_children): Adjust.
28001 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
28004 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
28005 * config/avr/avr.md: Only post-reload split REG-REG moves if
28006 either register is GENERAL_REG_P.
28008 2018-02-07 Jakub Jelinek <jakub@redhat.com>
28010 PR tree-optimization/84235
28011 * tree-ssa-scopedtables.c
28012 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
28013 if the subtraction is performed in floating point type where NaNs are
28014 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
28015 build 1. Formatting fix.
28017 2018-02-06 Jakub Jelinek <jakub@redhat.com>
28020 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
28021 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
28022 and skip it regardless of bb boundaries. Use CALL_P macro,
28023 don't test INSN_P (insn) together with CALL_P or JUMP_P check
28024 unnecessarily, formatting fix.
28026 2018-02-06 Michael Collison <michael.collison@arm.com>
28028 * config/arm/thumb2.md:
28029 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
28030 (*thumb_mov_notscc): Ditto.
28032 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
28035 * config/rs6000/rs6000.md (su code attribute): Use "u" for
28036 unsigned_fix, not "s".
28038 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28040 * configure.ac (gcc_fn_eh_frame_ro): New function.
28041 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
28042 correct .eh_frame permissions.
28043 * configure: Regenerate.
28045 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
28047 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
28048 irrelevant options.
28050 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28052 * config/rs6000/rs6000.c (rs6000_option_override_internal):
28053 Display warning message for -mno-speculate-indirect-jumps.
28055 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
28057 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
28059 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
28061 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
28063 PR tree-optimization/84225
28064 * tree-eh.c (find_trapping_overflow): Only call
28065 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
28067 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
28070 * config/i386/i386.c: Reimplement the check of possible options
28071 -mibt/-mshstk conbination. Change error messages.
28072 * doc/invoke.texi: Fix a typo: remove extra '='.
28074 2018-02-06 Marek Polacek <polacek@redhat.com>
28076 PR tree-optimization/84228
28077 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
28079 2018-02-06 Tamar Christina <tamar.christina@arm.com>
28082 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
28083 emitted arch directives.
28084 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
28085 __ARM_FEATURE_COPROC before changing architectures.
28087 2018-02-06 Richard Biener <rguenther@suse.de>
28089 * config/i386/i386.c (print_reg): Fix typo.
28090 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
28092 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
28094 * configure: Regenerate.
28096 2018-02-05 Martin Sebor <msebor@redhat.com>
28098 PR tree-optimization/83369
28099 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
28102 2018-02-05 Martin Liska <mliska@suse.cz>
28104 * doc/invoke.texi: Cherry-pick upstream r323995.
28106 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
28108 * ira.c (ira_init_register_move_cost): Adjust comment.
28110 2018-02-05 Martin Liska <mliska@suse.cz>
28112 PR gcov-profile/84137
28113 * doc/gcov.texi: Fix typo in documentation.
28115 2018-02-05 Martin Liska <mliska@suse.cz>
28117 PR gcov-profile/83879
28118 * doc/gcov.texi: Document necessity of --dynamic-list-data when
28119 using dlopen functionality.
28121 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
28123 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
28124 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
28125 _mm_maskz_range_ss, _mm_mask_range_round_ss,
28126 _mm_maskz_range_round_ss): New intrinsics.
28127 (__builtin_ia32_rangesd128_round)
28128 (__builtin_ia32_rangess128_round): Remove.
28129 (__builtin_ia32_rangesd128_mask_round,
28130 __builtin_ia32_rangess128_mask_round): New builtins.
28131 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
28132 __builtin_ia32_rangess128_round): Remove.
28133 (__builtin_ia32_rangesd128_mask_round,
28134 __builtin_ia32_rangess128_mask_round): New builtins.
28135 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
28136 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
28137 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
28138 "<round_saeonly_constraint>")): Changed to ...
28139 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
28140 "<round_saeonly_scalar_constraint>")): ... this.
28141 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
28142 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
28143 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
28144 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
28145 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
28147 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
28149 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
28151 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
28152 Remove all values except native, 8540 and 8548.
28154 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
28156 * config/i386/i386.c (ix86_output_function_return): Pass
28157 INVALID_REGNUM, instead of -1, as invalid register number to
28158 indirect_thunk_name and output_indirect_thunk.
28160 2018-02-02 Julia Koval <julia.koval@intel.com>
28162 * config.gcc: Add -march=icelake.
28163 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
28164 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
28165 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
28166 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
28167 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
28168 (processor_target_table): Add icelake.
28169 (ix86_option_override_internal): Handle new PTAs.
28170 (get_builtin_code_for_version): Handle icelake.
28171 (M_INTEL_COREI7_ICELAKE): New.
28172 (fold_builtin_cpu): Handle icelake.
28173 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
28174 * doc/invoke.texi: Add -march=icelake.
28176 2018-02-02 Julia Koval <julia.koval@intel.com>
28178 * config/i386/i386.c (ix86_option_override_internal): Change flags type
28179 to wide_int_bitmask.
28180 * wide-int-bitmask.h: New.
28182 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
28185 * config/i386/i386.md: Replace Pmode with word_mode in
28186 builtin_setjmp_setup and builtin_longjmp to support x32.
28188 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
28192 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
28194 (rs6000_supported_cpu_names): New static variable.
28195 (linux_cpu_translation_table): Likewise.
28196 (elf_platform) <cpu>: Define new static variable and use it.
28197 Translate kernel AT_PLATFORM name to canonical name if needed.
28198 Error if platform name is unknown.
28200 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
28203 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
28205 2018-02-01 Jeff Law <law@redhat.com>
28208 * config/i386/i386.c (release_scratch_register_on_entry): Add new
28209 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
28210 the scratch if RELEASE_VIA_POP is false.
28211 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
28212 If we have to save a temporary register, decrement SIZE appropriately.
28213 Pass new arguments to release_scratch_register_on_entry.
28214 (ix86_adjust_stack_and_probe): Likewise.
28215 (ix86_emit_probe_stack_range): Pass new arguments to
28216 release_scratch_register_on_entry.
28218 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
28220 PR rtl-optimization/84157
28221 * combine.c (change_zero_ext): Use REG_P predicate in
28222 front of HARD_REGISTER_P predicate.
28224 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
28226 * config/avr/avr.c (avr_option_override): Move disabling of
28227 -fdelete-null-pointer-checks to...
28228 * common/config/avr/avr-common.c (avr_option_optimization_table):
28231 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
28233 PR tree-optimization/81635
28234 * tree-data-ref.c (split_constant_offset_1): For types that
28235 wrap on overflow, try to use range info to prove that wrapping
28238 2018-02-01 Renlin Li <renlin.li@arm.com>
28241 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
28242 TAILCALL_ADDR_REGS.
28243 (aarch64_register_move_cost): Likewise.
28244 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
28245 TAILCALL_ADDR_REGS.
28246 (REG_CLASS_NAMES): Likewise.
28247 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
28248 TAILCALL_ADDR_REGS. Remove IP registers.
28249 * config/aarch64/aarch64.md (Ucs): Update register constraint.
28251 2018-02-01 Richard Biener <rguenther@suse.de>
28253 * domwalk.h (dom_walker::dom_walker): Add additional constructor
28254 for specifying RPO order and allow NULL for that.
28255 * domwalk.c (dom_walker::dom_walker): Likewise.
28256 (dom_walker::walk): Handle NULL RPO order.
28257 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
28259 (rewrite_update_dom_walker): Likewise.
28260 (mark_def_dom_walker): Likewise.
28262 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
28264 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
28265 (aarch64_maybe_expand_sve_subreg_move): Declare.
28266 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
28267 * config/aarch64/predicates.md (aarch64_any_register_operand): New
28269 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
28270 that are semantically a reverse operation.
28271 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
28272 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
28273 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
28275 (aarch64_can_change_mode_class): For big-endian, forbid changes
28276 between two SVE modes if they have different element sizes.
28278 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
28280 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
28281 the TImode handling for big-endian targets.
28283 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
28285 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
28286 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
28288 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
28289 Remove BSWAP handing for big-endian targets and use the form of
28290 LD1RQ appropariate for the mode.
28292 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
28294 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
28295 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
28296 duplicated element.
28298 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
28301 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
28302 check for operands that need to go through aarch64_sve_reload_be.
28304 2018-02-01 Jakub Jelinek <jakub@redhat.com>
28306 PR tree-optimization/81661
28307 PR tree-optimization/84117
28308 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
28309 * tree-eh.c: Include gimplify.h.
28310 (find_trapping_overflow, replace_trapping_overflow,
28311 rewrite_to_non_trapping_overflow): New functions.
28312 * tree-vect-loop.c: Include tree-eh.h.
28313 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
28314 * tree-data-ref.c: Include tree-eh.h.
28315 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
28317 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
28319 PR rtl-optimization/84123
28320 * combine.c (change_zero_ext): Check if hard register satisfies
28321 can_change_dest_mode before calling gen_lowpart_SUBREG.
28323 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
28326 * ira.c (ira_init_register_move_cost): Remove assert.
28328 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
28330 PR rtl-optimization/84071
28331 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
28332 * doc/tm.texi: Regenerate.
28334 2018-01-31 Richard Biener <rguenther@suse.de>
28336 PR tree-optimization/84132
28337 * tree-data-ref.c (analyze_miv_subscript): Properly
28338 check whether evolution_function_is_affine_multivariate_p
28339 before calling gcd_of_steps_may_divide_p.
28341 2018-01-31 Julia Koval <julia.koval@intel.com>
28344 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
28345 * config/i386/i386.md (rdpid_rex64) New.
28346 (rdpid): Make 32bit only.
28348 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
28351 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
28352 an IDENTIFIER_NODE for FUNCTION_TYPE's.
28354 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
28357 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
28359 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
28361 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
28363 PR rtl-optimization/84071
28364 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
28365 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
28367 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
28369 * config/arc/arc.c (arc_handle_aux_attribute): New function.
28370 (arc_attribute_table): Add 'aux' attribute.
28371 (arc_in_small_data_p): Consider aux like variables.
28372 (arc_is_aux_reg_p): New function.
28373 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
28374 (arc_get_aux_arg): New function.
28375 (prepare_move_operands): Handle aux-register access.
28376 (arc_handle_aux_attribute): New function.
28377 * doc/extend.texi (ARC Variable attributes): Add subsection.
28379 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
28381 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
28382 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
28383 (arc_attribute_table): Add 'uncached' attribute.
28384 (arc_print_operand): Print '.di' flag for uncached memory
28386 (arc_in_small_data_p): Do not consider for small data the uncached
28388 (arc_is_uncached_mem_p): New function.
28389 * config/arc/predicates.md (compact_store_memory_operand): Check
28390 for uncached memory accesses.
28391 (nonvol_nonimm_operand): Likewise.
28392 * doc/extend.texi (ARC Type Attribute): New subsection.
28394 2018-01-31 Jakub Jelinek <jakub@redhat.com>
28397 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
28398 falign-loops=): Add Optimization flag.
28400 2018-01-30 Jeff Law <law@redhat.com>
28403 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
28404 INT_REGISTERS_SAVED. Check it prior to calling
28405 get_scratch_register_on_entry.
28406 (ix86_adjust_stack_and_probe): Similarly.
28407 (ix86_emit_probe_stack_range): Similarly.
28408 (ix86_expand_prologue): Corresponding changes.
28410 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28413 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
28414 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
28416 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
28419 * lra-constraints.c (curr_insn_transform): Process AND in the
28422 2018-01-30 Jakub Jelinek <jakub@redhat.com>
28424 PR rtl-optimization/83986
28425 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
28426 dependence against last_pending_memory_flush in addition to
28427 pending_jump_insns.
28429 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
28431 PR tree-optimization/81611
28432 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
28435 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28438 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
28441 2018-01-30 Richard Biener <rguenther@suse.de>
28442 Jakub Jelinek <jakub@redhat.com>
28444 PR tree-optimization/84111
28445 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
28446 inner loops added during recursion, as they don't have up-to-date
28449 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
28452 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
28453 (can_inline_edge_by_limits_p): ... here.
28454 (can_early_inline_edge_p, check_callers,
28455 update_caller_keys, update_callee_keys, recursive_inlining,
28456 add_new_edges_to_heap, speculation_useful_p,
28457 inline_small_functions,
28458 inline_small_functions, flatten_function,
28459 inline_to_all_callers_1): Update.
28461 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
28463 * profile-count.c (profile_count::combine_with_ipa_count): Handle
28466 2018-01-30 Richard Biener <rguenther@suse.de>
28468 PR tree-optimization/83008
28469 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
28470 invariant and constant vector uses in stmts when they need
28471 more than one stmt.
28473 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28476 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
28477 * configure: Regenerate.
28479 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
28481 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
28483 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
28484 Use gen_rtx_REG rather than gen_lowpart.
28486 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
28488 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
28489 rather than 0 when creating partial subregs.
28491 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
28493 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
28496 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
28499 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
28500 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
28501 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
28502 flags. This restores the settings used before the 2017-07-24.
28503 Turning off pre increment/decrement/modify allows IVOPTS to
28504 optimize DF/SF loops where the index is an int.
28506 2018-01-29 Richard Biener <rguenther@suse.de>
28507 Kelvin Nilsen <kelvin@gcc.gnu.org>
28510 * tree-vect-stmts.c (vectorizable_call): Don't call
28511 targetm.vectorize_builtin_md_vectorized_function if callee is
28514 2018-01-22 Carl Love <cel@us.ibm.com>
28516 * doc/extend.tex: Fix typo in second arg in
28517 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
28519 2018-01-29 Richard Biener <rguenther@suse.de>
28521 PR tree-optimization/84086
28522 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
28523 (flush_ssaname_freelist): When SSA names were released reset
28524 the SCEV hash table.
28526 2018-01-29 Richard Biener <rguenther@suse.de>
28528 PR tree-optimization/84057
28529 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
28530 removed paths when removing edges.
28532 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
28534 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
28535 -mfunction-return=@var{choice}.
28537 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
28539 PR diagnostic/84034
28540 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
28541 Handle CR like TAB.
28542 (layout::print_source_line): Likewise.
28543 (test_get_line_width_without_trailing_whitespace): Add test cases.
28545 2018-01-27 Jakub Jelinek <jakub@redhat.com>
28547 PR middle-end/84040
28548 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
28551 2018-01-26 Jim Wilson <jimw@sifive.com>
28553 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
28555 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
28558 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28560 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
28561 and CMP + SUB-immediate -> SUBS.
28563 2018-01-26 Martin Sebor <msebor@redhat.com>
28565 PR tree-optimization/83896
28566 * tree-ssa-strlen.c (get_string_len): Rename...
28567 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
28568 Avoid assuming length is constant.
28569 (handle_char_store): Use HOST_WIDE_INT for string length.
28571 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
28574 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
28575 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
28577 2018-01-26 Richard Biener <rguenther@suse.de>
28579 PR rtl-optimization/84003
28580 * dse.c (record_store): Only record redundant stores when
28581 the earlier store aliases at least all accesses the later one does.
28583 2018-01-26 Jakub Jelinek <jakub@redhat.com>
28585 PR rtl-optimization/83985
28586 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
28587 REG_CFA_RESTORE insns.
28588 (delete_unmarked_insns): Don't ignore separate shrink wrapping
28589 REG_CFA_RESTORE insns here.
28592 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
28593 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
28595 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
28597 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
28598 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
28599 (arc_init): Likewise.
28600 (arc_override_options): Likewise.
28601 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
28603 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
28605 * config/arc/arc.h (TARGET_DBNZ): Define.
28606 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
28607 properly set the tune attribute.
28608 (dbnz): Use TARGET_DBNZ guard.
28609 * config/arc/arc.opt (mtune): Add core3 option.
28611 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
28613 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
28614 recognize new pic like addresses.
28615 (arc_delegitimize_address): Clean up.
28617 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
28619 * config/arc/arc-arches.def: Option mrf16 valid for all
28621 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
28622 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
28623 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
28624 * config/arc/arc-tables.opt: Regenerate.
28625 * config/arc/arc.c (arc_conditional_register_usage): Handle
28626 reduced register file case.
28627 (arc_file_start): Set must have build attributes.
28628 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
28629 mrf16 option value.
28630 * config/arc/arc.opt (mrf16): Add new option.
28631 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
28632 * config/arc/genmultilib.awk: Handle new mrf16 option.
28633 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
28634 * config/arc/t-multilib: Regenerate.
28635 * doc/invoke.texi (ARC Options): Document mrf16 option.
28637 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
28639 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
28640 * config/arc/arc.c (arc_handle_secure_attribute): New function.
28641 (arc_attribute_table): Add 'secure_call' attribute.
28642 (arc_print_operand): Print secure call operand.
28643 (arc_function_ok_for_sibcall): Don't optimize tail calls when
28645 (arc_is_secure_call_p): New function. * config/arc/arc.md
28646 (call_i): Add support for sjli instruction.
28647 (call_value_i): Likewise.
28648 * config/arc/constraints.md (Csc): New constraint.
28650 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
28651 John Eric Martin <John.Martin@emmicro-us.com>
28653 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
28654 * config/arc/arc.c (_arc_jli_section): New struct.
28655 (arc_jli_section): New type.
28656 (rc_jli_sections): New static variable.
28657 (arc_handle_jli_attribute): New function.
28658 (arc_attribute_table): Add jli_always and jli_fixed attribute.
28659 (arc_file_end): New function.
28660 (TARGET_ASM_FILE_END): Define.
28661 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
28662 (arc_add_jli_section): New function.
28663 (jli_call_scan): Likewise.
28664 (arc_reorg): Call jli_call_scan.
28665 (arc_output_addsi): Remove 'S' from printing asm operand.
28666 (arc_is_jli_call_p): New function.
28667 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
28669 (movhi_insn): Likewise.
28670 (movsi_insn): Likewise.
28671 (movsi_set_cc_insn): Likewise.
28672 (loadqi_update): Likewise.
28673 (load_zeroextendqisi_update): Likewise.
28674 (load_signextendqisi_update): Likewise.
28675 (loadhi_update): Likewise.
28676 (load_zeroextendhisi_update): Likewise.
28677 (load_signextendhisi_update): Likewise.
28678 (loadsi_update): Likewise.
28679 (loadsf_update): Likewise.
28680 (movsicc_insn): Likewise.
28681 (bset_insn): Likewise.
28682 (bxor_insn): Likewise.
28683 (bclr_insn): Likewise.
28684 (bmsk_insn): Likewise.
28685 (bicsi3_insn): Likewise.
28686 (cmpsi_cc_c_insn): Likewise.
28687 (movsi_ne): Likewise.
28688 (movsi_cond_exec): Likewise.
28689 (clrsbsi2): Likewise.
28690 (norm_f): Likewise.
28700 (call_i): Remove 'S' asm letter, add jli instruction.
28701 (call_value_i): Likewise.
28702 * config/arc/arc.op (mjli-always): New option.
28703 * config/arc/constraints.md (Cji): New constraint.
28704 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
28706 (subsf3_fpx): Likewise.
28707 (mulsf3_fpx): Likewise.
28708 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
28710 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
28711 function attrbutes.
28712 * doc/invoke.texi (ARC): Document mjli-always option.
28714 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
28716 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
28717 avoid addition with 0 and use incw and decw where possible.
28719 2018-01-26 Richard Biener <rguenther@suse.de>
28721 PR tree-optimization/81082
28722 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
28723 association if it requires casting to unsigned.
28724 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
28725 from fold_plusminus_mult_expr to catch important cases late when
28726 range info is available.
28728 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28730 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
28731 * configure.ac (hidden_linkonce): New test.
28732 * configure: Regenerate.
28733 * config.in: Regenerate.
28735 2018-01-26 Julia Koval <julia.koval@intel.com>
28737 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
28738 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
28739 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
28740 _mm_mask_bitshuffle_epi64_mask): Fix type.
28741 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
28742 USI_FTYPE_V4DI_V4DI_USI): Remove.
28743 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
28744 __builtin_ia32_vpshufbitqmb256_mask,
28745 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
28746 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
28747 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
28749 2018-01-26 Alan Modra <amodra@gmail.com>
28752 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
28753 UNSPEC_VBPERMQ. Sort other unspecs.
28755 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
28757 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
28759 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
28761 PR middle-end/83055
28762 * predict.c (drop_profile): Do not push/pop cfun; update also
28764 (handle_missing_profiles): Fix logic looking for zero profiles.
28766 2018-01-25 Jakub Jelinek <jakub@redhat.com>
28768 PR middle-end/83977
28769 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
28770 on functions with #pragma omp declare simd or functions with simd
28772 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
28773 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
28774 Remove trailing \n from warning_at calls.
28776 2018-01-25 Tom de Vries <tom@codesourcery.com>
28779 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
28780 for neutered workers.
28782 2018-01-24 Joseph Myers <joseph@codesourcery.com>
28785 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
28786 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
28788 2018-01-24 Jeff Law <law@redhat.com>
28791 * i386.c (get_probe_interval): Move to earlier point.
28792 (ix86_compute_frame_layout): If -fstack-clash-protection and
28793 the frame is larger than the probe interval, then use pushes
28794 to save registers rather than reg->mem moves.
28795 (ix86_expand_prologue): Remove conditional for int_registers_saved
28798 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
28801 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
28802 min/max for never referenced object.
28804 2018-01-24 Jakub Jelinek <jakub@redhat.com>
28806 PR middle-end/83977
28807 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
28809 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
28810 attributes from DECL_ATTRIBUTES (decl) without affecting
28811 DECL_ATTRIBUTES (current_function_decl).
28812 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
28813 functions with non-NULL DECL_ABSTRACT_ORIGIN.
28815 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
28817 PR tree-optimization/83979
28818 * fold-const.c (fold_comparison): Use constant_boolean_node
28819 instead of boolean_{true,false}_node.
28821 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
28823 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
28826 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28828 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
28829 Simplify the clause that sets the length attribute.
28830 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
28831 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
28832 clause that sets the length attribute.
28833 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
28835 2018-01-24 Tom de Vries <tom@codesourcery.com>
28838 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
28839 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
28840 Add strict parameter.
28841 (prevent_branch_around_nothing): Insert dummy insn between branch to
28842 label and label with no ptx insn inbetween.
28843 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
28845 2018-01-24 Tom de Vries <tom@codesourcery.com>
28848 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
28849 for neutered threads in warp.
28850 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
28852 2018-01-24 Richard Biener <rguenther@suse.de>
28854 PR tree-optimization/83176
28855 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
28858 2018-01-24 Richard Biener <rguenther@suse.de>
28860 PR tree-optimization/82819
28861 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
28862 code generating pluses that are no-ops in the target precision.
28864 2018-01-24 Richard Biener <rguenther@suse.de>
28866 PR middle-end/84000
28867 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
28869 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
28871 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
28872 to merge probabilities.
28873 * predict.c (probably_never_executed): Also mark as cold functions
28874 with global 0 profile and guessed local profile.
28875 * profile-count.c (profile_probability::combine_with_count): New
28877 * profile-count.h (profile_probability::operator*,
28878 profile_probability::operator*=, profile_probability::operator/,
28879 profile_probability::operator/=): Reduce precision to adjusted
28880 and set value to guessed on contradictory divisions.
28881 (profile_probability::combine_with_freq): Remove.
28882 (profile_probability::combine_wiht_count): Declare.
28883 (profile_count::force_nonzero):: Set to adjusted.
28884 (profile_count::probability_in):: Set quality to adjusted.
28885 * tree-ssa-tail-merge.c (replace_block_by): Use
28886 combine_with_count.
28888 2018-01-23 Andrew Waterman <andrew@sifive.com>
28889 Jim Wilson <jimw@sifive.com>
28891 * config/riscv/riscv.c (riscv_stack_boundary): New.
28892 (riscv_option_override): Set riscv_stack_boundary. Handle
28893 riscv_preferred_stack_boundary_arg.
28894 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
28895 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
28896 (STACK_BOUNDARY): Set to riscv_stack_boundary.
28897 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
28898 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
28899 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
28901 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
28904 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
28905 of struct ix86_frame.
28906 (ix86_expand_epilogue): Likewise. Add a local variable for
28907 the reg_save_offset field in struct ix86_frame.
28909 2018-01-23 Bin Cheng <bin.cheng@arm.com>
28911 PR tree-optimization/82604
28912 * tree-loop-distribution.c (enum partition_kind): New enum item
28913 PKIND_PARTIAL_MEMSET.
28914 (partition_builtin_p): Support above new enum item.
28915 (generate_code_for_partition): Ditto.
28916 (compute_access_range): Differentiate cases that equality can be
28917 proven at all loops, the innermost loops or no loops.
28918 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
28919 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
28920 (finalize_partitions, distribute_loop): Don't fuse partition of
28921 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
28922 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
28923 parloop is enabled.
28925 2018-01-23 Martin Liska <mliska@suse.cz>
28927 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
28928 order to ignore the predictor.
28929 (PRED_POLYMORPHIC_CALL): Likewise.
28930 (PRED_RECURSIVE_CALL): Likewise.
28932 2018-01-23 Martin Liska <mliska@suse.cz>
28934 * tree-profile.c (tree_profiling): Print function header to
28935 aware reader which function we are working on.
28936 * value-prof.c (gimple_find_values_to_profile): Do not print
28937 not interesting value histograms.
28939 2018-01-23 Martin Liska <mliska@suse.cz>
28941 * profile-count.h (enum profile_quality): Add
28942 profile_uninitialized as the first value. Do not number values
28943 as they are zero based.
28944 (profile_count::verify): Update sanity check.
28945 (profile_probability::verify): Likewise.
28947 2018-01-23 Nathan Sidwell <nathan@acm.org>
28949 * doc/invoke.texi (ffor-scope): Deprecate.
28951 2018-01-23 David Malcolm <dmalcolm@redhat.com>
28953 PR tree-optimization/83510
28954 * domwalk.c (set_all_edges_as_executable): New function.
28955 (dom_walker::dom_walker): Convert bool param
28956 "skip_unreachable_blocks" to enum reachability. Move setup of
28957 edge flags to set_all_edges_as_executable and only do it when
28958 reachability is REACHABLE_BLOCKS.
28959 * domwalk.h (enum dom_walker::reachability): New enum.
28960 (dom_walker::dom_walker): Convert bool param
28961 "skip_unreachable_blocks" to enum reachability.
28962 (set_all_edges_as_executable): New decl.
28963 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
28964 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
28966 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
28967 but converting true to REACHABLE_BLOCKS.
28968 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
28970 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
28971 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
28972 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
28974 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
28975 if check_all_array_refs will be called.
28977 2018-01-23 David Malcolm <dmalcolm@redhat.com>
28979 * tree.c (selftest::test_location_wrappers): Add more test
28982 2018-01-23 David Malcolm <dmalcolm@redhat.com>
28984 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
28985 (selftest::test_bit_in_range): Likewise.
28987 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
28990 * doc/sourcebuild.texi (vect_float): Say that the selector
28991 only describes the situation when -funsafe-math-optimizations is on.
28992 (vect_float_strict): Document.
28994 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
28996 PR tree-optimization/83965
28997 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
28998 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
28999 instead of checking only for a reduction.
29000 (vect_recog_widen_sum_pattern): Likewise.
29002 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
29004 * predict.c (probably_never_executed): Only use precise profile info.
29005 (compute_function_frequency): Skip after inlining hack since we now
29006 have quality checking.
29008 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
29010 * profile-count.h (profile_probability::very_unlikely,
29011 profile_probability::unlikely, profile_probability::even): Set
29012 precision to guessed.
29014 2018-01-23 Richard Biener <rguenther@suse.de>
29016 PR tree-optimization/83963
29017 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
29018 Properly terminate dominator walk when crossing the exit edge not
29019 when visiting its source block.
29021 2018-01-23 Jakub Jelinek <jakub@redhat.com>
29024 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
29025 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
29027 2018-01-22 Jakub Jelinek <jakub@redhat.com>
29029 PR tree-optimization/83957
29030 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
29031 semicolon after for body surrounded by braces.
29033 PR tree-optimization/83081
29034 * profile-count.h (profile_probability::split): New method.
29035 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
29036 Use profile_probability::split.
29037 (do_compare_rtx_and_jump): Fix adjustment of probabilities
29038 when splitting a single conditional jump into 2.
29040 2018-01-22 David Malcolm <dmalcolm@redhat.com>
29042 PR tree-optimization/69452
29043 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
29046 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29048 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
29049 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
29050 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
29052 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29054 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
29056 * config/rl78/rl78.md (movdi): New define_expand.
29057 * config/rl78/rl78.c (rl78_split_movdi): New function.
29059 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
29062 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
29064 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
29065 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
29066 128-bit to produce an UNSPEC move to get the double word with the
29067 signbit and then a shift directly to do signbit.
29068 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
29069 implementation with a new version that just does either a direct
29070 move or a regular move. Move memory interface to separate insns.
29071 Move insns so they are next to the expander.
29072 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
29073 with signbit move. Split big and little endian case.
29074 (signbit<mode>2_dm_mem_le): Likewise.
29075 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
29076 (signbit<mode>2_dm2): Likewise.
29078 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29080 * config/rl78/rl78.md (anddi3): New define_expand.
29082 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29084 * config/rl78/rl78.md (umindi3): New define_expand.
29086 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29088 * config/rl78/rl78.md (smindi3): New define_expand.
29090 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29092 * config/rl78/rl78.md (smaxdi3): New define_expand.
29094 2018-01-22 Carl Love <cel@us.ibm.com>
29096 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
29097 LVX_V1TI): Add macro expansion.
29098 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
29099 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
29100 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
29101 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
29102 Change check to determine if the instruction is a byte reversing
29103 entry. Fix typo in comment.
29104 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
29105 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
29106 Add def_builtin calls for new builtins.
29107 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
29108 Add define_insn expansion.
29110 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29112 * config/rl78/rl78.md (umaxdi3): New define_expand.
29114 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29116 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
29117 for non-QImode registers.
29119 2018-01-22 Richard Biener <rguenther@suse.de>
29121 PR tree-optimization/83963
29122 * graphite-scop-detection.c (scop_detection::get_sese): Delay
29123 including the loop exit block.
29124 (scop_detection::merge_sese): Likewise.
29125 (scop_detection::add_scop): Do it here instead.
29127 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29129 * doc/sourcebuild.texi (arm_softfloat): Document.
29131 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
29134 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
29135 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
29136 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
29138 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29139 David Edelsohn <dje.gcc@gmail.com>
29142 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
29143 Change "crset eq" to "crset 2".
29144 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
29145 (*call_indirect_aix<mode>_nospec): Likewise.
29146 (*call_value_indirect_aix<mode>_nospec): Likewise.
29147 (*call_indirect_elfv2<mode>_nospec): Likewise.
29148 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
29149 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
29150 change assembly output from . to $.
29151 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
29152 (indirect_jump<mode>_nospec): Change assembly output from . to $.
29153 (*tablejump<mode>_internal1_nospec): Likewise.
29155 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
29158 * config/sh/sh_optimize_sett_clrt.cc:
29159 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
29161 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
29163 PR tree-optimization/83940
29164 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
29165 offset_dt to vect_constant_def rather than vect_unknown_def_type.
29166 (vect_check_load_store_mask): Add a mask_dt_out parameter and
29167 use it to pass back the definition type.
29168 (vect_check_store_rhs): Likewise rhs_dt_out.
29169 (vect_build_gather_load_calls): Add a mask_dt argument and use
29170 it instead of a call to vect_is_simple_use.
29171 (vectorizable_store): Update calls to vect_check_load_store_mask
29172 and vect_check_store_rhs. Use the dt returned by the latter instead
29173 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
29174 instead of calls to vect_is_simple_use. Pass the scalar rather
29175 than the vector operand to vect_is_simple_use when handling
29176 second and subsequent copies of an rhs value.
29177 (vectorizable_load): Update calls to vect_check_load_store_mask
29178 and vect_build_gather_load_calls. Use the cached mask_dt and
29179 gs_info.offset_dt instead of calls to vect_is_simple_use.
29181 2018-01-20 Jakub Jelinek <jakub@redhat.com>
29183 PR middle-end/83945
29184 * tree-emutls.c: Include gimplify.h.
29185 (lower_emutls_2): New function.
29186 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
29187 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
29188 it before further processing.
29191 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
29192 UINTVAL (trueop1) instead of INTVAL (op1).
29194 2018-01-19 Jakub Jelinek <jakub@redhat.com>
29198 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
29199 INCOMING_FRAME_SP_OFFSET if not defined.
29200 (scan_trace): Add ENTRY argument. If true and
29201 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
29202 emit a note to adjust the CFA offset.
29203 (create_cfi_notes): Adjust scan_trace callers.
29204 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
29205 INCOMING_FRAME_SP_OFFSET in the CIE.
29206 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
29207 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
29209 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
29210 * doc/tm.texi: Regenerated.
29212 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29214 PR rtl-optimization/83147
29215 * lra-constraints.c (remove_inheritance_pseudos): Use
29216 lra_substitute_pseudo_within_insn.
29218 2018-01-19 Tom de Vries <tom@codesourcery.com>
29219 Cesar Philippidis <cesar@codesourcery.com>
29222 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
29224 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
29227 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
29228 spaces for function labels.
29230 2018-01-19 Martin Liska <mliska@suse.cz>
29232 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
29233 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
29234 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
29235 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
29236 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
29237 (PRED_CONST_RETURN): Change from 69 to 65.
29238 (PRED_NULL_RETURN): Change from 91 to 71.
29239 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
29240 (PRED_LOOP_GUARD): Change from 66 to 73.
29242 2018-01-19 Martin Liska <mliska@suse.cz>
29244 * predict.c (predict_insn_def): Add new assert.
29245 (struct branch_predictor): Change type to signed integer.
29246 (test_prediction_value_range): Amend test to cover
29247 PROB_UNINITIALIZED.
29248 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
29249 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
29250 (PRED_LOOP_ITERATIONS_MAX): Likewise.
29251 (PRED_LOOP_IV_COMPARE): Likewise.
29252 * predict.h (PROB_UNINITIALIZED): Define new constant.
29254 2018-01-19 Martin Liska <mliska@suse.cz>
29256 * predict.c (dump_prediction): Add new format for
29257 analyze_brprob.py script which is enabled with -details
29259 * profile-count.h (precise_p): New function.
29261 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
29263 PR tree-optimization/83922
29264 * tree-vect-loop.c (vect_verify_full_masking): Return false if
29265 there are no statements that need masking.
29266 (vect_active_double_reduction_p): New function.
29267 (vect_analyze_loop_operations): Use it when handling phis that
29268 are not in the loop header.
29270 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
29272 PR tree-optimization/83914
29273 * tree-vect-loop.c (vectorizable_induction): Don't convert
29274 init_expr or apply the peeling adjustment for inductions
29275 that are nested within the vectorized loop.
29277 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29279 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
29282 2018-01-18 Jakub Jelinek <jakub@redhat.com>
29286 * function.h (gimplify_parameters): Add gimple_seq * argument.
29287 * function.c: Include gimple.h and options.h.
29288 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
29289 for the added local temporaries if needed.
29290 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
29291 if there are any parameter cleanups, wrap whole body into a
29292 try/finally with the cleanups.
29294 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
29297 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
29298 Use GET_MODE_CLASS for scalar floating point.
29300 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
29304 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
29305 Fix call of call_cgraph_insertion_hooks.
29307 2018-01-18 Martin Sebor <msebor@redhat.com>
29309 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
29311 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
29314 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
29317 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
29320 * common.opt: (-ffile-prefix-map): New option.
29321 * opts.c (common_handle_option): Defer it.
29322 * opts-global.c (handle_common_deferred_options): Handle it.
29323 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
29324 * file-prefix-map.h: New file.
29325 (remap_debug_filename, add_debug_prefix_map): ...here.
29326 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
29327 * final.c (debug_prefix_map, add_debug_prefix_map
29328 remap_debug_filename): Move to...
29329 * file-prefix-map.c: New file.
29330 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
29331 generalize, get rid of alloca(), use strrchr() instead of strchr().
29332 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
29333 Implement in terms of add_prefix_map().
29334 (remap_macro_filename, remap_debug_filename): Implement in term of
29336 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
29337 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
29338 * dbxout.c: Include file-prefix-map.h.
29339 * varasm.c: Likewise.
29340 * vmsdbgout.c: Likewise.
29341 * xcoffout.c: Likewise.
29342 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
29343 * doc/cppopts.texi (-fmacro-prefix-map): Document.
29344 * doc/invoke.texi (-ffile-prefix-map): Document.
29345 (-fdebug-prefix-map): Update description.
29347 2018-01-18 Martin Liska <mliska@suse.cz>
29349 * config/i386/i386.c (indirect_thunk_name): Document that also
29351 (output_indirect_thunk): Document why both instructions
29352 (pause and lfence) are generated.
29354 2018-01-18 Richard Biener <rguenther@suse.de>
29356 PR tree-optimization/83887
29357 * graphite-scop-detection.c
29358 (scop_detection::get_nearest_dom_with_single_entry): Remove.
29359 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
29360 (scop_detection::merge_sese): Re-implement with a flood-fill
29361 algorithm that properly finds a SESE region if it exists.
29363 2018-01-18 Jakub Jelinek <jakub@redhat.com>
29366 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
29367 pointer_diff optimizations use view_convert instead of convert.
29369 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29371 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
29372 Generate different code for -mno-speculate-indirect-jumps.
29373 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
29374 (*call_indirect_aix<mode>): Disable for
29375 -mno-speculate-indirect-jumps.
29376 (*call_indirect_aix<mode>_nospec): New define_insn.
29377 (*call_value_indirect_aix<mode>): Disable for
29378 -mno-speculate-indirect-jumps.
29379 (*call_value_indirect_aix<mode>_nospec): New define_insn.
29380 (*sibcall_nonlocal_sysv<mode>): Generate different code for
29381 -mno-speculate-indirect-jumps.
29382 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
29384 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
29386 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
29387 long double type, set the flags for noting the default long double
29388 type, even if we don't pass or return a long double type.
29390 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
29393 * ipa-inline.c (flatten_function): Do not overwrite final inlining
29396 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
29398 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
29399 support for merge[hl].
29400 (fold_mergehl_helper): New helper function.
29401 (tree-vector-builder.h): New #include for tree_vector_builder usage.
29402 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
29403 (altivec_vmrglw_direct): Add xxmrglw insn.
29405 2018-01-17 Andrew Waterman <andrew@sifive.com>
29407 * config/riscv/riscv.c (riscv_conditional_register_usage): If
29408 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
29410 2018-01-17 David Malcolm <dmalcolm@redhat.com>
29413 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
29414 call the lto_location_cache before reading the
29415 DECL_SOURCE_LOCATION of the types.
29417 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
29418 Richard Sandiford <richard.sandiford@linaro.org>
29420 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
29421 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
29422 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
29423 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
29424 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
29426 * config/aarch64/constraints.md (aarch64_movti_operand):
29428 * config/aarch64/predicates.md (Uti): Add new constraint.
29430 2018-01-17 Carl Love <cel@us.ibm.com>
29432 * config/rs6000/vsx.md (define_expand xl_len_r,
29433 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
29434 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
29436 (define_expand, define_insn): Move the shift left from the
29437 define_insn to the define_expand for lxvl and stxvl instructions.
29438 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
29439 and XL_LEN_R definitions to PURE.
29441 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
29443 * config/i386/i386.c (indirect_thunk_name): Declare regno
29444 as unsigned int. Compare regno with INVALID_REGNUM.
29445 (output_indirect_thunk): Ditto.
29446 (output_indirect_thunk_function): Ditto.
29447 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
29448 in the call to output_indirect_thunk_function.
29450 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
29452 PR middle-end/83884
29453 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
29454 rather than the size of inner_type to determine the stack slot size
29455 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
29457 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
29460 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
29463 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
29465 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
29466 endian Linux systems to optionally enable multilibs for selecting
29467 the long double type if the user configured an explicit type.
29468 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
29469 have no long double multilibs if not defined.
29470 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
29471 warn if the user used -mabi={ieee,ibm}longdouble and we built
29472 multilibs for long double.
29473 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
29474 appropriate multilib option.
29475 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
29477 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
29478 for building long double multilibs.
29479 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
29481 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
29483 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
29486 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
29488 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
29491 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
29494 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
29497 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
29499 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
29500 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
29502 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
29504 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
29505 different rtl trees depending on TARGET_64BIT.
29506 (rs6000_gen_lvx): Likewise.
29508 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
29510 * config/visium/visium.md (nop): Tweak comment.
29511 (hazard_nop): Likewise.
29513 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29515 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
29516 -mspeculate-indirect-jumps.
29517 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
29518 for -mno-speculate-indirect-jumps.
29519 (*call_indirect_elfv2<mode>_nospec): New define_insn.
29520 (*call_value_indirect_elfv2<mode>): Disable for
29521 -mno-speculate-indirect-jumps.
29522 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
29523 (indirect_jump): Emit different RTL for
29524 -mno-speculate-indirect-jumps.
29525 (*indirect_jump<mode>): Disable for
29526 -mno-speculate-indirect-jumps.
29527 (*indirect_jump<mode>_nospec): New define_insn.
29528 (tablejump): Emit different RTL for
29529 -mno-speculate-indirect-jumps.
29530 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
29531 (tablejumpsi_nospec): New define_expand.
29532 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
29533 (tablejumpdi_nospec): New define_expand.
29534 (*tablejump<mode>_internal1): Disable for
29535 -mno-speculate-indirect-jumps.
29536 (*tablejump<mode>_internal1_nospec): New define_insn.
29537 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
29540 2018-01-16 Artyom Skrobov tyomitch@gmail.com
29542 * caller-save.c (insert_save): Drop unnecessary parameter. All
29545 2018-01-16 Jakub Jelinek <jakub@redhat.com>
29546 Richard Biener <rguenth@suse.de>
29549 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
29550 return early, inline manually is_gimple_sizepos. Make sure if we
29551 call gimplify_expr we don't end up with a gimple constant.
29552 * tree.c (variably_modified_type_p): Don't return true for
29553 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
29554 * gimplify.h (is_gimple_sizepos): Remove.
29556 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
29558 PR tree-optimization/83857
29559 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
29560 vectorizable_live_operation for pure SLP statements.
29561 (vectorizable_live_operation): Handle PHIs.
29563 2018-01-16 Richard Biener <rguenther@suse.de>
29565 PR tree-optimization/83867
29566 * tree-vect-stmts.c (vect_transform_stmt): Precompute
29567 nested_in_vect_loop_p since the scalar stmt may get invalidated.
29569 2018-01-16 Jakub Jelinek <jakub@redhat.com>
29572 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
29573 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
29574 If off is not INTEGER_CST, issue a may not be aligned warning
29575 rather than isn't aligned. Use isn%'t rather than isn't.
29576 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
29578 <case MULT_EXPR>: Improve the case when bottom and one of the
29579 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
29580 operand, in that case check if the other operand is multiple of
29581 bottom divided by the INTEGER_CST operand.
29583 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
29586 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
29587 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
29588 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
29589 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
29590 * config/pa/pa.c (pa_function_arg_advance): Likewise.
29591 (pa_function_arg, pa_arg_partial_bytes): Likewise.
29592 (pa_function_arg_size): New function.
29594 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
29596 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
29597 in a separate statement.
29599 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
29601 PR tree-optimization/83847
29602 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
29603 group gathers and scatters.
29605 2018-01-16 Jakub Jelinek <jakub@redhat.com>
29607 PR rtl-optimization/86620
29608 * params.def (max-sched-ready-insns): Bump minimum value to 1.
29610 PR rtl-optimization/83213
29611 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
29612 to last if both are JUMP_INSNs.
29614 PR tree-optimization/83843
29615 * gimple-ssa-store-merging.c
29616 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
29617 store_immediate_info for bswap/nop orig_stores.
29619 2018-01-15 Andrew Waterman <andrew@sifive.com>
29621 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
29623 <UDIV>: Increase cost if !TARGET_DIV.
29625 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
29627 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
29628 (define_attr "cr_logical_3op"): New.
29629 (cceq_ior_compare): Adjust.
29630 (cceq_ior_compare_complement): Adjust.
29631 (*cceq_rev_compare): Adjust.
29632 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
29633 (is_cracked_insn): Adjust.
29634 (insn_must_be_first_in_group): Adjust.
29635 * config/rs6000/40x.md: Adjust.
29636 * config/rs6000/440.md: Adjust.
29637 * config/rs6000/476.md: Adjust.
29638 * config/rs6000/601.md: Adjust.
29639 * config/rs6000/603.md: Adjust.
29640 * config/rs6000/6xx.md: Adjust.
29641 * config/rs6000/7450.md: Adjust.
29642 * config/rs6000/7xx.md: Adjust.
29643 * config/rs6000/8540.md: Adjust.
29644 * config/rs6000/cell.md: Adjust.
29645 * config/rs6000/e300c2c3.md: Adjust.
29646 * config/rs6000/e500mc.md: Adjust.
29647 * config/rs6000/e500mc64.md: Adjust.
29648 * config/rs6000/e5500.md: Adjust.
29649 * config/rs6000/e6500.md: Adjust.
29650 * config/rs6000/mpc.md: Adjust.
29651 * config/rs6000/power4.md: Adjust.
29652 * config/rs6000/power5.md: Adjust.
29653 * config/rs6000/power6.md: Adjust.
29654 * config/rs6000/power7.md: Adjust.
29655 * config/rs6000/power8.md: Adjust.
29656 * config/rs6000/power9.md: Adjust.
29657 * config/rs6000/rs64.md: Adjust.
29658 * config/rs6000/titan.md: Adjust.
29660 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
29662 * config/i386/predicates.md (indirect_branch_operand): Rewrite
29663 ix86_indirect_branch_register logic.
29665 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
29667 * config/i386/constraints.md (Bs): Update
29668 ix86_indirect_branch_register check. Don't check
29669 ix86_indirect_branch_register with GOT_memory_operand.
29671 * config/i386/predicates.md (GOT_memory_operand): Don't check
29672 ix86_indirect_branch_register here.
29673 (GOT32_symbol_operand): Likewise.
29675 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
29677 * config/i386/predicates.md (constant_call_address_operand):
29678 Rewrite ix86_indirect_branch_register logic.
29679 (sibcall_insn_operand): Likewise.
29681 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
29683 * config/i386/constraints.md (Bs): Replace
29684 ix86_indirect_branch_thunk_register with
29685 ix86_indirect_branch_register.
29687 * config/i386/i386.md (indirect_jump): Likewise.
29688 (tablejump): Likewise.
29689 (*sibcall_memory): Likewise.
29690 (*sibcall_value_memory): Likewise.
29691 Peepholes of indirect call and jump via memory: Likewise.
29692 * config/i386/i386.opt: Likewise.
29693 * config/i386/predicates.md (indirect_branch_operand): Likewise.
29694 (GOT_memory_operand): Likewise.
29695 (call_insn_operand): Likewise.
29696 (sibcall_insn_operand): Likewise.
29697 (GOT32_symbol_operand): Likewise.
29699 2018-01-15 Jakub Jelinek <jakub@redhat.com>
29701 PR middle-end/83837
29702 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
29703 type rather than type addr's type points to.
29704 (expand_omp_atomic_mutex): Likewise.
29705 (expand_omp_atomic): Likewise.
29707 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
29710 * config/i386/i386.c (output_indirect_thunk_function): Use
29711 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
29712 for __x86_return_thunk.
29714 2018-01-15 Richard Biener <rguenther@suse.de>
29716 PR middle-end/83850
29717 * expmed.c (extract_bit_field_1): Fix typo.
29719 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29722 * config/arm/iterators.md (VF): New mode iterator.
29723 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
29724 Remove integer-related logic from pattern.
29725 (neon_vabd<mode>_3): Likewise.
29727 2018-01-15 Jakub Jelinek <jakub@redhat.com>
29729 PR middle-end/82694
29730 * common.opt (fstrict-overflow): No longer an alias.
29731 (fwrapv-pointer): New option.
29732 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
29733 also for pointer types based on flag_wrapv_pointer.
29734 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
29735 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
29736 opts->x_flag_wrapv got set.
29737 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
29738 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
29739 POINTER_TYPE_OVERFLOW_UNDEFINED.
29740 * match.pd: Likewise in address comparison pattern.
29741 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
29743 2018-01-15 Richard Biener <rguenther@suse.de>
29746 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
29747 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
29748 Reset type names to their identifier if their TYPE_DECL doesn't
29749 have linkage (and thus is used for ODR and devirt).
29750 (save_debug_info_for_decl): Remove.
29751 (save_debug_info_for_type): Likewise.
29752 (add_tree_to_fld_list): Adjust.
29753 * tree-pretty-print.c (dump_generic_node): Make dumping of
29754 type names more robust.
29756 2018-01-15 Richard Biener <rguenther@suse.de>
29758 * BASE-VER: Bump to 8.0.1.
29760 2018-01-14 Martin Sebor <msebor@redhat.com>
29763 * builtins.c (check_access): Avoid warning when the no-warning bit
29766 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
29768 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
29769 * ira-color (allocno_hard_regs_compare): Likewise.
29771 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
29774 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
29775 Use .pushsection/.popsection.
29777 2018-01-14 Martin Sebor <msebor@redhat.com>
29780 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
29782 2018-01-14 Jakub Jelinek <jakub@redhat.com>
29784 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
29785 entry from extra_headers.
29786 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
29787 extra_headers, make the list bitwise identical to the i?86-*-* one.
29789 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
29791 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
29792 -mcmodel=large with -mindirect-branch=thunk,
29793 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
29794 -mfunction-return=thunk-extern.
29795 * doc/invoke.texi: Document -mcmodel=large is incompatible with
29796 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
29797 -mfunction-return=thunk and -mfunction-return=thunk-extern.
29799 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
29801 * config/i386/i386.c (print_reg): Print the name of the full
29802 integer register without '%'.
29803 (ix86_print_operand): Handle 'V'.
29804 * doc/extend.texi: Document 'V' modifier.
29806 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
29808 * config/i386/constraints.md (Bs): Disallow memory operand for
29809 -mindirect-branch-register.
29811 * config/i386/predicates.md (indirect_branch_operand): Likewise.
29812 (GOT_memory_operand): Likewise.
29813 (call_insn_operand): Likewise.
29814 (sibcall_insn_operand): Likewise.
29815 (GOT32_symbol_operand): Likewise.
29816 * config/i386/i386.md (indirect_jump): Call convert_memory_address
29817 for -mindirect-branch-register.
29818 (tablejump): Likewise.
29819 (*sibcall_memory): Likewise.
29820 (*sibcall_value_memory): Likewise.
29821 Disallow peepholes of indirect call and jump via memory for
29822 -mindirect-branch-register.
29823 (*call_pop): Replace m with Bw.
29824 (*call_value_pop): Likewise.
29825 (*sibcall_pop_memory): Replace m with Bs.
29826 * config/i386/i386.opt (mindirect-branch-register): New option.
29827 * doc/invoke.texi: Document -mindirect-branch-register option.
29829 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
29831 * config/i386/i386-protos.h (ix86_output_function_return): New.
29832 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
29833 set function_return_type.
29834 (indirect_thunk_name): Add ret_p to indicate thunk for function
29836 (output_indirect_thunk_function): Pass false to
29837 indirect_thunk_name.
29838 (ix86_output_indirect_branch_via_reg): Likewise.
29839 (ix86_output_indirect_branch_via_push): Likewise.
29840 (output_indirect_thunk_function): Create alias for function
29841 return thunk if regno < 0.
29842 (ix86_output_function_return): New function.
29843 (ix86_handle_fndecl_attribute): Handle function_return.
29844 (ix86_attribute_table): Add function_return.
29845 * config/i386/i386.h (machine_function): Add
29846 function_return_type.
29847 * config/i386/i386.md (simple_return_internal): Use
29848 ix86_output_function_return.
29849 (simple_return_internal_long): Likewise.
29850 * config/i386/i386.opt (mfunction-return=): New option.
29851 (indirect_branch): Mention -mfunction-return=.
29852 * doc/extend.texi: Document function_return function attribute.
29853 * doc/invoke.texi: Document -mfunction-return= option.
29855 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
29857 * config/i386/i386-opts.h (indirect_branch): New.
29858 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
29859 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
29860 with local indirect jump when converting indirect call and jump.
29861 (ix86_set_indirect_branch_type): New.
29862 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
29863 (indirectlabelno): New.
29864 (indirect_thunk_needed): Likewise.
29865 (indirect_thunk_bnd_needed): Likewise.
29866 (indirect_thunks_used): Likewise.
29867 (indirect_thunks_bnd_used): Likewise.
29868 (INDIRECT_LABEL): Likewise.
29869 (indirect_thunk_name): Likewise.
29870 (output_indirect_thunk): Likewise.
29871 (output_indirect_thunk_function): Likewise.
29872 (ix86_output_indirect_branch_via_reg): Likewise.
29873 (ix86_output_indirect_branch_via_push): Likewise.
29874 (ix86_output_indirect_branch): Likewise.
29875 (ix86_output_indirect_jmp): Likewise.
29876 (ix86_code_end): Call output_indirect_thunk_function if needed.
29877 (ix86_output_call_insn): Call ix86_output_indirect_branch if
29879 (ix86_handle_fndecl_attribute): Handle indirect_branch.
29880 (ix86_attribute_table): Add indirect_branch.
29881 * config/i386/i386.h (machine_function): Add indirect_branch_type
29882 and has_local_indirect_jump.
29883 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
29885 (tablejump): Likewise.
29886 (*indirect_jump): Use ix86_output_indirect_jmp.
29887 (*tablejump_1): Likewise.
29888 (simple_return_indirect_internal): Likewise.
29889 * config/i386/i386.opt (mindirect-branch=): New option.
29890 (indirect_branch): New.
29893 (thunk-inline): Likewise.
29894 (thunk-extern): Likewise.
29895 * doc/extend.texi: Document indirect_branch function attribute.
29896 * doc/invoke.texi: Document -mindirect-branch= option.
29898 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
29901 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
29903 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
29905 * ipa-inline.c (want_inline_small_function_p): Return false if
29906 inlining has already failed with CIF_FINAL_ERROR.
29907 (update_caller_keys): Call want_inline_small_function_p before
29909 (update_callee_keys): Likewise.
29911 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29913 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
29915 (rs6000_quadword_masked_address_p): Likewise.
29916 (quad_aligned_load_p): Likewise.
29917 (quad_aligned_store_p): Likewise.
29918 (const_load_sequence_p): Add comment to describe the outer-most loop.
29919 (mimic_memory_attributes_and_flags): New function.
29920 (rs6000_gen_stvx): Likewise.
29921 (replace_swapped_aligned_store): Likewise.
29922 (rs6000_gen_lvx): Likewise.
29923 (replace_swapped_aligned_load): Likewise.
29924 (replace_swapped_load_constant): Capitalize argument name in
29925 comment describing this function.
29926 (rs6000_analyze_swaps): Add a third pass to search for vector loads
29927 and stores that access quad-word aligned addresses and replace
29928 with stvx or lvx instructions when appropriate.
29929 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
29930 New function prototype.
29931 (rs6000_quadword_masked_address_p): Likewise.
29932 (rs6000_gen_lvx): Likewise.
29933 (rs6000_gen_stvx): Likewise.
29934 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
29935 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
29936 when memory address is aligned.
29937 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
29938 this split to select lvx instruction when memory address is aligned.
29939 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
29940 instruction when memory address is aligned.
29941 (*vsx_le_perm_load_v16qi): Likewise.
29942 (four unnamed splitters): Modify to select the stvx instruction
29943 when memory is aligned.
29945 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
29947 * predict.c (determine_unlikely_bbs): Handle correctly BBs
29948 which appears in the queue multiple times.
29950 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29951 Alan Hayward <alan.hayward@arm.com>
29952 David Sherwood <david.sherwood@arm.com>
29954 * tree-vectorizer.h (vec_lower_bound): New structure.
29955 (_loop_vec_info): Add check_nonzero and lower_bounds.
29956 (LOOP_VINFO_CHECK_NONZERO): New macro.
29957 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
29958 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
29959 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
29960 fields. Make seg_len the distance travelled, not including the
29962 (dr_direction_indicator): Declare.
29963 (dr_zero_step_indicator): Likewise.
29964 (dr_known_forward_stride_p): Likewise.
29965 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
29967 (runtime_alias_check_p): Allow runtime alias checks with
29969 (operator ==): Compare access_size and align.
29970 (prune_runtime_alias_test_list): Rework for new distinction between
29971 the access_size and seg_len.
29972 (create_intersect_range_checks_index): Likewise. Cope with polynomial
29974 (get_segment_min_max): New function.
29975 (create_intersect_range_checks): Use it.
29976 (dr_step_indicator): New function.
29977 (dr_direction_indicator): Likewise.
29978 (dr_zero_step_indicator): Likewise.
29979 (dr_known_forward_stride_p): Likewise.
29980 * tree-loop-distribution.c (data_ref_segment_size): Return
29981 DR_STEP * (niters - 1).
29982 (compute_alias_check_pairs): Update call to the dr_with_seg_len
29984 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
29985 (vect_preserves_scalar_order_p): New function, split out from...
29986 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
29987 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
29988 (vect_vfa_access_size): New function.
29989 (vect_vfa_align): Likewise.
29990 (vect_compile_time_alias): Take access_size_a and access_b arguments.
29991 (dump_lower_bound): New function.
29992 (vect_check_lower_bound): Likewise.
29993 (vect_small_gap_p): Likewise.
29994 (vectorizable_with_step_bound_p): Likewise.
29995 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
29996 depencies if the vectorization factor is 1. Convert the checks
29997 for nonzero steps into checks on the bounds of DR_STEP. Try using
29998 a bunds check for variable steps if the minimum required step is
29999 relatively small. Update calls to the dr_with_seg_len
30000 constructor and to vect_compile_time_alias.
30001 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
30003 (vect_loop_versioning): Call it.
30004 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
30006 (vect_estimate_min_profitable_iters): Account for any bounds checks.
30008 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30009 Alan Hayward <alan.hayward@arm.com>
30010 David Sherwood <david.sherwood@arm.com>
30012 * doc/sourcebuild.texi (vect_scatter_store): Document.
30013 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
30015 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
30017 * genopinit.c (main): Add supports_vec_scatter_store and
30018 supports_vec_scatter_store_cached to target_optabs.
30019 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
30020 IFN_MASK_SCATTER_STORE.
30021 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
30023 * internal-fn.h (internal_store_fn_p): Declare.
30024 (internal_fn_stored_value_index): Likewise.
30025 * internal-fn.c (scatter_store_direct): New macro.
30026 (expand_scatter_store_optab_fn): New function.
30027 (direct_scatter_store_optab_supported_p): New macro.
30028 (internal_store_fn_p): New function.
30029 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
30030 IFN_MASK_SCATTER_STORE.
30031 (internal_fn_mask_index): Likewise.
30032 (internal_fn_stored_value_index): New function.
30033 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
30034 for scatter stores.
30035 * optabs-query.h (supports_vec_scatter_store_p): Declare.
30036 * optabs-query.c (supports_vec_scatter_store_p): New function.
30037 * tree-vectorizer.h (vect_get_store_rhs): Declare.
30038 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
30039 true for scatter stores.
30040 (vect_gather_scatter_fn_p): Handle scatter stores too.
30041 (vect_check_gather_scatter): Consider using scatter stores if
30042 supports_vec_scatter_store_p.
30043 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
30044 scatter stores too.
30045 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
30046 internal_fn_stored_value_index.
30047 (check_load_store_masking): Handle scatter stores too.
30048 (vect_get_store_rhs): Make public.
30049 (vectorizable_call): Use internal_store_fn_p.
30050 (vectorizable_store): Handle scatter store internal functions.
30051 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
30052 when deciding whether the end of the group has been reached.
30053 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
30054 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
30055 (mask_scatter_store<mode>): New insns.
30057 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30058 Alan Hayward <alan.hayward@arm.com>
30059 David Sherwood <david.sherwood@arm.com>
30061 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
30062 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
30063 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
30065 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
30066 Use vect_truncate_gather_scatter_offset if we can't treat the
30067 operation as a normal gather load or scatter store.
30068 (get_group_load_store_type): Take the gather_scatter_info
30069 as argument. Try using a gather load or scatter store for
30070 single-element groups.
30071 (get_load_store_type): Update calls to get_group_load_store_type
30072 and vect_use_strided_gather_scatters_p.
30074 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30075 Alan Hayward <alan.hayward@arm.com>
30076 David Sherwood <david.sherwood@arm.com>
30078 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
30079 optional tree argument.
30080 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
30082 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
30083 but continue to use the current value as a fallback.
30084 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
30085 to compare the updates.
30086 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
30087 (get_load_store_type): Use it when handling a strided access.
30088 (vect_get_strided_load_store_ops): New function.
30089 (vect_get_data_ptr_increment): Likewise.
30090 (vectorizable_load): Handle strided gather loads. Always pass
30091 a step to vect_create_data_ref_ptr and bump_vector_ptr.
30093 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30094 Alan Hayward <alan.hayward@arm.com>
30095 David Sherwood <david.sherwood@arm.com>
30097 * doc/md.texi (gather_load@var{m}): Document.
30098 (mask_gather_load@var{m}): Likewise.
30099 * genopinit.c (main): Add supports_vec_gather_load and
30100 supports_vec_gather_load_cached to target_optabs.
30101 * optabs-tree.c (init_tree_optimization_optabs): Use
30102 ggc_cleared_alloc to allocate target_optabs.
30103 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
30104 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
30106 * internal-fn.h (internal_load_fn_p): Declare.
30107 (internal_gather_scatter_fn_p): Likewise.
30108 (internal_fn_mask_index): Likewise.
30109 (internal_gather_scatter_fn_supported_p): Likewise.
30110 * internal-fn.c (gather_load_direct): New macro.
30111 (expand_gather_load_optab_fn): New function.
30112 (direct_gather_load_optab_supported_p): New macro.
30113 (direct_internal_fn_optab): New function.
30114 (internal_load_fn_p): Likewise.
30115 (internal_gather_scatter_fn_p): Likewise.
30116 (internal_fn_mask_index): Likewise.
30117 (internal_gather_scatter_fn_supported_p): Likewise.
30118 * optabs-query.c (supports_at_least_one_mode_p): New function.
30119 (supports_vec_gather_load_p): Likewise.
30120 * optabs-query.h (supports_vec_gather_load_p): Declare.
30121 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
30122 and memory_type field.
30123 (NUM_PATTERNS): Bump to 15.
30124 * tree-vect-data-refs.c: Include internal-fn.h.
30125 (vect_gather_scatter_fn_p): New function.
30126 (vect_describe_gather_scatter_call): Likewise.
30127 (vect_check_gather_scatter): Try using internal functions for
30128 gather loads. Recognize existing calls to a gather load function.
30129 (vect_analyze_data_refs): Consider using gather loads if
30130 supports_vec_gather_load_p.
30131 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
30132 (vect_get_gather_scatter_offset_type): Likewise.
30133 (vect_convert_mask_for_vectype): Likewise.
30134 (vect_add_conversion_to_patterm): Likewise.
30135 (vect_try_gather_scatter_pattern): Likewise.
30136 (vect_recog_gather_scatter_pattern): New pattern recognizer.
30137 (vect_vect_recog_func_ptrs): Add it.
30138 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
30139 internal_fn_mask_index and internal_gather_scatter_fn_p.
30140 (check_load_store_masking): Take the gather_scatter_info as an
30141 argument and handle gather loads.
30142 (vect_get_gather_scatter_ops): New function.
30143 (vectorizable_call): Check internal_load_fn_p.
30144 (vectorizable_load): Likewise. Handle gather load internal
30146 (vectorizable_store): Update call to check_load_store_masking.
30147 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
30148 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
30149 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
30150 (aarch64_gather_scale_operand_d): New predicates.
30151 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
30152 (mask_gather_load<mode>): New insns.
30154 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30155 Alan Hayward <alan.hayward@arm.com>
30156 David Sherwood <david.sherwood@arm.com>
30158 * optabs.def (fold_left_plus_optab): New optab.
30159 * doc/md.texi (fold_left_plus_@var{m}): Document.
30160 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
30161 * internal-fn.c (fold_left_direct): Define.
30162 (expand_fold_left_optab_fn): Likewise.
30163 (direct_fold_left_optab_supported_p): Likewise.
30164 * fold-const-call.c (fold_const_fold_left): New function.
30165 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
30166 * tree-parloops.c (valid_reduction_p): New function.
30167 (gather_scalar_reductions): Use it.
30168 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
30169 (vect_finish_replace_stmt): Declare.
30170 * tree-vect-loop.c (fold_left_reduction_fn): New function.
30171 (needs_fold_left_reduction_p): New function, split out from...
30172 (vect_is_simple_reduction): ...here. Accept reductions that
30173 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
30174 (vect_force_simple_reduction): Also store the reduction type in
30175 the assignment's STMT_VINFO_REDUC_TYPE.
30176 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
30177 (merge_with_identity): New function.
30178 (vect_expand_fold_left): Likewise.
30179 (vectorize_fold_left_reduction): Likewise.
30180 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
30181 scalar phi in place for it. Check for target support and reject
30182 cases that would reassociate the operation. Defer the transform
30183 phase to vectorize_fold_left_reduction.
30184 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
30185 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
30186 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
30188 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30190 * tree-if-conv.c (predicate_mem_writes): Remove redundant
30191 call to ifc_temp_var.
30193 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30194 Alan Hayward <alan.hayward@arm.com>
30195 David Sherwood <david.sherwood@arm.com>
30197 * target.def (legitimize_address_displacement): Take the original
30198 offset as a poly_int.
30199 * targhooks.h (default_legitimize_address_displacement): Update
30201 * targhooks.c (default_legitimize_address_displacement): Likewise.
30202 * doc/tm.texi: Regenerate.
30203 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
30204 as an argument, moving assert of ad->disp == ad->disp_term to...
30205 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
30206 Try calling targetm.legitimize_address_displacement before expanding
30207 the address rather than afterwards, and adjust for the new interface.
30208 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
30209 Match the new hook interface. Handle SVE addresses.
30210 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
30211 new hook interface.
30213 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30215 * Makefile.in (OBJS): Add early-remat.o.
30216 * target.def (select_early_remat_modes): New hook.
30217 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
30218 * doc/tm.texi: Regenerate.
30219 * targhooks.h (default_select_early_remat_modes): Declare.
30220 * targhooks.c (default_select_early_remat_modes): New function.
30221 * timevar.def (TV_EARLY_REMAT): New timevar.
30222 * passes.def (pass_early_remat): New pass.
30223 * tree-pass.h (make_pass_early_remat): Declare.
30224 * early-remat.c: New file.
30225 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
30227 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
30229 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30230 Alan Hayward <alan.hayward@arm.com>
30231 David Sherwood <david.sherwood@arm.com>
30233 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
30234 vfm1 with a bound_epilog parameter.
30235 (vect_do_peeling): Update calls accordingly, and move the prologue
30236 call earlier in the function. Treat the base bound_epilog as 0 for
30237 fully-masked loops and retain vf - 1 for other loops. Add 1 to
30238 this base when peeling for gaps.
30239 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
30240 with fully-masked loops.
30241 (vect_estimate_min_profitable_iters): Handle the single peeled
30242 iteration in that case.
30244 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30245 Alan Hayward <alan.hayward@arm.com>
30246 David Sherwood <david.sherwood@arm.com>
30248 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
30249 single-element interleaving even if the size is not a power of 2.
30250 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
30251 accesses for single-element interleaving if the group size is
30254 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30255 Alan Hayward <alan.hayward@arm.com>
30256 David Sherwood <david.sherwood@arm.com>
30258 * doc/md.texi (fold_extract_last_@var{m}): Document.
30259 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
30260 * optabs.def (fold_extract_last_optab): New optab.
30261 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
30262 * internal-fn.c (fold_extract_direct): New macro.
30263 (expand_fold_extract_optab_fn): Likewise.
30264 (direct_fold_extract_optab_supported_p): Likewise.
30265 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
30266 * tree-vect-loop.c (vect_model_reduction_cost): Handle
30267 EXTRACT_LAST_REDUCTION.
30268 (get_initial_def_for_reduction): Do not create an initial vector
30269 for EXTRACT_LAST_REDUCTION reductions.
30270 (vectorizable_reduction): Leave the scalar phi in place for
30271 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
30272 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
30273 epilogue code for EXTRACT_LAST_REDUCTION and defer the
30274 transform phase to vectorizable_condition.
30275 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
30277 (vect_finish_stmt_generation): ...here.
30278 (vect_finish_replace_stmt): New function.
30279 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
30280 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
30282 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
30284 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30285 Alan Hayward <alan.hayward@arm.com>
30286 David Sherwood <david.sherwood@arm.com>
30288 * doc/md.texi (extract_last_@var{m}): Document.
30289 * optabs.def (extract_last_optab): New optab.
30290 * internal-fn.def (EXTRACT_LAST): New internal function.
30291 * internal-fn.c (cond_unary_direct): New macro.
30292 (expand_cond_unary_optab_fn): Likewise.
30293 (direct_cond_unary_optab_supported_p): Likewise.
30294 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
30295 loops using EXTRACT_LAST.
30296 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
30297 (extract_last_<mode>): ...this optab.
30298 (vec_extract<mode><Vel>): Update accordingly.
30300 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30301 Alan Hayward <alan.hayward@arm.com>
30302 David Sherwood <david.sherwood@arm.com>
30304 * target.def (empty_mask_is_expensive): New hook.
30305 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
30306 * doc/tm.texi: Regenerate.
30307 * targhooks.h (default_empty_mask_is_expensive): Declare.
30308 * targhooks.c (default_empty_mask_is_expensive): New function.
30309 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
30310 if the target says that empty masks are expensive.
30311 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
30313 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
30315 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30316 Alan Hayward <alan.hayward@arm.com>
30317 David Sherwood <david.sherwood@arm.com>
30319 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
30320 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
30321 (vect_use_loop_mask_for_alignment_p): New function.
30322 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
30323 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
30324 niters_skip argument. Make sure that the first niters_skip elements
30325 of the first iteration are inactive.
30326 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
30327 Update call to vect_set_loop_masks_directly.
30328 (get_misalign_in_elems): New function, split out from...
30329 (vect_gen_prolog_loop_niters): ...here.
30330 (vect_update_init_of_dr): Take a code argument that specifies whether
30331 the adjustment should be added or subtracted.
30332 (vect_update_init_of_drs): Likewise.
30333 (vect_prepare_for_masked_peels): New function.
30334 (vect_do_peeling): Skip prologue peeling if we're using a mask
30335 instead. Update call to vect_update_inits_of_drs.
30336 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
30338 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
30339 alignment. Do not include the number of peeled iterations in
30340 the minimum threshold in that case.
30341 (vectorizable_induction): Adjust the start value down by
30342 LOOP_VINFO_MASK_SKIP_NITERS iterations.
30343 (vect_transform_loop): Call vect_prepare_for_masked_peels.
30344 Take the number of skipped iterations into account when calculating
30346 * tree-vect-stmts.c (vect_gen_while_not): New function.
30348 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30349 Alan Hayward <alan.hayward@arm.com>
30350 David Sherwood <david.sherwood@arm.com>
30352 * doc/sourcebuild.texi (vect_fully_masked): Document.
30353 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
30354 default value to 0.
30355 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
30357 (vect_analyze_loop_2): ...here. Don't check the vectorization
30358 factor against the number of loop iterations if the loop is
30361 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30362 Alan Hayward <alan.hayward@arm.com>
30363 David Sherwood <david.sherwood@arm.com>
30365 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
30366 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
30367 (dump_groups): Update accordingly.
30368 (iv_use::mem_type): New member variable.
30369 (address_p): New function.
30370 (record_use): Add a mem_type argument and initialize the new
30372 (record_group_use): Add a mem_type argument. Use address_p.
30373 Remove obsolete null checks of base_object. Update call to record_use.
30374 (find_interesting_uses_op): Update call to record_group_use.
30375 (find_interesting_uses_cond): Likewise.
30376 (find_interesting_uses_address): Likewise.
30377 (get_mem_type_for_internal_fn): New function.
30378 (find_address_like_use): Likewise.
30379 (find_interesting_uses_stmt): Try find_address_like_use before
30380 calling find_interesting_uses_op.
30381 (addr_offset_valid_p): Use the iv mem_type field as the type
30382 of the addressed memory.
30383 (add_autoinc_candidates): Likewise.
30384 (get_address_cost): Likewise.
30385 (split_small_address_groups_p): Use address_p.
30386 (split_address_groups): Likewise.
30387 (add_iv_candidate_for_use): Likewise.
30388 (autoinc_possible_for_pair): Likewise.
30389 (rewrite_groups): Likewise.
30390 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
30391 (determine_group_iv_cost): Update after split of USE_ADDRESS.
30392 (get_alias_ptr_type_for_ptr_address): New function.
30393 (rewrite_use_address): Rewrite address uses in calls that were
30394 identified by find_address_like_use.
30396 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30397 Alan Hayward <alan.hayward@arm.com>
30398 David Sherwood <david.sherwood@arm.com>
30400 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
30402 * gimple-expr.h (is_gimple_addressable: Likewise.
30403 * gimple-expr.c (is_gimple_address): Likewise.
30404 * internal-fn.c (expand_call_mem_ref): New function.
30405 (expand_mask_load_optab_fn): Use it.
30406 (expand_mask_store_optab_fn): Likewise.
30408 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30409 Alan Hayward <alan.hayward@arm.com>
30410 David Sherwood <david.sherwood@arm.com>
30412 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
30413 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
30414 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
30415 (cond_umax@var{mode}): Document.
30416 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
30417 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
30418 (cond_umin_optab, cond_umax_optab): New optabs.
30419 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
30420 (COND_IOR, COND_XOR): New internal functions.
30421 * internal-fn.h (get_conditional_internal_fn): Declare.
30422 * internal-fn.c (cond_binary_direct): New macro.
30423 (expand_cond_binary_optab_fn): Likewise.
30424 (direct_cond_binary_optab_supported_p): Likewise.
30425 (get_conditional_internal_fn): New function.
30426 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
30427 Cope with reduction statements that are vectorized as calls rather
30429 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
30430 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
30431 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
30432 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
30433 (UNSPEC_COND_EOR): New unspecs.
30434 (optab): Add mappings for them.
30435 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
30436 (sve_int_op, sve_fp_op): New int attributes.
30438 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30439 Alan Hayward <alan.hayward@arm.com>
30440 David Sherwood <david.sherwood@arm.com>
30442 * optabs.def (while_ult_optab): New optab.
30443 * doc/md.texi (while_ult@var{m}@var{n}): Document.
30444 * internal-fn.def (WHILE_ULT): New internal function.
30445 * internal-fn.h (direct_internal_fn_supported_p): New override
30446 that takes two types as argument.
30447 * internal-fn.c (while_direct): New macro.
30448 (expand_while_optab_fn): New function.
30449 (convert_optab_supported_p): Likewise.
30450 (direct_while_optab_supported_p): New macro.
30451 * wide-int.h (wi::udiv_ceil): New function.
30452 * tree-vectorizer.h (rgroup_masks): New structure.
30453 (vec_loop_masks): New typedef.
30454 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
30455 and fully_masked_p.
30456 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
30457 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
30458 (vect_max_vf): New function.
30459 (slpeel_make_loop_iterate_ntimes): Delete.
30460 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
30461 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
30462 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
30463 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
30464 internal-fn.h, stor-layout.h and optabs-query.h.
30465 (vect_set_loop_mask): New function.
30466 (add_preheader_seq): Likewise.
30467 (add_header_seq): Likewise.
30468 (interleave_supported_p): Likewise.
30469 (vect_maybe_permute_loop_masks): Likewise.
30470 (vect_set_loop_masks_directly): Likewise.
30471 (vect_set_loop_condition_masked): Likewise.
30472 (vect_set_loop_condition_unmasked): New function, split out from
30473 slpeel_make_loop_iterate_ntimes.
30474 (slpeel_make_loop_iterate_ntimes): Rename to..
30475 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
30476 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
30477 (vect_do_peeling): Update call accordingly.
30478 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
30480 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
30481 mask_compare_type, can_fully_mask_p and fully_masked_p.
30482 (release_vec_loop_masks): New function.
30483 (_loop_vec_info): Use it to free the loop masks.
30484 (can_produce_all_loop_masks_p): New function.
30485 (vect_get_max_nscalars_per_iter): Likewise.
30486 (vect_verify_full_masking): Likewise.
30487 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
30488 retries, and free the mask rgroups before retrying. Check loop-wide
30489 reasons for disallowing fully-masked loops. Make the final decision
30490 about whether use a fully-masked loop or not.
30491 (vect_estimate_min_profitable_iters): Do not assume that peeling
30492 for the number of iterations will be needed for fully-masked loops.
30493 (vectorizable_reduction): Disable fully-masked loops.
30494 (vectorizable_live_operation): Likewise.
30495 (vect_halve_mask_nunits): New function.
30496 (vect_double_mask_nunits): Likewise.
30497 (vect_record_loop_mask): Likewise.
30498 (vect_get_loop_mask): Likewise.
30499 (vect_transform_loop): Handle the case in which the final loop
30500 iteration might handle a partial vector. Call vect_set_loop_condition
30501 instead of slpeel_make_loop_iterate_ntimes.
30502 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
30503 (check_load_store_masking): New function.
30504 (prepare_load_store_mask): Likewise.
30505 (vectorizable_store): Handle fully-masked loops.
30506 (vectorizable_load): Likewise.
30507 (supportable_widening_operation): Use vect_halve_mask_nunits for
30509 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
30510 (vect_gen_while): New function.
30511 * config/aarch64/aarch64.md (umax<mode>3): New expander.
30512 (aarch64_uqdec<mode>): New insn.
30514 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30515 Alan Hayward <alan.hayward@arm.com>
30516 David Sherwood <david.sherwood@arm.com>
30518 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
30519 (reduc_xor_scal_optab): New optabs.
30520 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
30521 (reduc_xor_scal_@var{m}): Document.
30522 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
30523 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
30524 internal functions.
30525 * fold-const-call.c (fold_const_call): Handle them.
30526 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
30527 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
30528 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
30529 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
30530 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
30531 (UNSPEC_XORV): New unspecs.
30532 (optab): Add entries for them.
30533 (BITWISEV): New int iterator.
30534 (bit_reduc_op): New int attributes.
30536 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30537 Alan Hayward <alan.hayward@arm.com>
30538 David Sherwood <david.sherwood@arm.com>
30540 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
30541 * internal-fn.def (VEC_SHL_INSERT): New internal function.
30542 * optabs.def (vec_shl_insert_optab): New optab.
30543 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
30544 (duplicate_and_interleave): Likewise.
30545 * tree-vect-loop.c: Include internal-fn.h.
30546 (neutral_op_for_slp_reduction): New function, split out from
30547 get_initial_defs_for_reduction.
30548 (get_initial_def_for_reduction): Handle option 2 for variable-length
30549 vectors by loading the neutral value into a vector and then shifting
30550 the initial value into element 0.
30551 (get_initial_defs_for_reduction): Replace the code argument with
30552 the neutral value calculated by neutral_op_for_slp_reduction.
30553 Use gimple_build_vector for constant-length vectors.
30554 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
30555 but the first group_size elements have a neutral value.
30556 Use duplicate_and_interleave otherwise.
30557 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
30558 Update call to get_initial_defs_for_reduction. Handle SLP
30559 reductions for variable-length vectors by creating one vector
30560 result for each scalar result, with the elements associated
30561 with other scalar results stubbed out with the neutral value.
30562 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
30563 Require IFN_VEC_SHL_INSERT for double reductions on
30564 variable-length vectors, or SLP reductions that have
30565 a neutral value. Require can_duplicate_and_interleave_p
30566 support for variable-length unchained SLP reductions if there
30567 is no neutral value, such as for MIN/MAX reductions. Also require
30568 the number of vector elements to be a multiple of the number of
30569 SLP statements when doing variable-length unchained SLP reductions.
30570 Update call to vect_create_epilog_for_reduction.
30571 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
30572 and remove initial values.
30573 (duplicate_and_interleave): Make public.
30574 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
30575 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
30577 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30578 Alan Hayward <alan.hayward@arm.com>
30579 David Sherwood <david.sherwood@arm.com>
30581 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
30582 (can_duplicate_and_interleave_p): New function.
30583 (vect_get_and_check_slp_defs): Take the vector of statements
30584 rather than just the current one. Remove excess parentheses.
30585 Restriction rejectinon of vect_constant_def and vect_external_def
30586 for variable-length vectors to boolean types, or types for which
30587 can_duplicate_and_interleave_p is false.
30588 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
30589 (duplicate_and_interleave): New function.
30590 (vect_get_constant_vectors): Use gimple_build_vector for
30591 constant-length vectors and suitable variable-length constant
30592 vectors. Use duplicate_and_interleave for other variable-length
30593 vectors. Don't defer the update when inserting new statements.
30595 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30596 Alan Hayward <alan.hayward@arm.com>
30597 David Sherwood <david.sherwood@arm.com>
30599 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
30600 min_profitable_iters doesn't go negative.
30602 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30603 Alan Hayward <alan.hayward@arm.com>
30604 David Sherwood <david.sherwood@arm.com>
30606 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
30607 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
30608 * optabs.def (vec_mask_load_lanes_optab): New optab.
30609 (vec_mask_store_lanes_optab): Likewise.
30610 * internal-fn.def (MASK_LOAD_LANES): New internal function.
30611 (MASK_STORE_LANES): Likewise.
30612 * internal-fn.c (mask_load_lanes_direct): New macro.
30613 (mask_store_lanes_direct): Likewise.
30614 (expand_mask_load_optab_fn): Handle masked operations.
30615 (expand_mask_load_lanes_optab_fn): New macro.
30616 (expand_mask_store_optab_fn): Handle masked operations.
30617 (expand_mask_store_lanes_optab_fn): New macro.
30618 (direct_mask_load_lanes_optab_supported_p): Likewise.
30619 (direct_mask_store_lanes_optab_supported_p): Likewise.
30620 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
30622 (vect_load_lanes_supported): Likewise.
30623 * tree-vect-data-refs.c (strip_conversion): New function.
30624 (can_group_stmts_p): Likewise.
30625 (vect_analyze_data_ref_accesses): Use it instead of checking
30626 for a pair of assignments.
30627 (vect_store_lanes_supported): Take a masked_p parameter.
30628 (vect_load_lanes_supported): Likewise.
30629 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
30630 vect_store_lanes_supported and vect_load_lanes_supported.
30631 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
30632 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
30633 parameter. Don't allow gaps for masked accesses.
30634 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
30635 and vect_load_lanes_supported.
30636 (get_load_store_type): Take a masked_p parameter and update
30637 call to get_group_load_store_type.
30638 (vectorizable_store): Update call to get_load_store_type.
30639 Handle IFN_MASK_STORE_LANES.
30640 (vectorizable_load): Update call to get_load_store_type.
30641 Handle IFN_MASK_LOAD_LANES.
30643 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30644 Alan Hayward <alan.hayward@arm.com>
30645 David Sherwood <david.sherwood@arm.com>
30647 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
30649 * config/aarch64/aarch64-protos.h
30650 (aarch64_sve_struct_memory_operand_p): Declare.
30651 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
30652 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
30653 (VPRED, vpred): Handle SVE structure modes.
30654 * config/aarch64/constraints.md (Utx): New constraint.
30655 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
30656 (aarch64_sve_struct_nonimmediate_operand): New predicates.
30657 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
30658 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
30659 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
30660 structure modes. Split into pieces after RA.
30661 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
30662 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
30664 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
30665 SVE structure modes.
30666 (aarch64_classify_address): Likewise.
30667 (sizetochar): Move earlier in file.
30668 (aarch64_print_operand): Handle SVE register lists.
30669 (aarch64_array_mode): New function.
30670 (aarch64_sve_struct_memory_operand_p): Likewise.
30671 (TARGET_ARRAY_MODE): Redefine.
30673 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30674 Alan Hayward <alan.hayward@arm.com>
30675 David Sherwood <david.sherwood@arm.com>
30677 * target.def (array_mode): New target hook.
30678 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
30679 * doc/tm.texi: Regenerate.
30680 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
30681 * hooks.c (hook_optmode_mode_uhwi_none): New function.
30682 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
30683 targetm.array_mode.
30684 * stor-layout.c (mode_for_array): Likewise. Support polynomial
30687 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30688 Alan Hayward <alan.hayward@arm.com>
30689 David Sherwood <david.sherwood@arm.com>
30691 * fold-const.c (fold_binary_loc): Check the argument types
30692 rather than the result type when testing for a vector operation.
30694 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30696 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
30697 * doc/tm.texi: Regenerate.
30699 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30700 Alan Hayward <alan.hayward@arm.com>
30701 David Sherwood <david.sherwood@arm.com>
30703 * doc/invoke.texi (-msve-vector-bits=): Document new option.
30704 (sve): Document new AArch64 extension.
30705 * doc/md.texi (w): Extend the description of the AArch64
30706 constraint to include SVE vectors.
30707 (Upl, Upa): Document new AArch64 predicate constraints.
30708 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
30710 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
30711 (msve-vector-bits=): New option.
30712 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
30713 SVE when these are disabled.
30714 (sve): New extension.
30715 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
30716 modes. Adjust their number of units based on aarch64_sve_vg.
30717 (MAX_BITSIZE_MODE_ANY_MODE): Define.
30718 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
30719 aarch64_addr_query_type.
30720 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
30721 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
30722 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
30723 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
30724 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
30725 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
30726 (aarch64_simd_imm_zero_p): Delete.
30727 (aarch64_check_zero_based_sve_index_immediate): Declare.
30728 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
30729 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
30730 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
30731 (aarch64_sve_float_mul_immediate_p): Likewise.
30732 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
30733 rather than an rtx.
30734 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
30735 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
30736 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
30737 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
30738 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
30739 (aarch64_regmode_natural_size): Likewise.
30740 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
30741 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
30743 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
30744 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
30745 for VG and the SVE predicate registers.
30746 (V_ALIASES): Add a "z"-prefixed alias.
30747 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
30748 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
30749 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
30750 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
30751 (REG_CLASS_NAMES): Add entries for them.
30752 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
30753 and the predicate registers.
30754 (aarch64_sve_vg): Declare.
30755 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
30756 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
30757 (REGMODE_NATURAL_SIZE): Define.
30758 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
30760 * config/aarch64/aarch64.c: Include cfgrtl.h.
30761 (simd_immediate_info): Add a constructor for series vectors,
30762 and an associated step field.
30763 (aarch64_sve_vg): New variable.
30764 (aarch64_dbx_register_number): Handle VG and the predicate registers.
30765 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
30766 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
30767 (VEC_ANY_DATA, VEC_STRUCT): New constants.
30768 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
30769 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
30770 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
30771 (aarch64_get_mask_mode): New functions.
30772 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
30773 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
30774 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
30775 predicate modes and predicate registers. Explicitly restrict
30776 GPRs to modes of 16 bytes or smaller. Only allow FP registers
30777 to store a vector mode if it is recognized by
30778 aarch64_classify_vector_mode.
30779 (aarch64_regmode_natural_size): New function.
30780 (aarch64_hard_regno_caller_save_mode): Return the original mode
30782 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
30783 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
30784 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
30785 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
30787 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
30788 does not overlap dest if the function is frame-related. Handle
30790 (aarch64_split_add_offset): New function.
30791 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
30792 them aarch64_add_offset.
30793 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
30794 and update call to aarch64_sub_sp.
30795 (aarch64_add_cfa_expression): New function.
30796 (aarch64_expand_prologue): Pass extra temporary registers to the
30797 functions above. Handle the case in which we need to emit new
30798 DW_CFA_expressions for registers that were originally saved
30799 relative to the stack pointer, but now have to be expressed
30800 relative to the frame pointer.
30801 (aarch64_output_mi_thunk): Pass extra temporary registers to the
30803 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
30804 IP0 and IP1 values for SVE frames.
30805 (aarch64_expand_vec_series): New function.
30806 (aarch64_expand_sve_widened_duplicate): Likewise.
30807 (aarch64_expand_sve_const_vector): Likewise.
30808 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
30809 Handle SVE constants. Use emit_move_insn to move a force_const_mem
30810 into the register, rather than emitting a SET directly.
30811 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
30812 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
30813 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
30814 (offset_9bit_signed_scaled_p): New functions.
30815 (aarch64_replicate_bitmask_imm): New function.
30816 (aarch64_bitmask_imm): Use it.
30817 (aarch64_cannot_force_const_mem): Reject expressions involving
30818 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
30819 (aarch64_classify_index): Handle SVE indices, by requiring
30820 a plain register index with a scale that matches the element size.
30821 (aarch64_classify_address): Handle SVE addresses. Assert that
30822 the mode of the address is VOIDmode or an integer mode.
30823 Update call to aarch64_classify_symbol.
30824 (aarch64_classify_symbolic_expression): Update call to
30825 aarch64_classify_symbol.
30826 (aarch64_const_vec_all_in_range_p): New function.
30827 (aarch64_print_vector_float_operand): Likewise.
30828 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
30829 "vN" for FP registers with SVE modes. Handle (const ...) vectors
30830 and the FP immediates 1.0 and 0.5.
30831 (aarch64_print_address_internal): Handle SVE addresses.
30832 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
30833 (aarch64_regno_regclass): Handle predicate registers.
30834 (aarch64_secondary_reload): Handle big-endian reloads of SVE
30836 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
30837 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
30838 (aarch64_convert_sve_vector_bits): New function.
30839 (aarch64_override_options): Use it to handle -msve-vector-bits=.
30840 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
30841 rather than an rtx.
30842 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
30843 Handle SVE vector and predicate modes. Accept VL-based constants
30844 that need only one temporary register, and VL offsets that require
30845 no temporary registers.
30846 (aarch64_conditional_register_usage): Mark the predicate registers
30847 as fixed if SVE isn't available.
30848 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
30849 Return true for SVE vector and predicate modes.
30850 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
30851 rather than an unsigned int. Handle SVE modes.
30852 (aarch64_preferred_simd_mode): Update call accordingly. Handle
30854 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
30856 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
30857 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
30858 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
30859 (aarch64_sve_float_mul_immediate_p): New functions.
30860 (aarch64_sve_valid_immediate): New function.
30861 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
30862 Explicitly reject structure modes. Check for INDEX constants.
30863 Handle PTRUE and PFALSE constants.
30864 (aarch64_check_zero_based_sve_index_immediate): New function.
30865 (aarch64_simd_imm_zero_p): Delete.
30866 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
30867 vector modes. Accept constants in the range of CNT[BHWD].
30868 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
30869 ask for an Advanced SIMD mode.
30870 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
30871 (aarch64_simd_vector_alignment): Handle SVE predicates.
30872 (aarch64_vectorize_preferred_vector_alignment): New function.
30873 (aarch64_simd_vector_alignment_reachable): Use it instead of
30875 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
30876 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
30878 (MAX_VECT_LEN): Delete.
30879 (expand_vec_perm_d): Add a vec_flags field.
30880 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
30881 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
30882 (aarch64_evpc_ext): Don't apply a big-endian lane correction
30884 (aarch64_evpc_rev): Rename to...
30885 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
30886 (aarch64_evpc_rev_global): New function.
30887 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
30888 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
30890 (aarch64_evpc_sve_tbl): New function.
30891 (aarch64_expand_vec_perm_const_1): Update after rename of
30892 aarch64_evpc_rev. Handle SVE permutes too, trying
30893 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
30894 than aarch64_evpc_tbl.
30895 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
30896 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
30897 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
30898 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
30899 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
30900 (aarch64_expand_sve_vcond): New functions.
30901 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
30902 of aarch64_vector_mode_p.
30903 (aarch64_dwarf_poly_indeterminate_value): New function.
30904 (aarch64_compute_pressure_classes): Likewise.
30905 (aarch64_can_change_mode_class): Likewise.
30906 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
30907 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
30908 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
30909 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
30910 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
30911 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
30912 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
30913 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
30915 (Dn, Dl, Dr): Accept const as well as const_vector.
30916 (Dz): Likewise. Compare against CONST0_RTX.
30917 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
30918 of "vector" where appropriate.
30919 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
30920 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
30921 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
30922 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
30923 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
30924 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
30925 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
30926 (v_int_equiv): Extend to SVE modes.
30927 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
30929 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
30930 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
30931 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
30932 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
30933 (SVE_COND_FP_CMP): New int iterators.
30934 (perm_hilo): Handle the new unpack unspecs.
30935 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
30937 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
30938 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
30939 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
30940 (aarch64_equality_operator, aarch64_constant_vector_operand)
30941 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
30942 (aarch64_sve_nonimmediate_operand): Likewise.
30943 (aarch64_sve_general_operand): Likewise.
30944 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
30945 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
30946 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
30947 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
30948 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
30949 (aarch64_sve_float_arith_immediate): Likewise.
30950 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
30951 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
30952 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
30953 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
30954 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
30955 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
30956 (aarch64_sve_float_arith_operand): Likewise.
30957 (aarch64_sve_float_arith_with_sub_operand): Likewise.
30958 (aarch64_sve_float_mul_operand): Likewise.
30959 (aarch64_sve_vec_perm_operand): Likewise.
30960 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
30961 (aarch64_mov_operand): Accept const_poly_int and const_vector.
30962 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
30963 as well as const_vector.
30964 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
30965 in file. Use CONST0_RTX and CONSTM1_RTX.
30966 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
30967 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
30968 Use aarch64_simd_imm_zero.
30969 * config/aarch64/aarch64-sve.md: New file.
30970 * config/aarch64/aarch64.md: Include it.
30971 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
30972 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
30973 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
30974 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
30975 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
30976 (sve): New attribute.
30977 (enabled): Disable instructions with the sve attribute unless
30979 (movqi, movhi): Pass CONST_POLY_INT operaneds through
30980 aarch64_expand_mov_immediate.
30981 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
30982 CNT[BHSD] immediates.
30983 (movti): Split CONST_POLY_INT moves into two halves.
30984 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
30985 Split additions that need a temporary here if the destination
30986 is the stack pointer.
30987 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
30988 (*add<mode>3_poly_1): New instruction.
30989 (set_clobber_cc): New expander.
30991 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30993 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
30994 parameter and use it instead of GET_MODE_SIZE (innermode). Use
30995 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
30996 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
30997 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
30998 Change innermode from fixed_mode_size to machine_mode.
30999 (simplify_subreg): Update call accordingly. Handle a constant-sized
31000 subreg of a variable-length CONST_VECTOR.
31002 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31003 Alan Hayward <alan.hayward@arm.com>
31004 David Sherwood <david.sherwood@arm.com>
31006 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
31007 (add_offset_to_base): New function, split out from...
31008 (create_mem_ref): ...here. When handling a scale other than 1,
31009 check first whether the address is valid without the offset.
31010 Add it into the base if so, leaving the index and scale as-is.
31012 2018-01-12 Jakub Jelinek <jakub@redhat.com>
31015 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
31016 fold_for_warn before checking if arg2 is INTEGER_CST.
31018 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
31020 * config/rs6000/predicates.md (load_multiple_operation): Delete.
31021 (store_multiple_operation): Delete.
31022 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
31023 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
31024 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
31025 guarded by TARGET_STRING.
31026 (rs6000_output_load_multiple): Delete.
31027 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
31028 OPTION_MASK_STRING / TARGET_STRING handling.
31029 (print_operand) <'N', 'O'>: Add comment that these are unused now.
31030 (const rs6000_opt_masks) <"string">: Change mask to 0.
31031 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
31032 (MASK_STRING): Delete.
31033 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
31035 (load_multiple): Delete.
31042 (store_multiple): Delete.
31049 (movmemsi_8reg): Delete.
31050 (corresponding unnamed define_insn): Delete.
31051 (movmemsi_6reg): Delete.
31052 (corresponding unnamed define_insn): Delete.
31053 (movmemsi_4reg): Delete.
31054 (corresponding unnamed define_insn): Delete.
31055 (movmemsi_2reg): Delete.
31056 (corresponding unnamed define_insn): Delete.
31057 (movmemsi_1reg): Delete.
31058 (corresponding unnamed define_insn): Delete.
31059 * config/rs6000/rs6000.opt (mno-string): New.
31060 (mstring): Replace by deprecation warning stub.
31061 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
31063 2018-01-12 Jakub Jelinek <jakub@redhat.com>
31065 * regrename.c (regrename_do_replace): If replacing the same
31066 reg multiple times, try to reuse last created gen_raw_REG.
31069 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
31070 main to workaround a bug in GDB.
31072 2018-01-12 Tom de Vries <tom@codesourcery.com>
31075 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
31077 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
31079 PR rtl-optimization/80481
31080 * ira-color.c (get_cap_member): New function.
31081 (allocnos_conflict_by_live_ranges_p): Use it.
31082 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
31083 (setup_slot_coalesced_allocno_live_ranges): Ditto.
31085 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
31088 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
31089 (*saddl_se_1): Ditto.
31090 (*ssubsi_1): Ditto.
31091 (*ssubl_se_1): Ditto.
31093 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
31095 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
31096 rather than wi::to_widest for DR_INITs.
31097 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
31098 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
31099 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
31101 (vect_analyze_group_access_1): Note that here.
31103 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
31105 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
31106 polynomial type sizes.
31108 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
31110 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
31111 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
31112 (gimple_add_tmp_var): Likewise.
31114 2018-01-12 Martin Liska <mliska@suse.cz>
31116 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
31117 (gimple_alloc_sizes): Likewise.
31118 (dump_gimple_statistics): Use PRIu64 in printf format.
31119 * gimple.h: Change uint64_t to int.
31121 2018-01-12 Martin Liska <mliska@suse.cz>
31123 * tree-core.h: Use uint64_t instead of int.
31124 * tree.c (tree_node_counts): Likewise.
31125 (tree_node_sizes): Likewise.
31126 (dump_tree_statistics): Use PRIu64 in printf format.
31128 2018-01-12 Martin Liska <mliska@suse.cz>
31130 * Makefile.in: As qsort_chk is implemented in vec.c, add
31131 vec.o to linkage of gencfn-macros.
31132 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
31133 passing the info to record_node_allocation_statistics.
31134 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
31136 * ggc-common.c (struct ggc_usage): Add operator== and use
31137 it in operator< and compare function.
31138 * mem-stats.h (struct mem_usage): Likewise.
31139 * vec.c (struct vec_usage): Remove operator< and compare
31140 function. Can be simply inherited.
31142 2018-01-12 Martin Jambor <mjambor@suse.cz>
31145 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
31146 * tree-ssa-math-opts.c: Include domwalk.h.
31147 (convert_mult_to_fma_1): New function.
31148 (fma_transformation_info): New type.
31149 (fma_deferring_state): Likewise.
31150 (cancel_fma_deferring): New function.
31151 (result_of_phi): Likewise.
31152 (last_fma_candidate_feeds_initial_phi): Likewise.
31153 (convert_mult_to_fma): Added deferring logic, split actual
31154 transformation to convert_mult_to_fma_1.
31155 (math_opts_dom_walker): New type.
31156 (math_opts_dom_walker::after_dom_children): New method, body moved
31157 here from pass_optimize_widening_mul::execute, added deferring logic
31159 (pass_optimize_widening_mul::execute): Moved most of code to
31160 math_opts_dom_walker::after_dom_children.
31161 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
31162 * config/i386/i386.c (ix86_option_override_internal): Added
31163 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
31165 2018-01-12 Richard Biener <rguenther@suse.de>
31168 * dwarf2out.c (gen_variable_die): Do not reset old_die for
31169 inline instance vars.
31171 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
31174 * config/rx/rx.c (rx_is_restricted_memory_address):
31175 Handle SUBREG case.
31177 2018-01-12 Richard Biener <rguenther@suse.de>
31179 PR tree-optimization/80846
31180 * target.def (split_reduction): New target hook.
31181 * targhooks.c (default_split_reduction): New function.
31182 * targhooks.h (default_split_reduction): Declare.
31183 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
31184 target requests first reduce vectors by combining low and high
31186 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
31187 (get_vectype_for_scalar_type_and_size): Export.
31188 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
31189 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
31190 * doc/tm.texi: Regenerate.
31191 * config/i386/i386.c (ix86_split_reduction): Implement
31192 TARGET_VECTORIZE_SPLIT_REDUCTION.
31194 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
31197 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
31198 in PIC mode except for TARGET_VXWORKS_RTP.
31199 * config/sparc/sparc.c: Include cfgrtl.h.
31200 (TARGET_INIT_PIC_REG): Define.
31201 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
31202 (sparc_pic_register_p): New predicate.
31203 (sparc_legitimate_address_p): Use it.
31204 (sparc_legitimize_pic_address): Likewise.
31205 (sparc_delegitimize_address): Likewise.
31206 (sparc_mode_dependent_address_p): Likewise.
31207 (gen_load_pcrel_sym): Remove 4th parameter.
31208 (load_got_register): Adjust call to above. Remove obsolete stuff.
31209 (sparc_expand_prologue): Do not call load_got_register here.
31210 (sparc_flat_expand_prologue): Likewise.
31211 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
31212 (sparc_use_pseudo_pic_reg): New function.
31213 (sparc_init_pic_reg): Likewise.
31214 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
31215 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
31217 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
31219 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
31220 Add item for branch_cost.
31222 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
31224 PR rtl-optimization/83565
31225 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
31226 not extend the result to a larger mode for rotate operations.
31227 (num_sign_bit_copies1): Likewise.
31229 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31232 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
31234 Use values-Xc.o for -pedantic.
31235 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
31237 2018-01-12 Martin Liska <mliska@suse.cz>
31240 * ipa-devirt.c (final_warning_record::grow_type_warnings):
31242 (possible_polymorphic_call_targets): Use it.
31243 (ipa_devirt): Likewise.
31245 2018-01-12 Martin Liska <mliska@suse.cz>
31247 * profile-count.h (enum profile_quality): Use 0 as invalid
31248 enum value of profile_quality.
31250 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
31252 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
31253 -mext-string options.
31255 2018-01-12 Richard Biener <rguenther@suse.de>
31257 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
31258 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
31259 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
31261 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
31263 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
31265 * configure.ac (--with-long-double-format): Add support for the
31266 configuration option to change the default long double format on
31268 * config.gcc (powerpc*-linux*-*): Likewise.
31269 * configure: Regenerate.
31270 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
31271 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
31272 used without modification.
31274 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31276 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
31277 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
31278 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
31279 MISC_BUILTIN_SPEC_BARRIER.
31280 (rs6000_init_builtins): Likewise.
31281 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
31283 (speculation_barrier): New define_insn.
31284 * doc/extend.texi: Document __builtin_speculation_barrier.
31286 2018-01-11 Jakub Jelinek <jakub@redhat.com>
31289 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
31290 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
31291 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
31293 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
31294 integral modes instead of "ss" and "sd".
31295 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
31296 vectors with 32-bit and 64-bit elements.
31297 (vecdupssescalarmodesuffix): New mode attribute.
31298 (vec_dup<mode>): Use it.
31300 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
31303 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
31304 frame if argument is passed on stack.
31306 2018-01-11 Jakub Jelinek <jakub@redhat.com>
31309 * ree.c (combine_reaching_defs): Optimize also
31310 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
31311 reg2=any_extend(exp); reg1=reg2;, formatting fix.
31313 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
31315 PR middle-end/83189
31316 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
31318 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
31320 PR middle-end/83718
31321 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
31322 after they are computed.
31324 2018-01-11 Bin Cheng <bin.cheng@arm.com>
31326 PR tree-optimization/83695
31327 * gimple-loop-linterchange.cc
31328 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
31329 reset cached scev information after interchange.
31330 (pass_linterchange::execute): Remove call to scev_reset_htab.
31332 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31334 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
31335 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
31336 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
31337 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
31338 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
31339 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
31340 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
31341 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
31342 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
31343 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
31344 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
31345 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
31346 (V_lane_reg): Likewise.
31347 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
31349 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
31350 (vfmal_lane_low<mode>_intrinsic,
31351 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
31352 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
31353 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
31354 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
31355 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
31356 vfmsl_lane_high<mode>_intrinsic): New define_insns.
31358 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31360 * config/arm/arm-cpus.in (fp16fml): New feature.
31361 (ALL_SIMD): Add fp16fml.
31362 (armv8.2-a): Add fp16fml as an option.
31363 (armv8.3-a): Likewise.
31364 (armv8.4-a): Add fp16fml as part of fp16.
31365 * config/arm/arm.h (TARGET_FP16FML): Define.
31366 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
31368 * config/arm/arm-modes.def (V2HF): Define.
31369 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
31370 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
31371 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
31372 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
31373 vfmsl_low, vfmsl_high): New set of builtins.
31374 * config/arm/iterators.md (PLUSMINUS): New code iterator.
31375 (vfml_op): New code attribute.
31376 (VFMLHALVES): New int iterator.
31377 (VFML, VFMLSEL): New mode attributes.
31378 (V_reg): Define mapping for V2HF.
31379 (V_hi, V_lo): New mode attributes.
31380 (VF_constraint): Likewise.
31381 (vfml_half, vfml_half_selector): New int attributes.
31382 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
31384 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
31385 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
31387 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
31388 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
31389 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
31390 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
31392 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
31393 Document new effective target and option set.
31395 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31397 * config/arm/arm-cpus.in (armv8_4): New feature.
31398 (ARMv8_4a): New fgroup.
31399 (armv8.4-a): New arch.
31400 * config/arm/arm-tables.opt: Regenerate.
31401 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
31402 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
31403 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
31404 Add matching rules for -march=armv8.4-a and extensions.
31405 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
31407 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
31410 * config/rx/rx.md (BW): New mode attribute.
31411 (sync_lock_test_and_setsi): Add mode suffix to insn output.
31413 2018-01-11 Richard Biener <rguenther@suse.de>
31415 PR tree-optimization/83435
31416 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
31417 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
31418 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
31420 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
31421 Alan Hayward <alan.hayward@arm.com>
31422 David Sherwood <david.sherwood@arm.com>
31424 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
31426 (aarch64_classify_address): Initialize it. Track polynomial offsets.
31427 (aarch64_print_address_internal): Use it to check for a zero offset.
31429 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
31430 Alan Hayward <alan.hayward@arm.com>
31431 David Sherwood <david.sherwood@arm.com>
31433 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
31434 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
31435 Return a poly_int64 rather than a HOST_WIDE_INT.
31436 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
31437 rather than a HOST_WIDE_INT.
31438 * config/aarch64/aarch64.h (aarch64_frame): Protect with
31439 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
31440 hard_fp_offset, frame_size, initial_adjust, callee_offset and
31441 final_offset from HOST_WIDE_INT to poly_int64.
31442 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
31443 to_constant when getting the number of units in an Advanced SIMD
31445 (aarch64_builtin_vectorized_function): Check for a constant number
31447 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
31449 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
31450 attribute instead of GET_MODE_NUNITS.
31451 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
31452 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
31453 GET_MODE_SIZE for fixed-size registers.
31454 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
31455 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
31456 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
31457 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
31458 (aarch64_print_operand, aarch64_print_address_internal)
31459 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
31460 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
31461 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
31462 Handle polynomial GET_MODE_SIZE.
31463 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
31464 wider than SImode without modification.
31465 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
31466 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
31467 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
31468 passing and returning SVE modes.
31469 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
31470 rather than GEN_INT.
31471 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
31472 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
31473 (aarch64_allocate_and_probe_stack_space): Likewise.
31474 (aarch64_layout_frame): Cope with polynomial offsets.
31475 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
31476 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
31477 polynomial offsets.
31478 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
31479 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
31480 poly_int64 rather than a HOST_WIDE_INT.
31481 (aarch64_get_separate_components, aarch64_process_components)
31482 (aarch64_expand_prologue, aarch64_expand_epilogue)
31483 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
31484 (aarch64_anchor_offset): New function, split out from...
31485 (aarch64_legitimize_address): ...here.
31486 (aarch64_builtin_vectorization_cost): Handle polynomial
31487 TYPE_VECTOR_SUBPARTS.
31488 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
31490 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
31491 number of elements from the PARALLEL rather than the mode.
31492 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
31493 rather than GET_MODE_BITSIZE.
31494 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
31495 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
31496 (aarch64_expand_vec_perm_const_1): Handle polynomial
31497 d->perm.length () and d->perm elements.
31498 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
31499 Apply to_constant to d->perm elements.
31500 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
31501 polynomial CONST_VECTOR_NUNITS.
31502 (aarch64_move_pointer): Take amount as a poly_int64 rather
31504 (aarch64_progress_pointer): Avoid temporary variable.
31505 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
31506 the mode attribute instead of GET_MODE.
31508 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
31509 Alan Hayward <alan.hayward@arm.com>
31510 David Sherwood <david.sherwood@arm.com>
31512 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
31513 x exists before using it.
31514 (aarch64_add_constant_internal): Rename to...
31515 (aarch64_add_offset_1): ...this. Replace regnum with separate
31516 src and dest rtxes. Handle the case in which they're different,
31517 including when the offset is zero. Replace scratchreg with an rtx.
31518 Use 2 additions if there is no spare register into which we can
31519 move a 16-bit constant.
31520 (aarch64_add_constant): Delete.
31521 (aarch64_add_offset): Replace reg with separate src and dest
31522 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
31523 Use aarch64_add_offset_1.
31524 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
31525 an rtx rather than an int. Take the delta as a poly_int64
31526 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
31527 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
31528 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
31529 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
31530 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
31531 and aarch64_add_sp.
31532 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
31533 aarch64_add_constant.
31535 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
31537 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
31538 Use scalar_float_mode.
31540 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
31542 * config/aarch64/aarch64-simd.md
31543 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
31544 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
31545 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
31546 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
31547 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
31548 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
31549 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
31550 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
31551 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
31552 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
31554 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31557 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
31558 targ_options->x_arm_arch_string is non NULL.
31560 2018-01-11 Tamar Christina <tamar.christina@arm.com>
31562 * config/aarch64/aarch64.h
31563 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
31565 2018-01-11 Sudakshina Das <sudi.das@arm.com>
31568 * expmed.c (emit_store_flag_force): Swap if const op0
31569 and change VOIDmode to mode of op0.
31571 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
31573 PR rtl-optimization/83761
31574 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
31575 than bytes to mode_for_size.
31577 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
31579 PR middle-end/83189
31580 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
31581 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
31584 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
31586 PR middle-end/83575
31587 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
31588 when in layout mode.
31589 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
31590 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
31593 2018-01-10 Michael Collison <michael.collison@arm.com>
31595 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
31596 * config/aarch64/aarch64-option-extension.def: Add
31597 AARCH64_OPT_EXTENSION of 'fp16fml'.
31598 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
31599 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
31600 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
31601 * config/aarch64/constraints.md (Ui7): New constraint.
31602 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
31603 (VFMLA_SEL_W): Ditto.
31606 (VFMLA16_LOW): New int iterator.
31607 (VFMLA16_HIGH): Ditto.
31608 (UNSPEC_FMLAL): New unspec.
31609 (UNSPEC_FMLSL): Ditto.
31610 (UNSPEC_FMLAL2): Ditto.
31611 (UNSPEC_FMLSL2): Ditto.
31612 (f16mac): New code attribute.
31613 * config/aarch64/aarch64-simd-builtins.def
31614 (aarch64_fmlal_lowv2sf): Ditto.
31615 (aarch64_fmlsl_lowv2sf): Ditto.
31616 (aarch64_fmlalq_lowv4sf): Ditto.
31617 (aarch64_fmlslq_lowv4sf): Ditto.
31618 (aarch64_fmlal_highv2sf): Ditto.
31619 (aarch64_fmlsl_highv2sf): Ditto.
31620 (aarch64_fmlalq_highv4sf): Ditto.
31621 (aarch64_fmlslq_highv4sf): Ditto.
31622 (aarch64_fmlal_lane_lowv2sf): Ditto.
31623 (aarch64_fmlsl_lane_lowv2sf): Ditto.
31624 (aarch64_fmlal_laneq_lowv2sf): Ditto.
31625 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
31626 (aarch64_fmlalq_lane_lowv4sf): Ditto.
31627 (aarch64_fmlsl_lane_lowv4sf): Ditto.
31628 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
31629 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
31630 (aarch64_fmlal_lane_highv2sf): Ditto.
31631 (aarch64_fmlsl_lane_highv2sf): Ditto.
31632 (aarch64_fmlal_laneq_highv2sf): Ditto.
31633 (aarch64_fmlsl_laneq_highv2sf): Ditto.
31634 (aarch64_fmlalq_lane_highv4sf): Ditto.
31635 (aarch64_fmlsl_lane_highv4sf): Ditto.
31636 (aarch64_fmlalq_laneq_highv4sf): Ditto.
31637 (aarch64_fmlsl_laneq_highv4sf): Ditto.
31638 * config/aarch64/aarch64-simd.md:
31639 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
31640 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
31641 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
31642 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
31643 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
31644 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
31645 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
31646 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
31647 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
31648 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
31649 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
31650 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
31651 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
31652 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
31653 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
31654 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
31655 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
31656 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
31657 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
31658 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
31659 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
31660 (vfmlsl_low_u32): Ditto.
31661 (vfmlalq_low_u32): Ditto.
31662 (vfmlslq_low_u32): Ditto.
31663 (vfmlal_high_u32): Ditto.
31664 (vfmlsl_high_u32): Ditto.
31665 (vfmlalq_high_u32): Ditto.
31666 (vfmlslq_high_u32): Ditto.
31667 (vfmlal_lane_low_u32): Ditto.
31668 (vfmlsl_lane_low_u32): Ditto.
31669 (vfmlal_laneq_low_u32): Ditto.
31670 (vfmlsl_laneq_low_u32): Ditto.
31671 (vfmlalq_lane_low_u32): Ditto.
31672 (vfmlslq_lane_low_u32): Ditto.
31673 (vfmlalq_laneq_low_u32): Ditto.
31674 (vfmlslq_laneq_low_u32): Ditto.
31675 (vfmlal_lane_high_u32): Ditto.
31676 (vfmlsl_lane_high_u32): Ditto.
31677 (vfmlal_laneq_high_u32): Ditto.
31678 (vfmlsl_laneq_high_u32): Ditto.
31679 (vfmlalq_lane_high_u32): Ditto.
31680 (vfmlslq_lane_high_u32): Ditto.
31681 (vfmlalq_laneq_high_u32): Ditto.
31682 (vfmlslq_laneq_high_u32): Ditto.
31683 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
31684 (AARCH64_FL_FOR_ARCH8_4): New.
31685 (AARCH64_ISA_F16FML): New ISA flag.
31686 (TARGET_F16FML): New feature flag for fp16fml.
31687 (doc/invoke.texi): Document new fp16fml option.
31689 2018-01-10 Michael Collison <michael.collison@arm.com>
31691 * config/aarch64/aarch64-builtins.c:
31692 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
31693 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
31694 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
31695 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
31696 (AARCH64_ISA_SHA3): New ISA flag.
31697 (TARGET_SHA3): New feature flag for sha3.
31698 * config/aarch64/iterators.md (sha512_op): New int attribute.
31699 (CRYPTO_SHA512): New int iterator.
31700 (UNSPEC_SHA512H): New unspec.
31701 (UNSPEC_SHA512H2): Ditto.
31702 (UNSPEC_SHA512SU0): Ditto.
31703 (UNSPEC_SHA512SU1): Ditto.
31704 * config/aarch64/aarch64-simd-builtins.def
31705 (aarch64_crypto_sha512hqv2di): New builtin.
31706 (aarch64_crypto_sha512h2qv2di): Ditto.
31707 (aarch64_crypto_sha512su0qv2di): Ditto.
31708 (aarch64_crypto_sha512su1qv2di): Ditto.
31709 (aarch64_eor3qv8hi): Ditto.
31710 (aarch64_rax1qv2di): Ditto.
31711 (aarch64_xarqv2di): Ditto.
31712 (aarch64_bcaxqv8hi): Ditto.
31713 * config/aarch64/aarch64-simd.md:
31714 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
31715 (aarch64_crypto_sha512su0qv2di): Ditto.
31716 (aarch64_crypto_sha512su1qv2di): Ditto.
31717 (aarch64_eor3qv8hi): Ditto.
31718 (aarch64_rax1qv2di): Ditto.
31719 (aarch64_xarqv2di): Ditto.
31720 (aarch64_bcaxqv8hi): Ditto.
31721 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
31722 (vsha512h2q_u64): Ditto.
31723 (vsha512su0q_u64): Ditto.
31724 (vsha512su1q_u64): Ditto.
31725 (veor3q_u16): Ditto.
31726 (vrax1q_u64): Ditto.
31727 (vxarq_u64): Ditto.
31728 (vbcaxq_u16): Ditto.
31729 * config/arm/types.md (crypto_sha512): New type attribute.
31730 (crypto_sha3): Ditto.
31731 (doc/invoke.texi): Document new sha3 option.
31733 2018-01-10 Michael Collison <michael.collison@arm.com>
31735 * config/aarch64/aarch64-builtins.c:
31736 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
31737 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
31738 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
31739 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
31740 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
31741 (AARCH64_ISA_SM4): New ISA flag.
31742 (TARGET_SM4): New feature flag for sm4.
31743 * config/aarch64/aarch64-simd-builtins.def
31744 (aarch64_sm3ss1qv4si): Ditto.
31745 (aarch64_sm3tt1aq4si): Ditto.
31746 (aarch64_sm3tt1bq4si): Ditto.
31747 (aarch64_sm3tt2aq4si): Ditto.
31748 (aarch64_sm3tt2bq4si): Ditto.
31749 (aarch64_sm3partw1qv4si): Ditto.
31750 (aarch64_sm3partw2qv4si): Ditto.
31751 (aarch64_sm4eqv4si): Ditto.
31752 (aarch64_sm4ekeyqv4si): Ditto.
31753 * config/aarch64/aarch64-simd.md:
31754 (aarch64_sm3ss1qv4si): Ditto.
31755 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
31756 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
31757 (aarch64_sm4eqv4si): Ditto.
31758 (aarch64_sm4ekeyqv4si): Ditto.
31759 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
31760 (sm3part_op): Ditto.
31761 (CRYPTO_SM3TT): Ditto.
31762 (CRYPTO_SM3PART): Ditto.
31763 (UNSPEC_SM3SS1): New unspec.
31764 (UNSPEC_SM3TT1A): Ditto.
31765 (UNSPEC_SM3TT1B): Ditto.
31766 (UNSPEC_SM3TT2A): Ditto.
31767 (UNSPEC_SM3TT2B): Ditto.
31768 (UNSPEC_SM3PARTW1): Ditto.
31769 (UNSPEC_SM3PARTW2): Ditto.
31770 (UNSPEC_SM4E): Ditto.
31771 (UNSPEC_SM4EKEY): Ditto.
31772 * config/aarch64/constraints.md (Ui2): New constraint.
31773 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
31774 * config/arm/types.md (crypto_sm3): New type attribute.
31775 (crypto_sm4): Ditto.
31776 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
31777 (vsm3tt1aq_u32): Ditto.
31778 (vsm3tt1bq_u32): Ditto.
31779 (vsm3tt2aq_u32): Ditto.
31780 (vsm3tt2bq_u32): Ditto.
31781 (vsm3partw1q_u32): Ditto.
31782 (vsm3partw2q_u32): Ditto.
31783 (vsm4eq_u32): Ditto.
31784 (vsm4ekeyq_u32): Ditto.
31785 (doc/invoke.texi): Document new sm4 option.
31787 2018-01-10 Michael Collison <michael.collison@arm.com>
31789 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
31790 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
31791 (AARCH64_FL_FOR_ARCH8_4): New.
31792 (AARCH64_FL_V8_4): New flag.
31793 (doc/invoke.texi): Document new armv8.4-a option.
31795 2018-01-10 Michael Collison <michael.collison@arm.com>
31797 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
31798 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
31799 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
31800 * config/aarch64/aarch64-option-extension.def: Add
31801 AARCH64_OPT_EXTENSION of 'sha2'.
31802 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
31803 (crypto): Disable sha2 and aes if crypto disabled.
31804 (crypto): Enable aes and sha2 if enabled.
31805 (simd): Disable sha2 and aes if simd disabled.
31806 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
31808 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
31809 (TARGET_SHA2): New feature flag for sha2.
31810 (TARGET_AES): New feature flag for aes.
31811 * config/aarch64/aarch64-simd.md:
31812 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
31813 conditional on TARGET_AES.
31814 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
31815 (aarch64_crypto_sha1hsi): Make pattern conditional
31817 (aarch64_crypto_sha1hv4si): Ditto.
31818 (aarch64_be_crypto_sha1hv4si): Ditto.
31819 (aarch64_crypto_sha1su1v4si): Ditto.
31820 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
31821 (aarch64_crypto_sha1su0v4si): Ditto.
31822 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
31823 (aarch64_crypto_sha256su0v4si): Ditto.
31824 (aarch64_crypto_sha256su1v4si): Ditto.
31825 (doc/invoke.texi): Document new aes and sha2 options.
31827 2018-01-10 Martin Sebor <msebor@redhat.com>
31829 PR tree-optimization/83781
31830 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
31833 2018-01-11 Martin Sebor <msebor@gmail.com>
31834 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31836 PR tree-optimization/83501
31837 PR tree-optimization/81703
31839 * tree-ssa-strlen.c (get_string_cst): Rename...
31840 (get_string_len): ...to this. Handle global constants.
31841 (handle_char_store): Adjust.
31843 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
31844 Jim Wilson <jimw@sifive.com>
31846 * config/riscv/riscv-protos.h (riscv_output_return): New.
31847 * config/riscv/riscv.c (struct machine_function): New naked_p field.
31848 (riscv_attribute_table, riscv_output_return),
31849 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
31850 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
31851 (riscv_compute_frame_info): Only compute frame->mask if not a naked
31853 (riscv_expand_prologue): Add early return for naked function.
31854 (riscv_expand_epilogue): Likewise.
31855 (riscv_function_ok_for_sibcall): Return false for naked function.
31856 (riscv_set_current_function): New.
31857 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
31858 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
31859 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
31860 * doc/extend.texi (RISC-V Function Attributes): New.
31862 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
31864 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
31865 check for 128-bit long double before checking TCmode.
31866 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
31867 128-bit long doubles before checking TFmode or TCmode.
31868 (FLOAT128_IBM_P): Likewise.
31870 2018-01-10 Martin Sebor <msebor@redhat.com>
31872 PR tree-optimization/83671
31873 * builtins.c (c_strlen): Unconditionally return zero for the empty
31875 Use -Warray-bounds for warnings.
31876 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
31877 for non-constant array indices with COMPONENT_REF, arrays of
31878 arrays, and pointers to arrays.
31879 (gimple_fold_builtin_strlen): Determine and set length range for
31880 non-constant character arrays.
31882 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
31884 PR middle-end/81897
31885 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
31888 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
31890 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
31892 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
31895 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
31896 VECTOR_MEM_ALTIVEC_OR_VSX_P.
31897 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
31898 indexed_or_indirect_operand predicate.
31899 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
31900 (*vsx_le_perm_load_v8hi): Likewise.
31901 (*vsx_le_perm_load_v16qi): Likewise.
31902 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
31903 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
31904 (*vsx_le_perm_store_v8hi): Likewise.
31905 (*vsx_le_perm_store_v16qi): Likewise.
31906 (eight unnamed splitters): Likewise.
31908 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
31910 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
31911 * config/rs6000/emmintrin.h: Likewise.
31912 * config/rs6000/mmintrin.h: Likewise.
31913 * config/rs6000/xmmintrin.h: Likewise.
31915 2018-01-10 David Malcolm <dmalcolm@redhat.com>
31918 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
31920 * tree.c (tree_nop_conversion): Return true for location wrapper
31922 (maybe_wrap_with_location): New function.
31923 (selftest::check_strip_nops): New function.
31924 (selftest::test_location_wrappers): New function.
31925 (selftest::tree_c_tests): Call it.
31926 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
31927 (maybe_wrap_with_location): New decl.
31928 (EXPR_LOCATION_WRAPPER_P): New macro.
31929 (location_wrapper_p): New inline function.
31930 (tree_strip_any_location_wrapper): New inline function.
31932 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
31935 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
31936 stack_realign_offset for the largest alignment of stack slot
31938 (ix86_find_max_used_stack_alignment): New function.
31939 (ix86_finalize_stack_frame_flags): Use it. Set
31940 max_used_stack_alignment if we don't realign stack.
31941 * config/i386/i386.h (machine_function): Add
31942 max_used_stack_alignment.
31944 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
31946 * config/arm/arm.opt (-mbranch-cost): New option.
31947 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
31950 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
31953 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
31954 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
31956 2018-01-10 Richard Biener <rguenther@suse.de>
31959 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
31960 early out so it also covers the case where we have a non-NULL
31963 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
31965 PR tree-optimization/83753
31966 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
31967 for non-strided grouped accesses if the number of elements is 1.
31969 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
31972 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
31973 * i386.h (TARGET_USE_GATHER): Define.
31974 * x86-tune.def (X86_TUNE_USE_GATHER): New.
31976 2018-01-10 Martin Liska <mliska@suse.cz>
31979 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
31980 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
31982 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
31983 CLEANUP_NO_PARTITIONING is not set.
31985 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
31987 * doc/rtl.texi: Remove documentation of (const ...) wrappers
31988 for vectors, as a partial revert of r254296.
31989 * rtl.h (const_vec_p): Delete.
31990 (const_vec_duplicate_p): Don't test for vector CONSTs.
31991 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
31992 * expmed.c (make_tree): Likewise.
31995 * common.md (E, F): Use CONSTANT_P instead of checking for
31997 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
31998 checking for CONST_VECTOR.
32000 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
32002 PR middle-end/83575
32003 * predict.c (force_edge_cold): Handle in more sane way edges
32004 with no prediction.
32006 2018-01-09 Carl Love <cel@us.ibm.com>
32008 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
32010 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
32011 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
32012 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
32013 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
32014 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
32015 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
32016 * config/rs6000/rs6000-protos.h: Add extern defition for
32017 rs6000_generate_float2_double_code.
32018 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
32020 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
32021 (float2_v2df): Add define_expand.
32023 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
32026 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
32027 op_mode in the force_to_mode call.
32029 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
32031 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
32032 instead of checking each element individually.
32033 (aarch64_evpc_uzp): Likewise.
32034 (aarch64_evpc_zip): Likewise.
32035 (aarch64_evpc_ext): Likewise.
32036 (aarch64_evpc_rev): Likewise.
32037 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
32038 instead of checking each element individually. Return true without
32040 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
32041 whether all selected elements come from the same input, instead of
32042 checking each element individually. Remove calls to gen_rtx_REG,
32043 start_sequence and end_sequence and instead assert that no rtl is
32046 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
32048 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
32049 order of HIGH and CONST checks.
32051 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
32053 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
32054 if the destination isn't an SSA_NAME.
32056 2018-01-09 Richard Biener <rguenther@suse.de>
32058 PR tree-optimization/83668
32059 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
32061 (canonicalize_loop_form): ... here, renamed from ...
32062 (canonicalize_loop_closed_ssa_form): ... this and amended to
32063 swap successor edges for loop exit blocks to make us use
32064 the RPO order we need for initial schedule generation.
32066 2018-01-09 Joseph Myers <joseph@codesourcery.com>
32068 PR tree-optimization/64811
32069 * match.pd: When optimizing comparisons with Inf, avoid
32070 introducing or losing exceptions from comparisons with NaN.
32072 2018-01-09 Martin Liska <mliska@suse.cz>
32075 * asan.c (shadow_mem_size): Add gcc_assert.
32077 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
32079 Don't save registers in main().
32082 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
32083 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
32084 * config/avr/avr.c (avr_set_current_function): Don't error if
32085 naked, OS_task or OS_main are specified at the same time.
32086 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
32088 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
32090 * common/config/avr/avr-common.c (avr_option_optimization_table):
32091 Switch on -mmain-is-OS_task for optimizing compilations.
32093 2018-01-09 Richard Biener <rguenther@suse.de>
32095 PR tree-optimization/83572
32096 * graphite.c: Include cfganal.h.
32097 (graphite_transform_loops): Connect infinite loops to exit
32098 and remove fake edges at the end.
32100 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
32102 * ipa-inline.c (edge_badness): Revert accidental checkin.
32104 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
32107 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
32108 symbols; not inline clones.
32110 2018-01-09 Jakub Jelinek <jakub@redhat.com>
32113 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
32114 hard registers. Formatting fixes.
32116 PR preprocessor/83722
32117 * gcc.c (try_generate_repro): Pass
32118 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
32119 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
32122 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
32123 Kito Cheng <kito.cheng@gmail.com>
32125 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
32126 (riscv_leaf_function_p): Delete.
32127 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
32129 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32131 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
32133 (do_ifelse): New function.
32134 (do_isel): New function.
32135 (do_sub3): New function.
32136 (do_add3): New function.
32137 (do_load_mask_compare): New function.
32138 (do_overlap_load_compare): New function.
32139 (expand_compare_loop): New function.
32140 (expand_block_compare): Call expand_compare_loop() when appropriate.
32141 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
32142 option description.
32143 (-mblock-compare-inline-loop-limit): New option.
32145 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32148 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
32149 Reverse order of second and third operands in first alternative.
32150 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
32151 of first and second elements in UNSPEC_VPERMR vector.
32152 (altivec_expand_vec_perm_le): Likewise.
32154 2018-01-08 Jeff Law <law@redhat.com>
32156 PR rtl-optimizatin/81308
32157 * tree-switch-conversion.c (cfg_altered): New file scoped static.
32158 (process_switch): If group_case_labels makes a change, then set
32160 (pass_convert_switch::execute): If a switch is converted, then
32161 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
32163 PR rtl-optimization/81308
32164 * recog.c (split_all_insns): Conditionally cleanup the CFG after
32167 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
32169 PR target/83663 - Revert r255946
32170 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
32171 generation for cases where splatting a value is not useful.
32172 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
32173 across a vec_duplicate and a paradoxical subreg forming a vector
32174 mode to a vec_concat.
32176 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32178 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
32179 -march=armv8.3-a variants.
32180 * config/arm/t-multilib: Likewise.
32181 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
32183 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32185 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
32187 (cceq_ior_compare_complement): Give it a name so I can use it, and
32188 change boolean_or_operator predicate to boolean_operator so it can
32189 be used to generate a crand.
32190 (eqne): New code iterator.
32191 (bd/bd_neg): New code_attrs.
32192 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
32193 a single define_insn.
32194 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
32195 decrement (bdnzt/bdnzf/bdzt/bdzf).
32196 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
32197 with the new names of the branch decrement patterns, and added the
32198 names of the branch decrement conditional patterns.
32200 2018-01-08 Richard Biener <rguenther@suse.de>
32202 PR tree-optimization/83563
32203 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
32206 2018-01-08 Richard Biener <rguenther@suse.de>
32208 PR middle-end/83713
32209 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
32211 2018-01-08 Richard Biener <rguenther@suse.de>
32213 PR tree-optimization/83685
32214 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
32215 references to abnormals.
32217 2018-01-08 Richard Biener <rguenther@suse.de>
32220 * dwarf2out.c (output_indirect_strings): Handle empty
32221 skeleton_debug_str_hash.
32222 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
32224 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
32226 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
32227 (emit_store_direct): Likewise.
32228 (arc_trampoline_adjust_address): Likewise.
32229 (arc_asm_trampoline_template): New function.
32230 (arc_initialize_trampoline): Use asm_trampoline_template.
32231 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
32232 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
32233 * config/arc/arc.md (flush_icache): Delete pattern.
32235 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
32237 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
32238 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
32241 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
32244 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
32245 by not USED_FOR_TARGET.
32246 (make_pass_resolve_sw_modes): Likewise.
32248 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
32250 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
32253 2018-01-08 Richard Biener <rguenther@suse.de>
32255 PR middle-end/83580
32256 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
32258 2018-01-08 Richard Biener <rguenther@suse.de>
32260 PR middle-end/83517
32261 * match.pd ((t * 2) / 2) -> t): Add missing :c.
32263 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
32265 PR middle-end/81897
32266 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
32267 basic blocks with a small number of successors.
32268 (convert_control_dep_chain_into_preds): Improve handling of
32270 (dump_predicates): Split apart into...
32271 (dump_pred_chain): ...here...
32272 (dump_pred_info): ...and here.
32273 (can_one_predicate_be_invalidated_p): Add debugging printfs.
32274 (can_chain_union_be_invalidated_p): Improve check for invalidation
32276 (uninit_uses_cannot_happen): Avoid unnecessary if
32277 convert_control_dep_chain_into_preds yielded nothing.
32279 2018-01-06 Martin Sebor <msebor@redhat.com>
32281 PR tree-optimization/83640
32282 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
32283 subtracting negative offset from size.
32284 (builtin_access::overlap): Adjust offset bounds of the access to fall
32285 within the size of the object if possible.
32287 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
32289 PR rtl-optimization/83699
32290 * expmed.c (extract_bit_field_1): Restrict the vector usage of
32291 extract_bit_field_as_subreg to cases in which the extracted
32292 value is also a vector.
32294 * lra-constraints.c (process_alt_operands): Test for the equivalence
32295 substitutions when detecting a possible reload cycle.
32297 2018-01-06 Jakub Jelinek <jakub@redhat.com>
32300 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
32301 by default if flag_selective_schedling{,2}. Formatting fixes.
32303 PR rtl-optimization/83682
32304 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
32305 if it has non-VECTOR_MODE element mode.
32306 (vec_duplicate_p): Likewise.
32308 PR middle-end/83694
32309 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
32310 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
32312 2018-01-05 Jakub Jelinek <jakub@redhat.com>
32315 * config/i386/i386-builtin.def
32316 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
32317 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
32318 Require also OPTION_MASK_ISA_AVX512F in addition to
32319 OPTION_MASK_ISA_GFNI.
32320 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
32321 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
32322 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
32323 to OPTION_MASK_ISA_GFNI.
32324 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
32325 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
32326 OPTION_MASK_ISA_AVX512BW.
32327 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
32328 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
32329 addition to OPTION_MASK_ISA_GFNI.
32330 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
32331 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
32332 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
32333 to OPTION_MASK_ISA_GFNI.
32334 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
32335 a requirement for all ISAs rather than any of them with a few
32337 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
32339 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
32340 bitmasks to be enabled with 3 exceptions, instead of requiring any
32341 enabled ISA with lots of exceptions.
32342 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
32343 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
32344 Change avx512bw in isa attribute to avx512f.
32345 * config/i386/sgxintrin.h: Add license boilerplate.
32346 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
32347 to __AVX512F__ and __AVX512VL to __AVX512VL__.
32348 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
32349 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
32351 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
32352 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
32353 temporarily sse2 rather than sse if not enabled already.
32356 * config/i386/sse.md (VI248_VLBW): Rename to ...
32357 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
32358 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
32359 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
32360 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
32361 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
32362 mode iterator instead of VI248_VLBW.
32364 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
32366 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
32367 (record_modified): Skip clobbers; add debug output.
32368 (param_change_prob): Use sreal frequencies.
32370 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
32372 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
32373 punt for user-aligned variables.
32375 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
32377 * tree-chrec.c (chrec_contains_symbols): Return true for
32380 2018-01-05 Sudakshina Das <sudi.das@arm.com>
32383 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
32384 of (x|y) == x for BICS pattern.
32386 2018-01-05 Jakub Jelinek <jakub@redhat.com>
32388 PR tree-optimization/83605
32389 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
32390 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
32393 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
32395 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
32396 * config/epiphany/rtems.h: New file.
32398 2018-01-04 Jakub Jelinek <jakub@redhat.com>
32399 Uros Bizjak <ubizjak@gmail.com>
32402 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
32403 QIreg_operand instead of register_operand predicate.
32404 * config/i386/i386.c (ix86_rop_should_change_byte_p,
32405 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
32406 comments instead of -fmitigate[-_]rop.
32408 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32411 * cgraphunit.c (symbol_table::compile): Switch to text_section
32412 before calling assembly_start debug hook.
32413 * run-rtl-passes.c (run_rtl_passes): Likewise.
32416 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
32418 * tree-vrp.c (extract_range_from_binary_expr_1): Check
32419 range_int_cst_p rather than !symbolic_range_p before calling
32420 extract_range_from_multiplicative_op_1.
32422 2018-01-04 Jeff Law <law@redhat.com>
32424 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
32425 redundant test in assertion.
32427 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
32429 * doc/rtl.texi: Document machine_mode wrapper classes.
32431 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
32433 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
32434 using tree_to_uhwi.
32436 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
32438 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
32439 the VEC_PERM_EXPR fold to fail.
32441 2018-01-04 Jakub Jelinek <jakub@redhat.com>
32444 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
32445 to switched_sections.
32447 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
32450 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
32451 test for d.testing.
32453 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
32456 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
32457 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
32459 2018-01-04 Jakub Jelinek <jakub@redhat.com>
32462 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
32463 is BLKmode and bitpos not zero or mode change is needed.
32465 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
32468 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
32471 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
32474 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
32475 instead of MULT rtx. Update all corresponding splitters.
32476 (*saddl_se): Ditto.
32477 (*ssub<modesuffix>): Ditto.
32478 (*ssubl_se): Ditto.
32479 (*cmp_sadd_di): Update split patterns.
32480 (*cmp_sadd_si): Ditto.
32481 (*cmp_sadd_sidi): Ditto.
32482 (*cmp_ssub_di): Ditto.
32483 (*cmp_ssub_si): Ditto.
32484 (*cmp_ssub_sidi): Ditto.
32485 * config/alpha/predicates.md (const23_operand): New predicate.
32486 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
32487 Look for ASHIFT, not MULT inner operand.
32488 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
32490 2018-01-04 Martin Liska <mliska@suse.cz>
32492 PR gcov-profile/83669
32493 * gcov.c (output_intermediate_file): Add version to intermediate
32495 * doc/gcov.texi: Document new field 'version' in intermediate
32496 file format. Fix location of '-k' option of gcov command.
32498 2018-01-04 Martin Liska <mliska@suse.cz>
32501 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
32503 2018-01-04 Jakub Jelinek <jakub@redhat.com>
32505 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
32507 2018-01-03 Martin Sebor <msebor@redhat.com>
32509 PR tree-optimization/83655
32510 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
32511 checking calls with invalid arguments.
32513 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32515 * tree-vect-stmts.c (vect_get_store_rhs): New function.
32516 (vectorizable_mask_load_store): Delete.
32517 (vectorizable_call): Return false for masked loads and stores.
32518 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
32519 instead of gimple_assign_rhs1.
32520 (vectorizable_load): Handle IFN_MASK_LOAD.
32521 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
32523 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32525 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
32527 (vectorizable_mask_load_store): ...here.
32528 (vectorizable_load): ...and here.
32530 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32532 * tree-vect-stmts.c (vect_build_all_ones_mask)
32533 (vect_build_zero_merge_argument): New functions, split out from...
32534 (vectorizable_load): ...here.
32536 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32538 * tree-vect-stmts.c (vect_check_store_rhs): New function,
32540 (vectorizable_mask_load_store): ...here.
32541 (vectorizable_store): ...and here.
32543 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32545 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
32547 (vectorizable_mask_load_store): ...here.
32549 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32551 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
32552 (vect_model_store_cost): Take a vec_load_store_type instead of a
32554 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
32555 (vect_model_store_cost): Take a vec_load_store_type instead of a
32557 (vectorizable_mask_load_store): Update accordingly.
32558 (vectorizable_store): Likewise.
32559 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
32561 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32563 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
32564 IFN_MASK_LOAD calls here rather than...
32565 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
32567 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32568 Alan Hayward <alan.hayward@arm.com>
32569 David Sherwood <david.sherwood@arm.com>
32571 * expmed.c (extract_bit_field_1): For vector extracts,
32572 fall back to extract_bit_field_as_subreg if vec_extract
32575 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32576 Alan Hayward <alan.hayward@arm.com>
32577 David Sherwood <david.sherwood@arm.com>
32579 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
32580 they are variable or constant sized.
32581 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
32582 slots for constant-sized data.
32584 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32585 Alan Hayward <alan.hayward@arm.com>
32586 David Sherwood <david.sherwood@arm.com>
32588 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
32589 handling COND_EXPRs with boolean comparisons, try to find a better
32590 basis for the mask type than the boolean itself.
32592 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32594 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
32595 is calculated and how it can be overridden.
32596 * genmodes.c (max_bitsize_mode_any_mode): New variable.
32597 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
32599 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
32602 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32603 Alan Hayward <alan.hayward@arm.com>
32604 David Sherwood <david.sherwood@arm.com>
32606 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
32607 Remove the mode argument.
32608 (aarch64_simd_valid_immediate): Remove the mode and inverse
32610 * config/aarch64/iterators.md (bitsize): New iterator.
32611 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
32612 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
32613 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
32614 aarch64_simd_valid_immediate.
32615 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
32616 (aarch64_reg_or_bic_imm): Likewise.
32617 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
32618 with an insn_type enum and msl with a modifier_type enum.
32619 Replace element_width with a scalar_mode. Change the shift
32620 to unsigned int. Add constructors for scalar_float_mode and
32621 scalar_int_mode elements.
32622 (aarch64_vect_float_const_representable_p): Delete.
32623 (aarch64_can_const_movi_rtx_p)
32624 (aarch64_simd_scalar_immediate_valid_for_move)
32625 (aarch64_simd_make_constant): Update call to
32626 aarch64_simd_valid_immediate.
32627 (aarch64_advsimd_valid_immediate_hs): New function.
32628 (aarch64_advsimd_valid_immediate): Likewise.
32629 (aarch64_simd_valid_immediate): Remove mode and inverse
32630 arguments. Rewrite to use the above. Use const_vec_duplicate_p
32631 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
32632 and aarch64_float_const_representable_p on the result.
32633 (aarch64_output_simd_mov_immediate): Remove mode argument.
32634 Update call to aarch64_simd_valid_immediate and use of
32635 simd_immediate_info.
32636 (aarch64_output_scalar_simd_mov_immediate): Update call
32639 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32640 Alan Hayward <alan.hayward@arm.com>
32641 David Sherwood <david.sherwood@arm.com>
32643 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
32644 (mode_nunits): Likewise CONST_MODE_NUNITS.
32645 * machmode.def (ADJUST_NUNITS): Document.
32646 * genmodes.c (mode_data::need_nunits_adj): New field.
32647 (blank_mode): Update accordingly.
32648 (adj_nunits): New variable.
32649 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
32651 (emit_mode_size_inline): Set need_bytesize_adj for all modes
32652 listed in adj_nunits.
32653 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
32654 listed in adj_nunits. Don't emit case statements for such modes.
32655 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
32656 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
32657 nothing if adj_nunits is nonnull.
32658 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
32659 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
32660 (emit_mode_fbit): Update use of print_maybe_const_decl.
32661 (emit_move_size): Likewise. Treat the array as non-const
32663 (emit_mode_adjustments): Handle adj_nunits.
32665 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32667 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
32668 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
32669 (VECTOR_MODES): Use it.
32670 (make_vector_modes): Take the prefix as an argument.
32672 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32673 Alan Hayward <alan.hayward@arm.com>
32674 David Sherwood <david.sherwood@arm.com>
32676 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
32677 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
32678 for MODE_VECTOR_BOOL.
32679 * machmode.def (VECTOR_BOOL_MODE): Document.
32680 * genmodes.c (VECTOR_BOOL_MODE): New macro.
32681 (make_vector_bool_mode): New function.
32682 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
32684 * lto-streamer-in.c (lto_input_mode_table): Likewise.
32685 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
32687 * stor-layout.c (int_mode_for_mode): Likewise.
32688 * tree.c (build_vector_type_for_mode): Likewise.
32689 * varasm.c (output_constant_pool_2): Likewise.
32690 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
32691 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
32692 for MODE_VECTOR_BOOL.
32693 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
32694 of mode class checks.
32695 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
32696 instead of a list of mode class checks.
32697 (expand_vector_scalar_condition): Likewise.
32698 (type_for_widest_vector_mode): Handle BImode as an inner mode.
32700 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32701 Alan Hayward <alan.hayward@arm.com>
32702 David Sherwood <david.sherwood@arm.com>
32704 * machmode.h (mode_size): Change from unsigned short to
32706 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
32707 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
32708 or if measurement_type is not polynomial.
32709 (fixed_size_mode::includes_p): Check for constant-sized modes.
32710 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
32711 return a poly_uint16 rather than an unsigned short.
32712 (emit_mode_size): Change the type of mode_size from unsigned short
32713 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
32714 (emit_mode_adjustments): Cope with polynomial vector sizes.
32715 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
32717 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
32719 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
32720 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
32721 * caller-save.c (setup_save_areas): Likewise.
32722 (replace_reg_with_saved_mem): Likewise.
32723 * calls.c (emit_library_call_value_1): Likewise.
32724 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
32725 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
32726 (gen_lowpart_for_combine): Likewise.
32727 * convert.c (convert_to_integer_1): Likewise.
32728 * cse.c (equiv_constant, cse_insn): Likewise.
32729 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
32730 (cselib_subst_to_values): Likewise.
32731 * dce.c (word_dce_process_block): Likewise.
32732 * df-problems.c (df_word_lr_mark_ref): Likewise.
32733 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
32734 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
32735 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
32736 (rtl_for_decl_location): Likewise.
32737 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
32738 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
32739 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
32740 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
32741 (expand_expr_real_1): Likewise.
32742 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
32743 (pad_below): Likewise.
32744 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
32745 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
32746 * ira.c (get_subreg_tracking_sizes): Likewise.
32747 * ira-build.c (ira_create_allocno_objects): Likewise.
32748 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
32749 (ira_sort_regnos_for_alter_reg): Likewise.
32750 * ira-costs.c (record_operand_costs): Likewise.
32751 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
32752 (resolve_simple_move): Likewise.
32753 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
32754 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
32755 (lra_constraints): Likewise.
32756 (CONST_POOL_OK_P): Reject variable-sized modes.
32757 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
32758 (add_pseudo_to_slot, lra_spill): Likewise.
32759 * omp-low.c (omp_clause_aligned_alignment): Likewise.
32760 * optabs-query.c (get_best_extraction_insn): Likewise.
32761 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
32762 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
32763 (expand_mult_highpart, valid_multiword_target_p): Likewise.
32764 * recog.c (offsettable_address_addr_space_p): Likewise.
32765 * regcprop.c (maybe_mode_change): Likewise.
32766 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
32767 * regrename.c (build_def_use): Likewise.
32768 * regstat.c (dump_reg_info): Likewise.
32769 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
32770 (find_reloads, find_reloads_subreg_address): Likewise.
32771 * reload1.c (eliminate_regs_1): Likewise.
32772 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
32773 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
32774 (simplify_binary_operation_1, simplify_subreg): Likewise.
32775 * targhooks.c (default_function_arg_padding): Likewise.
32776 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
32777 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
32778 (verify_gimple_assign_ternary): Likewise.
32779 * tree-inline.c (estimate_move_cost): Likewise.
32780 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
32781 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
32782 (get_address_cost_ainc): Likewise.
32783 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
32784 (vect_supportable_dr_alignment): Likewise.
32785 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
32786 (vectorizable_reduction): Likewise.
32787 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
32788 (vectorizable_operation, vectorizable_load): Likewise.
32789 * tree.c (build_same_sized_truth_vector_type): Likewise.
32790 * valtrack.c (cleanup_auto_inc_dec): Likewise.
32791 * var-tracking.c (emit_note_insn_var_location): Likewise.
32792 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
32793 (ADDR_VEC_ALIGN): Likewise.
32795 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32796 Alan Hayward <alan.hayward@arm.com>
32797 David Sherwood <david.sherwood@arm.com>
32799 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
32801 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
32802 or if measurement_type is polynomial.
32803 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
32804 * combine.c (make_extraction): Likewise.
32805 * dse.c (find_shift_sequence): Likewise.
32806 * dwarf2out.c (mem_loc_descriptor): Likewise.
32807 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
32808 (extract_bit_field, extract_low_bits): Likewise.
32809 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
32810 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
32811 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
32812 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
32813 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
32814 * reload.c (find_reloads): Likewise.
32815 * reload1.c (alter_reg): Likewise.
32816 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
32817 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
32818 * tree-if-conv.c (predicate_mem_writes): Likewise.
32819 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
32820 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
32821 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
32822 * valtrack.c (dead_debug_insert_temp): Likewise.
32823 * varasm.c (mergeable_constant_section): Likewise.
32824 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
32826 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32827 Alan Hayward <alan.hayward@arm.com>
32828 David Sherwood <david.sherwood@arm.com>
32830 * expr.c (expand_assignment): Cope with polynomial mode sizes
32831 when assigning to a CONCAT.
32833 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32834 Alan Hayward <alan.hayward@arm.com>
32835 David Sherwood <david.sherwood@arm.com>
32837 * machmode.h (mode_precision): Change from unsigned short to
32839 (mode_to_precision): Return a poly_uint16 rather than an unsigned
32841 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
32842 or if measurement_type is not polynomial.
32843 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
32844 in which the mode is already known to be a scalar_int_mode.
32845 * genmodes.c (emit_mode_precision): Change the type of mode_precision
32846 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
32848 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
32849 for GET_MODE_PRECISION.
32850 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
32851 for GET_MODE_PRECISION.
32852 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
32854 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
32855 (expand_field_assignment, make_extraction): Likewise.
32856 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
32857 (get_last_value): Likewise.
32858 * convert.c (convert_to_integer_1): Likewise.
32859 * cse.c (cse_insn): Likewise.
32860 * expr.c (expand_expr_real_1): Likewise.
32861 * lra-constraints.c (simplify_operand_subreg): Likewise.
32862 * optabs-query.c (can_atomic_load_p): Likewise.
32863 * optabs.c (expand_atomic_load): Likewise.
32864 (expand_atomic_store): Likewise.
32865 * ree.c (combine_reaching_defs): Likewise.
32866 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
32867 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
32868 * tree.h (type_has_mode_precision_p): Likewise.
32869 * ubsan.c (instrument_si_overflow): Likewise.
32871 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32872 Alan Hayward <alan.hayward@arm.com>
32873 David Sherwood <david.sherwood@arm.com>
32875 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
32876 polynomial numbers of units.
32877 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
32878 (valid_vector_subparts_p): New function.
32879 (build_vector_type): Remove temporary shim and take the number
32880 of units as a poly_uint64 rather than an int.
32881 (build_opaque_vector_type): Take the number of units as a
32882 poly_uint64 rather than an int.
32883 * tree.c (build_vector_from_ctor): Handle polynomial
32884 TYPE_VECTOR_SUBPARTS.
32885 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
32886 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
32887 (build_vector_from_val): If the number of units is variable,
32888 use build_vec_duplicate_cst for constant operands and
32889 VEC_DUPLICATE_EXPR otherwise.
32890 (make_vector_type): Remove temporary is_constant ().
32891 (build_vector_type, build_opaque_vector_type): Take the number of
32892 units as a poly_uint64 rather than an int.
32893 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
32895 * cfgexpand.c (expand_debug_expr): Likewise.
32896 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
32897 (store_constructor, expand_expr_real_1): Likewise.
32898 (const_scalar_mask_from_tree): Likewise.
32899 * fold-const-call.c (fold_const_reduction): Likewise.
32900 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
32901 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
32902 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
32903 (fold_relational_const): Likewise.
32904 (native_interpret_vector): Likewise. Change the size from an
32905 int to an unsigned int.
32906 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
32907 TYPE_VECTOR_SUBPARTS.
32908 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
32909 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
32910 duplicating a non-constant operand into a variable-length vector.
32911 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
32912 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
32913 * ipa-icf.c (sem_variable::equals): Likewise.
32914 * match.pd: Likewise.
32915 * omp-simd-clone.c (simd_clone_subparts): Likewise.
32916 * print-tree.c (print_node): Likewise.
32917 * stor-layout.c (layout_type): Likewise.
32918 * targhooks.c (default_builtin_vectorization_cost): Likewise.
32919 * tree-cfg.c (verify_gimple_comparison): Likewise.
32920 (verify_gimple_assign_binary): Likewise.
32921 (verify_gimple_assign_ternary): Likewise.
32922 (verify_gimple_assign_single): Likewise.
32923 * tree-pretty-print.c (dump_generic_node): Likewise.
32924 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
32925 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
32926 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
32927 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
32928 (vect_shift_permute_load_chain): Likewise.
32929 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
32930 (expand_vector_condition, optimize_vector_constructor): Likewise.
32931 (lower_vec_perm, get_compute_type): Likewise.
32932 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
32933 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
32934 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
32935 (vect_recog_mask_conversion_pattern): Likewise.
32936 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
32937 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
32938 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
32939 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
32940 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
32941 (vectorizable_shift, vectorizable_operation, vectorizable_store)
32942 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
32943 (supportable_widening_operation): Likewise.
32944 (supportable_narrowing_operation): Likewise.
32945 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
32947 * varasm.c (output_constant): Likewise.
32949 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32950 Alan Hayward <alan.hayward@arm.com>
32951 David Sherwood <david.sherwood@arm.com>
32953 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
32954 so that both the length == 3 and length != 3 cases set up their
32955 own permute vectors. Add comments explaining why we know the
32956 number of elements is constant.
32957 (vect_permute_load_chain): Likewise.
32959 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32960 Alan Hayward <alan.hayward@arm.com>
32961 David Sherwood <david.sherwood@arm.com>
32963 * machmode.h (mode_nunits): Change from unsigned char to
32965 (ONLY_FIXED_SIZE_MODES): New macro.
32966 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
32967 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
32968 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
32970 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
32971 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
32972 or if measurement_type is not polynomial.
32973 * genmodes.c (ZERO_COEFFS): New macro.
32974 (emit_mode_nunits_inline): Make mode_nunits_inline return a
32976 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
32977 Use ZERO_COEFFS when emitting initializers.
32978 * data-streamer.h (bp_pack_poly_value): New function.
32979 (bp_unpack_poly_value): Likewise.
32980 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
32981 for GET_MODE_NUNITS.
32982 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
32983 for GET_MODE_NUNITS.
32984 * tree.c (make_vector_type): Remove temporary shim and make
32985 the real function take the number of units as a poly_uint64
32986 rather than an int.
32987 (build_vector_type_for_mode): Handle polynomial nunits.
32988 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
32989 * emit-rtl.c (const_vec_series_p_1): Likewise.
32990 (gen_rtx_CONST_VECTOR): Likewise.
32991 * fold-const.c (test_vec_duplicate_folding): Likewise.
32992 * genrecog.c (validate_pattern): Likewise.
32993 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
32994 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
32995 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
32996 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
32997 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
32998 * rtlanal.c (subreg_get_info): Likewise.
32999 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
33000 (vect_grouped_load_supported): Likewise.
33001 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
33002 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
33003 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
33004 (simplify_const_unary_operation, simplify_binary_operation_1)
33005 (simplify_const_binary_operation, simplify_ternary_operation)
33006 (test_vector_ops_duplicate, test_vector_ops): Likewise.
33007 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
33008 instead of CONST_VECTOR_NUNITS.
33009 * varasm.c (output_constant_pool_2): Likewise.
33010 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
33011 explicit-encoded elements in the XVEC for variable-length vectors.
33013 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33015 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
33017 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33018 Alan Hayward <alan.hayward@arm.com>
33019 David Sherwood <david.sherwood@arm.com>
33021 * coretypes.h (fixed_size_mode): Declare.
33022 (fixed_size_mode_pod): New typedef.
33023 * builtins.h (target_builtins::x_apply_args_mode)
33024 (target_builtins::x_apply_result_mode): Change type to
33025 fixed_size_mode_pod.
33026 * builtins.c (apply_args_size, apply_result_size, result_vector)
33027 (expand_builtin_apply_args_1, expand_builtin_apply)
33028 (expand_builtin_return): Update accordingly.
33030 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33032 * cse.c (hash_rtx_cb): Hash only the encoded elements.
33033 * cselib.c (cselib_hash_rtx): Likewise.
33034 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
33035 CONST_VECTOR encoding.
33037 2018-01-03 Jakub Jelinek <jakub@redhat.com>
33038 Jeff Law <law@redhat.com>
33041 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
33042 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
33043 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
33044 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
33047 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
33048 explicitly probe *sp in a noreturn function if there were any callee
33049 register saves or frame pointer is needed.
33051 2018-01-03 Jakub Jelinek <jakub@redhat.com>
33054 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
33055 BLKmode for ternary, binary or unary expressions.
33058 * var-tracking.c (delete_vta_debug_insn): New inline function.
33059 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
33060 insns from get_insns () to NULL instead of each bb separately.
33061 Use delete_vta_debug_insn. No longer static.
33062 (vt_debug_insns_local, variable_tracking_main_1): Adjust
33063 delete_vta_debug_insns callers.
33064 * rtl.h (delete_vta_debug_insns): Declare.
33065 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
33066 instead of variable_tracking_main.
33068 2018-01-03 Martin Sebor <msebor@redhat.com>
33070 PR tree-optimization/83603
33071 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
33072 arguments past the endof the argument list in functions declared
33073 without a prototype.
33074 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
33075 Avoid checking when arguments are null.
33077 2018-01-03 Martin Sebor <msebor@redhat.com>
33080 * doc/extend.texi (attribute const): Fix a typo.
33081 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
33082 issuing -Wsuggest-attribute for void functions.
33084 2018-01-03 Martin Sebor <msebor@redhat.com>
33086 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
33087 offset_int::from instead of wide_int::to_shwi.
33088 (maybe_diag_overlap): Remove assertion.
33089 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
33090 * gimple-ssa-sprintf.c (format_directive): Same.
33091 (parse_directive): Same.
33092 (sprintf_dom_walker::compute_format_length): Same.
33093 (try_substitute_return_value): Same.
33095 2018-01-03 Jeff Law <law@redhat.com>
33097 PR middle-end/83654
33098 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
33099 non-constant residual for zero at runtime and avoid probing in
33100 that case. Reorganize code for trailing problem to mirror handling
33103 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33105 PR tree-optimization/83501
33106 * tree-ssa-strlen.c (get_string_cst): New.
33107 (handle_char_store): Call get_string_cst.
33109 2018-01-03 Martin Liska <mliska@suse.cz>
33111 PR tree-optimization/83593
33112 * tree-ssa-strlen.c: Include tree-cfg.h.
33113 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
33114 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
33115 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
33117 (strlen_dom_walker::before_dom_children): Call
33118 gimple_purge_dead_eh_edges. Dump tranformation with details
33120 (strlen_dom_walker::before_dom_children): Update call by adding
33121 new argument cleanup_eh.
33122 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
33124 2018-01-03 Martin Liska <mliska@suse.cz>
33127 * cif-code.def (VARIADIC_THUNK): New enum value.
33128 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
33131 2018-01-03 Jan Beulich <jbeulich@suse.com>
33133 * sse.md (mov<mode>_internal): Tighten condition for when to use
33134 vmovdqu<ssescalarsize> for TI and OI modes.
33136 2018-01-03 Jakub Jelinek <jakub@redhat.com>
33138 Update copyright years.
33140 2018-01-03 Martin Liska <mliska@suse.cz>
33143 * ipa-visibility.c (function_and_variable_visibility): Skip
33144 functions with noipa attribure.
33146 2018-01-03 Jakub Jelinek <jakub@redhat.com>
33148 * gcc.c (process_command): Update copyright notice dates.
33149 * gcov-dump.c (print_version): Ditto.
33150 * gcov.c (print_version): Ditto.
33151 * gcov-tool.c (print_version): Ditto.
33152 * gengtype.c (create_file): Ditto.
33153 * doc/cpp.texi: Bump @copying's copyright year.
33154 * doc/cppinternals.texi: Ditto.
33155 * doc/gcc.texi: Ditto.
33156 * doc/gccint.texi: Ditto.
33157 * doc/gcov.texi: Ditto.
33158 * doc/install.texi: Ditto.
33159 * doc/invoke.texi: Ditto.
33161 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33163 * vector-builder.h (vector_builder::m_full_nelts): Change from
33164 unsigned int to poly_uint64.
33165 (vector_builder::full_nelts): Update prototype accordingly.
33166 (vector_builder::new_vector): Likewise.
33167 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
33168 (vector_builder::operator ==): Likewise.
33169 (vector_builder::finalize): Likewise.
33170 * int-vector-builder.h (int_vector_builder::int_vector_builder):
33171 Take the number of elements as a poly_uint64 rather than an
33173 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
33174 from unsigned int to poly_uint64.
33175 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
33176 (vec_perm_indices::new_vector): Likewise.
33177 (vec_perm_indices::length): Likewise.
33178 (vec_perm_indices::nelts_per_input): Likewise.
33179 (vec_perm_indices::input_nelts): Likewise.
33180 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
33181 number of elements per input as a poly_uint64 rather than an
33182 unsigned int. Use the original encoding for variable-length
33183 vectors, rather than clamping each individual element.
33184 For the second and subsequent elements in each pattern,
33185 clamp the step and base before clamping their sum.
33186 (vec_perm_indices::series_p): Handle polynomial element counts.
33187 (vec_perm_indices::all_in_range_p): Likewise.
33188 (vec_perm_indices_to_tree): Likewise.
33189 (vec_perm_indices_to_rtx): Likewise.
33190 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
33191 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
33192 (tree_vector_builder::new_binary_operation): Handle polynomial
33193 element counts. Return false if we need to know the number
33194 of elements at compile time.
33195 * fold-const.c (fold_vec_perm): Punt if the number of elements
33196 isn't known at compile time.
33198 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33200 * vec-perm-indices.h (vec_perm_builder): Change element type
33201 from HOST_WIDE_INT to poly_int64.
33202 (vec_perm_indices::element_type): Update accordingly.
33203 (vec_perm_indices::clamp): Handle polynomial element_types.
33204 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
33205 (vec_perm_indices::all_in_range_p): Likewise.
33206 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
33208 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
33209 polynomial vec_perm_indices element types.
33210 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
33211 * fold-const.c (fold_vec_perm): Likewise.
33212 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
33213 * tree-vect-generic.c (lower_vec_perm): Likewise.
33214 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
33215 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
33216 element type to HOST_WIDE_INT.
33218 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33219 Alan Hayward <alan.hayward@arm.com>
33220 David Sherwood <david.sherwood@arm.com>
33222 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
33223 rather than an int. Use plus_constant.
33224 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
33225 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
33227 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33228 Alan Hayward <alan.hayward@arm.com>
33229 David Sherwood <david.sherwood@arm.com>
33231 * calls.c (emit_call_1, expand_call): Change struct_value_size from
33232 a HOST_WIDE_INT to a poly_int64.
33234 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33235 Alan Hayward <alan.hayward@arm.com>
33236 David Sherwood <david.sherwood@arm.com>
33238 * calls.c (load_register_parameters): Cope with polynomial
33239 mode sizes. Require a constant size for BLKmode parameters
33240 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
33241 forces a parameter to be padded at the lsb end in order to
33242 fill a complete number of words, require the parameter size
33243 to be ordered wrt UNITS_PER_WORD.
33245 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33246 Alan Hayward <alan.hayward@arm.com>
33247 David Sherwood <david.sherwood@arm.com>
33249 * reload1.c (spill_stack_slot_width): Change element type
33250 from unsigned int to poly_uint64_pod.
33251 (alter_reg): Treat mode sizes as polynomial.
33253 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33254 Alan Hayward <alan.hayward@arm.com>
33255 David Sherwood <david.sherwood@arm.com>
33257 * reload.c (complex_word_subreg_p): New function.
33258 (reload_inner_reg_of_subreg, push_reload): Use it.
33260 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33261 Alan Hayward <alan.hayward@arm.com>
33262 David Sherwood <david.sherwood@arm.com>
33264 * lra-constraints.c (process_alt_operands): Reject matched
33265 operands whose sizes aren't ordered.
33266 (match_reload): Refer to this check here.
33268 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33269 Alan Hayward <alan.hayward@arm.com>
33270 David Sherwood <david.sherwood@arm.com>
33272 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
33273 that the mode size is in the set {1, 2, 4, 8, 16}.
33275 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33276 Alan Hayward <alan.hayward@arm.com>
33277 David Sherwood <david.sherwood@arm.com>
33279 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
33280 Use plus_constant instead of gen_rtx_PLUS.
33282 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33283 Alan Hayward <alan.hayward@arm.com>
33284 David Sherwood <david.sherwood@arm.com>
33286 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
33287 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
33288 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
33289 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
33290 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
33291 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
33292 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
33293 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
33294 * config/i386/i386.c (ix86_push_rounding): ...this new function.
33295 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
33297 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
33298 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
33299 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
33300 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
33301 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
33302 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
33303 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
33304 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
33305 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
33306 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
33308 * expr.c (emit_move_resolve_push): Treat the input and result
33309 of PUSH_ROUNDING as a poly_int64.
33310 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
33311 (emit_push_insn): Likewise.
33312 * lra-eliminations.c (mark_not_eliminable): Likewise.
33313 * recog.c (push_operand): Likewise.
33314 * reload1.c (elimination_effects): Likewise.
33315 * rtlanal.c (nonzero_bits1): Likewise.
33316 * calls.c (store_one_arg): Likewise. Require the padding to be
33317 known at compile time.
33319 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33320 Alan Hayward <alan.hayward@arm.com>
33321 David Sherwood <david.sherwood@arm.com>
33323 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
33324 Use plus_constant instead of gen_rtx_PLUS.
33326 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33327 Alan Hayward <alan.hayward@arm.com>
33328 David Sherwood <david.sherwood@arm.com>
33330 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
33331 rather than an int.
33333 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33334 Alan Hayward <alan.hayward@arm.com>
33335 David Sherwood <david.sherwood@arm.com>
33337 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
33338 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
33339 via stack temporaries. Treat the mode size as polynomial too.
33341 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33342 Alan Hayward <alan.hayward@arm.com>
33343 David Sherwood <david.sherwood@arm.com>
33345 * expr.c (expand_expr_real_2): When handling conversions involving
33346 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
33347 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
33348 as a poly_uint64 too.
33350 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33351 Alan Hayward <alan.hayward@arm.com>
33352 David Sherwood <david.sherwood@arm.com>
33354 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
33356 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33357 Alan Hayward <alan.hayward@arm.com>
33358 David Sherwood <david.sherwood@arm.com>
33360 * combine.c (can_change_dest_mode): Handle polynomial
33361 REGMODE_NATURAL_SIZE.
33362 * expmed.c (store_bit_field_1): Likewise.
33363 * expr.c (store_constructor): Likewise.
33364 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
33365 and polynomial REGMODE_NATURAL_SIZE.
33366 (gen_lowpart_common): Likewise.
33367 * reginfo.c (record_subregs_of_mode): Likewise.
33368 * rtlanal.c (read_modify_subreg_p): Likewise.
33370 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33371 Alan Hayward <alan.hayward@arm.com>
33372 David Sherwood <david.sherwood@arm.com>
33374 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
33375 numbers of elements.
33377 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33378 Alan Hayward <alan.hayward@arm.com>
33379 David Sherwood <david.sherwood@arm.com>
33381 * match.pd: Cope with polynomial numbers of vector elements.
33383 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33384 Alan Hayward <alan.hayward@arm.com>
33385 David Sherwood <david.sherwood@arm.com>
33387 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
33388 in a POINTER_PLUS_EXPR.
33390 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33391 Alan Hayward <alan.hayward@arm.com>
33392 David Sherwood <david.sherwood@arm.com>
33394 * omp-simd-clone.c (simd_clone_subparts): New function.
33395 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
33396 (ipa_simd_modify_function_body): Likewise.
33398 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33399 Alan Hayward <alan.hayward@arm.com>
33400 David Sherwood <david.sherwood@arm.com>
33402 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
33403 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
33404 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
33405 (expand_vector_condition, vector_element): Likewise.
33406 (subparts_gt): New function.
33407 (get_compute_type): Use subparts_gt.
33408 (count_type_subparts): Delete.
33409 (expand_vector_operations_1): Use subparts_gt instead of
33410 count_type_subparts.
33412 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33413 Alan Hayward <alan.hayward@arm.com>
33414 David Sherwood <david.sherwood@arm.com>
33416 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
33417 (vect_compile_time_alias): ...this new function. Do the calculation
33418 on poly_ints rather than trees.
33419 (vect_prune_runtime_alias_test_list): Update call accordingly.
33421 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33422 Alan Hayward <alan.hayward@arm.com>
33423 David Sherwood <david.sherwood@arm.com>
33425 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
33427 (vect_schedule_slp_instance): Likewise.
33429 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33430 Alan Hayward <alan.hayward@arm.com>
33431 David Sherwood <david.sherwood@arm.com>
33433 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
33434 constant and extern definitions for variable-length vectors.
33435 (vect_get_constant_vectors): Note that the number of units
33436 is known to be constant.
33438 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33439 Alan Hayward <alan.hayward@arm.com>
33440 David Sherwood <david.sherwood@arm.com>
33442 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
33443 of units as polynomial. Choose between WIDE and NARROW based
33446 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33447 Alan Hayward <alan.hayward@arm.com>
33448 David Sherwood <david.sherwood@arm.com>
33450 * tree-vect-stmts.c (simd_clone_subparts): New function.
33451 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
33453 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33454 Alan Hayward <alan.hayward@arm.com>
33455 David Sherwood <david.sherwood@arm.com>
33457 * tree-vect-stmts.c (vectorizable_call): Treat the number of
33458 vectors as polynomial. Use build_index_vector for
33459 IFN_GOMP_SIMD_LANE.
33461 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33462 Alan Hayward <alan.hayward@arm.com>
33463 David Sherwood <david.sherwood@arm.com>
33465 * tree-vect-stmts.c (get_load_store_type): Treat the number of
33466 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
33467 for variable-length vectors.
33468 (vectorizable_mask_load_store): Treat the number of units as
33469 polynomial, asserting that it is constant if the condition has
33470 already been enforced.
33471 (vectorizable_store, vectorizable_load): Likewise.
33473 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33474 Alan Hayward <alan.hayward@arm.com>
33475 David Sherwood <david.sherwood@arm.com>
33477 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
33478 of units as polynomial. Punt if we can't tell at compile time
33479 which vector contains the final result.
33481 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33482 Alan Hayward <alan.hayward@arm.com>
33483 David Sherwood <david.sherwood@arm.com>
33485 * tree-vect-loop.c (vectorizable_induction): Treat the number
33486 of units as polynomial. Punt on SLP inductions. Use an integer
33487 VEC_SERIES_EXPR for variable-length integer reductions. Use a
33488 cast of such a series for variable-length floating-point
33491 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33492 Alan Hayward <alan.hayward@arm.com>
33493 David Sherwood <david.sherwood@arm.com>
33495 * tree.h (build_index_vector): Declare.
33496 * tree.c (build_index_vector): New function.
33497 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
33498 of units as polynomial, forcibly converting it to a constant if
33499 vectorizable_reduction has already enforced the condition.
33500 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
33501 to create a {1,2,3,...} vector.
33502 (vectorizable_reduction): Treat the number of units as polynomial.
33503 Choose vectype_in based on the largest scalar element size rather
33504 than the smallest number of units. Enforce the restrictions
33507 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33508 Alan Hayward <alan.hayward@arm.com>
33509 David Sherwood <david.sherwood@arm.com>
33511 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
33512 number of units as polynomial.
33514 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33515 Alan Hayward <alan.hayward@arm.com>
33516 David Sherwood <david.sherwood@arm.com>
33518 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
33519 * target.def (autovectorize_vector_sizes): Return the vector sizes
33520 by pointer, using vector_sizes rather than a bitmask.
33521 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
33522 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
33523 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
33525 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
33526 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
33527 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
33528 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
33529 * omp-general.c (omp_max_vf): Likewise.
33530 * omp-low.c (omp_clause_aligned_alignment): Likewise.
33531 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
33532 * tree-vect-loop.c (vect_analyze_loop): Likewise.
33533 * tree-vect-slp.c (vect_slp_bb): Likewise.
33534 * doc/tm.texi: Regenerate.
33535 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
33537 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
33538 the vector size as a poly_uint64 rather than an unsigned int.
33539 (current_vector_size): Change from an unsigned int to a poly_uint64.
33540 (get_vectype_for_scalar_type): Update accordingly.
33541 * tree.h (build_truth_vector_type): Take the size and number of
33542 units as a poly_uint64 rather than an unsigned int.
33543 (build_vector_type): Add a temporary overload that takes
33544 the number of units as a poly_uint64 rather than an unsigned int.
33545 * tree.c (make_vector_type): Likewise.
33546 (build_truth_vector_type): Take the number of units as a poly_uint64
33547 rather than an unsigned int.
33549 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33550 Alan Hayward <alan.hayward@arm.com>
33551 David Sherwood <david.sherwood@arm.com>
33553 * target.def (get_mask_mode): Take the number of units and length
33554 as poly_uint64s rather than unsigned ints.
33555 * targhooks.h (default_get_mask_mode): Update accordingly.
33556 * targhooks.c (default_get_mask_mode): Likewise.
33557 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
33558 * doc/tm.texi: Regenerate.
33560 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33561 Alan Hayward <alan.hayward@arm.com>
33562 David Sherwood <david.sherwood@arm.com>
33564 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
33565 * omp-general.c (omp_max_vf): Likewise.
33566 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
33567 (expand_omp_simd): Handle polynomial safelen.
33568 * omp-low.c (omplow_simd_context): Add a default constructor.
33569 (omplow_simd_context::max_vf): Change from int to poly_uint64.
33570 (lower_rec_simd_input_clauses): Update accordingly.
33571 (lower_rec_input_clauses): Likewise.
33573 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33574 Alan Hayward <alan.hayward@arm.com>
33575 David Sherwood <david.sherwood@arm.com>
33577 * tree-vectorizer.h (vect_nunits_for_cost): New function.
33578 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
33579 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
33580 (vect_analyze_slp_cost): Likewise.
33581 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
33582 (vect_model_load_cost): Likewise.
33584 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33585 Alan Hayward <alan.hayward@arm.com>
33586 David Sherwood <david.sherwood@arm.com>
33588 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
33589 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
33590 from an unsigned int * to a poly_uint64_pod *.
33591 (calculate_unrolling_factor): New function.
33592 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
33594 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33595 Alan Hayward <alan.hayward@arm.com>
33596 David Sherwood <david.sherwood@arm.com>
33598 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
33599 from an unsigned int to a poly_uint64.
33600 (_loop_vec_info::slp_unrolling_factor): Likewise.
33601 (_loop_vec_info::vectorization_factor): Change from an int
33603 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
33604 (vect_get_num_vectors): New function.
33605 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
33606 (vect_get_num_copies): Use vect_get_num_vectors.
33607 (vect_analyze_data_ref_dependences): Change max_vf from an int *
33608 to an unsigned int *.
33609 (vect_analyze_data_refs): Change min_vf from an int * to a
33611 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
33612 than an unsigned HOST_WIDE_INT.
33613 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
33614 (vect_analyze_data_ref_dependence): Change max_vf from an int *
33615 to an unsigned int *.
33616 (vect_analyze_data_ref_dependences): Likewise.
33617 (vect_compute_data_ref_alignment): Handle polynomial vf.
33618 (vect_enhance_data_refs_alignment): Likewise.
33619 (vect_prune_runtime_alias_test_list): Likewise.
33620 (vect_shift_permute_load_chain): Likewise.
33621 (vect_supportable_dr_alignment): Likewise.
33622 (dependence_distance_ge_vf): Take the vectorization factor as a
33623 poly_uint64 rather than an unsigned HOST_WIDE_INT.
33624 (vect_analyze_data_refs): Change min_vf from an int * to a
33626 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
33627 vfm1 as a poly_uint64 rather than an int. Make the same change
33628 for the returned bound_scalar.
33629 (vect_gen_vector_loop_niters): Handle polynomial vf.
33630 (vect_do_peeling): Likewise. Update call to
33631 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
33632 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
33634 * tree-vect-loop.c (vect_determine_vectorization_factor)
33635 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
33636 (vect_get_known_peeling_cost): Likewise.
33637 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
33638 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
33639 (vect_transform_loop): Likewise. Use the lowest possible VF when
33640 updating the upper bounds of the loop.
33641 (vect_min_worthwhile_factor): Make static. Return an unsigned int
33642 rather than an int.
33643 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
33644 polynomial unroll factors.
33645 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
33646 (vect_make_slp_decision): Likewise.
33647 (vect_supported_load_permutation_p): Likewise, and polynomial
33649 (vect_analyze_slp_cost): Handle polynomial vf.
33650 (vect_slp_analyze_node_operations): Likewise.
33651 (vect_slp_analyze_bb_1): Likewise.
33652 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
33653 than an unsigned HOST_WIDE_INT.
33654 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
33655 (vectorizable_load): Handle polynomial vf.
33656 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
33658 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
33660 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33661 Alan Hayward <alan.hayward@arm.com>
33662 David Sherwood <david.sherwood@arm.com>
33664 * match.pd: Handle bit operations involving three constants
33665 and try to fold one pair.
33667 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33669 * tree-vect-loop-manip.c: Include gimple-fold.h.
33670 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
33671 niters_maybe_zero parameters. Handle other cases besides a step of 1.
33672 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
33673 Add a path that uses a step of VF instead of 1, but disable it
33675 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
33676 and niters_no_overflow parameters. Update calls to
33677 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
33678 Create a new SSA name if the latter choses to use a ste other
33679 than zero, and return it via niters_vector_mult_vf_var.
33680 * tree-vect-loop.c (vect_transform_loop): Update calls to
33681 vect_do_peeling, vect_gen_vector_loop_niters and
33682 slpeel_make_loop_iterate_ntimes.
33683 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
33684 (vect_gen_vector_loop_niters): Update declarations after above changes.
33686 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
33688 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
33689 128-bit round to integer instructions.
33690 (ceil<mode>2): Likewise.
33691 (btrunc<mode>2): Likewise.
33692 (round<mode>2): Likewise.
33694 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33696 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
33697 unaligned VSX load/store on P8/P9.
33698 (expand_block_clear): Allow the use of unaligned VSX
33699 load/store on P8/P9.
33701 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33703 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
33705 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
33706 swap associated with both a load and a store.
33708 2018-01-02 Andrew Waterman <andrew@sifive.com>
33710 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
33711 * config/riscv/riscv.md (clear_cache): Use it.
33713 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
33715 * web.c: Remove out-of-date comment.
33717 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33719 * expr.c (fixup_args_size_notes): Check that any existing
33720 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
33721 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
33722 (emit_single_push_insn): ...here.
33724 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33726 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
33727 (const_vector_encoded_nelts): New function.
33728 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
33729 (const_vector_int_elt, const_vector_elt): Declare.
33730 * emit-rtl.c (const_vector_int_elt_1): New function.
33731 (const_vector_elt): Likewise.
33732 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
33733 of CONST_VECTOR_ELT.
33735 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33737 * expr.c: Include rtx-vector-builder.h.
33738 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
33739 directly on the tree encoding.
33740 (const_vector_from_tree): Likewise.
33741 * optabs.c: Include rtx-vector-builder.h.
33742 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
33743 sequence of "u" values.
33744 * vec-perm-indices.c: Include rtx-vector-builder.h.
33745 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
33746 directly on the vec_perm_indices encoding.
33748 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33750 * doc/rtl.texi (const_vector): Describe new encoding scheme.
33751 * Makefile.in (OBJS): Add rtx-vector-builder.o.
33752 * rtx-vector-builder.h: New file.
33753 * rtx-vector-builder.c: Likewise.
33754 * rtl.h (rtx_def::u2): Add a const_vector field.
33755 (CONST_VECTOR_NPATTERNS): New macro.
33756 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
33757 (CONST_VECTOR_DUPLICATE_P): Likewise.
33758 (CONST_VECTOR_STEPPED_P): Likewise.
33759 (CONST_VECTOR_ENCODED_ELT): Likewise.
33760 (const_vec_duplicate_p): Check for a duplicated vector encoding.
33761 (unwrap_const_vec_duplicate): Likewise.
33762 (const_vec_series_p): Check for a non-duplicated vector encoding.
33763 Say that the function only returns true for integer vectors.
33764 * emit-rtl.c: Include rtx-vector-builder.h.
33765 (gen_const_vec_duplicate_1): Delete.
33766 (gen_const_vector): Call gen_const_vec_duplicate instead of
33767 gen_const_vec_duplicate_1.
33768 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
33769 (gen_const_vec_duplicate): Use rtx_vector_builder.
33770 (gen_const_vec_series): Likewise.
33771 (gen_rtx_CONST_VECTOR): Likewise.
33772 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
33773 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
33774 Build a new vector rather than modifying a CONST_VECTOR in-place.
33775 (handle_special_swappables): Update call accordingly.
33776 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
33777 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
33778 Build a new vector rather than modifying a CONST_VECTOR in-place.
33779 (handle_special_swappables): Update call accordingly.
33781 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33783 * simplify-rtx.c (simplify_const_binary_operation): Use
33784 CONST_VECTOR_ELT instead of XVECEXP.
33786 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33788 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
33789 the selector elements to be different from the data elements
33790 if the selector is a VECTOR_CST.
33791 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
33792 ssizetype for the selector.
33794 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33796 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
33797 before testing each element individually.
33798 * tree-vect-generic.c (lower_vec_perm): Likewise.
33800 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33802 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
33803 * selftest-run-tests.c (selftest::run_tests): Call it.
33804 * vector-builder.h (vector_builder::operator ==): New function.
33805 (vector_builder::operator !=): Likewise.
33806 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
33807 (vec_perm_indices::all_from_input_p): New function.
33808 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
33809 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
33810 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
33811 instead of reading the VECTOR_CST directly. Detect whether both
33812 vector inputs are the same before constructing the vec_perm_indices,
33813 and update the number of inputs argument accordingly. Use the
33814 utility functions added above. Only construct sel2 if we need to.
33816 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33818 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
33819 the broadcast of the low byte.
33820 (expand_mult_highpart): Use an explicit encoding for the permutes.
33821 * optabs-query.c (can_mult_highpart_p): Likewise.
33822 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
33823 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
33824 (vectorizable_bswap): Likewise.
33825 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
33826 explicit encoding for the power-of-2 permutes.
33827 (vect_permute_store_chain): Likewise.
33828 (vect_grouped_load_supported): Likewise.
33829 (vect_permute_load_chain): Likewise.
33831 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33833 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
33834 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
33835 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
33836 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
33837 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
33838 (vect_gen_perm_mask_any): Likewise.
33840 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33842 * int-vector-builder.h: New file.
33843 * vec-perm-indices.h: Include int-vector-builder.h.
33844 (vec_perm_indices): Redefine as an int_vector_builder.
33845 (auto_vec_perm_indices): Delete.
33846 (vec_perm_builder): Redefine as a stand-alone class.
33847 (vec_perm_indices::vec_perm_indices): New function.
33848 (vec_perm_indices::clamp): Likewise.
33849 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
33850 (vec_perm_indices::new_vector): New function.
33851 (vec_perm_indices::new_expanded_vector): Update for new
33852 vec_perm_indices class.
33853 (vec_perm_indices::rotate_inputs): New function.
33854 (vec_perm_indices::all_in_range_p): Operate directly on the
33855 encoded form, without computing elided elements.
33856 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
33857 encoding. Update for new vec_perm_indices class.
33858 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
33859 the given vec_perm_builder.
33860 (expand_vec_perm_var): Update vec_perm_builder constructor.
33861 (expand_mult_highpart): Use vec_perm_builder instead of
33862 auto_vec_perm_indices.
33863 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
33864 vec_perm_indices instead of auto_vec_perm_indices. Use a single
33865 or double series encoding as appropriate.
33866 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
33867 vec_perm_indices instead of auto_vec_perm_indices.
33868 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
33869 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
33870 (vect_permute_store_chain): Likewise.
33871 (vect_grouped_load_supported): Likewise.
33872 (vect_permute_load_chain): Likewise.
33873 (vect_shift_permute_load_chain): Likewise.
33874 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
33875 (vect_transform_slp_perm_load): Likewise.
33876 (vect_schedule_slp_instance): Likewise.
33877 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
33878 (vectorizable_mask_load_store): Likewise.
33879 (vectorizable_bswap): Likewise.
33880 (vectorizable_store): Likewise.
33881 (vectorizable_load): Likewise.
33882 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
33883 vec_perm_indices instead of auto_vec_perm_indices. Use
33884 tree_to_vec_perm_builder to read the vector from a tree.
33885 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
33886 vec_perm_builder instead of a vec_perm_indices.
33887 (have_whole_vector_shift): Use vec_perm_builder and
33888 vec_perm_indices instead of auto_vec_perm_indices. Leave the
33889 truncation to calc_vec_perm_mask_for_shift.
33890 (vect_create_epilog_for_reduction): Likewise.
33891 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
33892 from auto_vec_perm_indices to vec_perm_indices.
33893 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
33894 instead of changing individual elements.
33895 (aarch64_vectorize_vec_perm_const): Use new_vector to install
33896 the vector in d.perm.
33897 * config/arm/arm.c (expand_vec_perm_d::perm): Change
33898 from auto_vec_perm_indices to vec_perm_indices.
33899 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
33900 instead of changing individual elements.
33901 (arm_vectorize_vec_perm_const): Use new_vector to install
33902 the vector in d.perm.
33903 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
33904 Update vec_perm_builder constructor.
33905 (rs6000_expand_interleave): Likewise.
33906 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
33907 (rs6000_expand_interleave): Likewise.
33909 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33911 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
33912 to qimode could truncate the indices.
33913 * optabs.c (expand_vec_perm_var): Likewise.
33915 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33917 * Makefile.in (OBJS): Add vec-perm-indices.o.
33918 * vec-perm-indices.h: New file.
33919 * vec-perm-indices.c: Likewise.
33920 * target.h (vec_perm_indices): Replace with a forward class
33922 (auto_vec_perm_indices): Move to vec-perm-indices.h.
33923 * optabs.h: Include vec-perm-indices.h.
33924 (expand_vec_perm): Delete.
33925 (selector_fits_mode_p, expand_vec_perm_var): Declare.
33926 (expand_vec_perm_const): Declare.
33927 * target.def (vec_perm_const_ok): Replace with...
33928 (vec_perm_const): ...this new hook.
33929 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
33930 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
33931 * doc/tm.texi: Regenerate.
33932 * optabs.def (vec_perm_const): Delete.
33933 * doc/md.texi (vec_perm_const): Likewise.
33934 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
33935 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
33936 expand_vec_perm for constant permutation vectors. Assert that
33937 the mode of variable permutation vectors is the integer equivalent
33938 of the mode that is being permuted.
33939 * optabs-query.h (selector_fits_mode_p): Declare.
33940 * optabs-query.c: Include vec-perm-indices.h.
33941 (selector_fits_mode_p): New function.
33942 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
33943 is defined, instead of checking whether the vec_perm_const_optab
33944 exists. Use targetm.vectorize.vec_perm_const instead of
33945 targetm.vectorize.vec_perm_const_ok. Check whether the indices
33946 fit in the vector mode before using a variable permute.
33947 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
33948 vec_perm_indices instead of an rtx.
33949 (expand_vec_perm): Replace with...
33950 (expand_vec_perm_const): ...this new function. Take the selector
33951 as a vec_perm_indices rather than an rtx. Also take the mode of
33952 the selector. Update call to shift_amt_for_vec_perm_mask.
33953 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
33954 Use vec_perm_indices::new_expanded_vector to expand the original
33955 selector into bytes. Check whether the indices fit in the vector
33956 mode before using a variable permute.
33957 (expand_vec_perm_var): Make global.
33958 (expand_mult_highpart): Use expand_vec_perm_const.
33959 * fold-const.c: Includes vec-perm-indices.h.
33960 * tree-ssa-forwprop.c: Likewise.
33961 * tree-vect-data-refs.c: Likewise.
33962 * tree-vect-generic.c: Likewise.
33963 * tree-vect-loop.c: Likewise.
33964 * tree-vect-slp.c: Likewise.
33965 * tree-vect-stmts.c: Likewise.
33966 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
33968 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
33969 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
33970 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
33971 (aarch64_vectorize_vec_perm_const): ...this new function.
33972 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
33973 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
33974 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
33975 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
33976 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
33977 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
33978 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
33980 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
33981 check for NEON modes.
33982 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
33983 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
33984 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
33985 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
33987 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
33988 the old VEC_PERM_CONST conditions.
33989 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
33990 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
33991 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
33992 (ia64_vectorize_vec_perm_const_ok): Merge into...
33993 (ia64_vectorize_vec_perm_const): ...this new function.
33994 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
33995 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
33996 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
33997 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
33998 * config/mips/mips.c (mips_expand_vec_perm_const)
33999 (mips_vectorize_vec_perm_const_ok): Merge into...
34000 (mips_vectorize_vec_perm_const): ...this new function.
34001 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
34002 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
34003 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
34004 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
34005 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
34006 (rs6000_expand_vec_perm_const): Delete.
34007 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
34009 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
34010 (altivec_expand_vec_perm_const_le): Take each operand individually.
34011 Operate on constant selectors rather than rtxes.
34012 (altivec_expand_vec_perm_const): Likewise. Update call to
34013 altivec_expand_vec_perm_const_le.
34014 (rs6000_expand_vec_perm_const): Delete.
34015 (rs6000_vectorize_vec_perm_const_ok): Delete.
34016 (rs6000_vectorize_vec_perm_const): New function.
34017 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
34018 an element count and rtx array.
34019 (rs6000_expand_extract_even): Update call accordingly.
34020 (rs6000_expand_interleave): Likewise.
34021 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
34022 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
34023 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
34024 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
34025 (rs6000_expand_vec_perm_const): Delete.
34026 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
34027 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
34028 (altivec_expand_vec_perm_const_le): Take each operand individually.
34029 Operate on constant selectors rather than rtxes.
34030 (altivec_expand_vec_perm_const): Likewise. Update call to
34031 altivec_expand_vec_perm_const_le.
34032 (rs6000_expand_vec_perm_const): Delete.
34033 (rs6000_vectorize_vec_perm_const_ok): Delete.
34034 (rs6000_vectorize_vec_perm_const): New function. Remove stray
34035 reference to the SPE evmerge intructions.
34036 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
34037 an element count and rtx array.
34038 (rs6000_expand_extract_even): Update call accordingly.
34039 (rs6000_expand_interleave): Likewise.
34040 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
34041 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
34043 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
34045 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34047 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
34048 vector mode and that that mode matches the mode of the data
34050 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
34051 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
34052 directly using expand_vec_perm_1 when forcing selectors into
34054 (expand_vec_perm_var): New function, split out from expand_vec_perm.
34056 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34058 * optabs-query.h (can_vec_perm_p): Delete.
34059 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
34060 * optabs-query.c (can_vec_perm_p): Split into...
34061 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
34062 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
34063 particular selector is valid.
34064 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
34065 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
34066 (vect_grouped_load_supported): Likewise.
34067 (vect_shift_permute_load_chain): Likewise.
34068 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
34069 (vect_transform_slp_perm_load): Likewise.
34070 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
34071 (vectorizable_bswap): Likewise.
34072 (vect_gen_perm_mask_checked): Likewise.
34073 * fold-const.c (fold_ternary_loc): Likewise. Don't take
34074 implementations of variable permutation vectors into account
34075 when deciding which selector to use.
34076 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
34077 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
34078 with a false third argument.
34079 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
34080 to test whether the constant selector is valid and can_vec_perm_var_p
34081 to test whether a variable selector is valid.
34083 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34085 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
34086 * optabs-query.c (can_vec_perm_p): Likewise.
34087 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
34088 instead of vec_perm_indices.
34089 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
34090 (vect_gen_perm_mask_checked): Likewise,
34091 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
34092 (vect_gen_perm_mask_checked): Likewise,
34094 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34096 * optabs-query.h (qimode_for_vec_perm): Declare.
34097 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
34098 (qimode_for_vec_perm): ...this new function.
34099 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
34101 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34103 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
34104 does not have a conditional at the top.
34106 2018-01-02 Richard Biener <rguenther@suse.de>
34108 * ipa-inline.c (big_speedup_p): Fix expression.
34110 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
34113 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
34116 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
34120 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
34121 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
34122 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
34123 cond_taken_branch_cost 3->4.
34125 2018-01-01 Jakub Jelinek <jakub@redhat.com>
34127 PR tree-optimization/83581
34128 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
34129 TODO_cleanup_cfg if any changes have been made.
34131 PR middle-end/83608
34132 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
34133 convert_modes if target mode has the right side, but different mode
34136 PR middle-end/83609
34137 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
34138 last argument when extracting from CONCAT. If either from_real or
34139 from_imag is NULL, use expansion through memory. If result is not
34140 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
34141 the parts directly to inner mode, if even that fails, use expansion
34144 PR middle-end/83623
34145 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
34146 check for bswap in mode rather than HImode and use that in expand_unop
34149 Copyright (C) 2018 Free Software Foundation, Inc.
34151 Copying and distribution of this file, with or without modification,
34152 are permitted in any medium without royalty provided the copyright
34153 notice and this notice are preserved.