1 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
3 * config/s390/s390.c (s390_default_align): Set default function
5 (s390_override_options_after_change): Call s390_default align.
6 (s390_option_override_internal): Call s390_default_align.
7 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
9 2018-07-17 Jakub Jelinek <jakub@redhat.com>
12 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
13 to _looptemp_ clauses, other than the first two.
15 2018-07-17 Martin Liska <mliska@suse.cz>
17 * opts.c: Do not enable OPT_falign_* for -Os.
19 2018-07-17 Martin Liska <mliska@suse.cz>
21 * align.h (MAX_CODE_ALIGN): New.
22 (MAX_CODE_ALIGN_VALUE): New.
23 * common/config/i386/i386-common.c (ix86_handle_option):
24 (MAX_CODE_ALIGN): Moved to align.h.
25 * final.c (MAX_CODE_ALIGN): Likewise.
26 * opts.c (parse_and_check_align_values):
27 (MAX_CODE_ALIGN): Likewise.
28 (MAX_CODE_ALIGN_VALUE): Likewise.
30 2018-07-17 Martin Liska <mliska@suse.cz>
32 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
33 in order to fulfil coding style.
34 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
35 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
36 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
37 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
38 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
39 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
40 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
41 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
42 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
44 2018-07-17 Martin Liska <mliska@suse.cz>
47 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions directly.
48 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
49 align_flags of label_to_alignment.
50 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into align_flags
52 * config/m68k/m68k.c: Do not use removed align_labels_value and
54 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
55 (LOOP_ALIGN): Likewise.
56 (LABEL_ALIGN): Likewise.
57 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
58 Remove not used macro.
59 (rs6000_loop_align): Change return type to align_flags.
60 (rs6000_loop_align_max_skip): Remove.
61 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
62 Change return type to align_flags.
63 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
64 Remove not used macro.
65 (rs6000_loop_align): Change return type to align_flags.
66 (rs6000_loop_align_max_skip): Remove.
67 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
68 * config/rx/rx-protos.h (rx_align_for_label): Make it
70 * config/rx/rx.c (rx_align_for_label): Change return type
72 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
74 into align_flags class.
75 (LABEL_ALIGN): Likewise.
76 (LOOP_ALIGN): Likewise.
77 * config/s390/s390.c (s390_label_align): Use align_flags
79 (s390_asm_output_function_label): Likewise.
80 * config/sh/sh.c (sh_override_options_after_change):
81 Use align_flags class directly without macros.
82 (find_barrier): Likewise.
83 (barrier_align): Likewise.
84 (sh_loop_align): Likewise.
85 * config/spu/spu.c (spu_option_override):
86 Use align_flags_tuple::get_value instead of removed macros.
87 (spu_sched_init): Likewise.
88 * config/spu/spu.h (GTY): Likewise.
89 * config/visium/visium.c (visium_option_override):
90 Set "8" as default secondary alignment.
91 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
92 in order to guarantee secondary alignment of 8.
93 * coretypes.h: Include align.h header file.
94 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
95 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
96 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
97 * doc/tm.texi.in: Likewise.
98 * final.c (struct label_alignment): Remove not used structure.
99 (LABEL_ALIGN): Change type to align_flags.
100 (LOOP_ALIGN): Likewise.
101 (JUMP_ALIGN): Likewise.
102 (default_loop_align_max_skip): Remove.
103 (default_label_align_max_skip): Likewise.
104 (default_jump_align_max_skip): Likewise.
105 (default_label_align_after_barrier_max_skip):
106 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
107 (LABEL_TO_MAX_SKIP): Remove.
108 (label_to_alignment): Return align_flags type instead of integer.
109 (label_to_max_skip): Remove.
110 (align_fuzz): Use align_flags type.
111 (compute_alignments): Use align_flags type and use align_flags::max
112 to combine multiple alignments.
113 (grow_label_align): Grow vec instead of C array.
114 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
115 (shorten_branches): Use align_flags type and use align_flags::max
116 to combine multiple alignments.
117 (final_scan_insn_1): Remove usage of secondary alignment that comes
118 from label alignment, but instead use proper secondary alignment
119 which is computed in grow_label_align.
120 * flags.h (struct align_flags_tuple): Move to align.h.
121 (struct align_flags): Likewise.
122 (state_align_loops): Rename to align_loops.
123 (state_align_jumps): Rename to align_jumps.
124 (state_align_labels): Rename to align_labels.
125 (state_align_functions): Rename to align_functions.
126 (align_loops_log): Remove.
127 (align_jumps_log): Remove.
128 (align_labels_log): Remove.
129 (align_functions_log): Remove.
130 (align_loops_max_skip): Remove.
131 (align_jumps_max_skip): Remove.
132 (align_labels_max_skip): Remove.
133 (align_functions_max_skip): Remove.
134 (align_loops_value): Remove.
135 (align_jumps_value): Remove.
136 (align_labels_value): Remove.
137 (align_functions_value): Remove.
138 * output.h (label_to_alignment): Change return type to align_flags.
139 (label_to_max_skip): Remove.
140 * target.def: Remove loop_align_max_skip, label_align_max_skip,
141 jump_align_max_skip macros.
142 * targhooks.h (default_loop_align_max_skip): Remove.
143 (default_label_align_max_skip): Likewise.
144 (default_jump_align_max_skip): Likewise.
145 (default_label_align_after_barrier_max_skip): Remove.
146 * toplev.c (read_log_maxskip): Use ::normalize function.
147 (parse_N_M): Remove not used argument and also call ::normalize.
148 (parse_alignment_opts): Do not pass unused arguments.
149 * varasm.c (assemble_start_function): Use directly align_functions
150 instead of removed macros.
151 * system.h: Do not poison removed macros.
153 2018-07-17 Jakub Jelinek <jakub@redhat.com>
156 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
157 and cond temporaries don't have reference type if iterator has
158 pointer type. For init use &for_pre_body instead of pre_p if
159 for_pre_body is non-empty.
161 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
163 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
164 double-double modes to SFmode directly directly.
165 (trunc<mode>sf2_fprs): Delete.
167 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
169 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
170 for conversions between IFmode and the decimal floating point modes.
171 (init_float128_ieee): Use the correct names for conversions between
172 KFmode and the decimal floating point modes.
174 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
176 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
177 for the conversions between TDmode and IFmode.
178 (init_float128_ieee): Use more correct names for the conversions
179 between TDmode and KFmode.
181 2018-07-16 Jakub Jelinek <jakub@redhat.com>
183 PR tree-optimization/86526
184 * builtins.c (expand_builtin_memcmp): Formatting fixes.
185 (inline_expand_builtin_string_cmp): Likewise.
186 (inline_string_cmp): Likewise. Use c_readstr instead of
187 builtin_memcpy_read_str. Add unit_mode temporary.
189 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
192 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
193 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
195 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
197 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
198 Alphabetize prototypes of built-in functions, separating out
199 built-in functions that are listed in this section but should be
202 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
205 * expmed.c (emit_store_flag): Do not emit setcc followed by a
206 conditional move when trapping comparison was split to a
207 non-trapping one (and vice versa).
209 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
211 * config/s390/s390.c (s390_function_profiler): Generate nops
212 instead of profiler call sequences.
213 * config/s390/s390.opt: Add the new option.
215 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
217 * config/s390/s390.c (s390_function_profiler): Generate
218 __mcount_loc section.
219 * config/s390/s390.opt: Add the new option.
221 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
223 * common.opt: Add the new warning.
224 * config/s390/s390.c (s390_function_profiler): Emit "brasl
225 %r0,__fentry__" when -mfentry is specified.
226 (s390_option_override_internal): Disallow -mfentry for 31-bit
228 * config/s390/s390.opt: Add the new option.
230 2018-07-16 Richard Biener <rguenther@suse.de>
233 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
234 for function-local FUNCTION_DECL and RESULT_DECL immediately.
236 2018-07-16 Martin Liska <mliska@suse.cz>
239 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
242 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
244 * config/arc/arcHS.md: Update ARCHS scheduling rules.
246 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
248 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
250 * config/arc/arc-cpus.def (hs4x): New cpu.
252 * config/arc/arc-tables.opt: Regenerate.
253 * config/arc/arc.c (arc_sched_issue_rate): New function.
254 (TARGET_SCHED_ISSUE_RATE): Define.
255 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
256 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
258 (attr tune): Add ARCHS4x tune values.
259 (attr tune_dspmpy): Define.
260 (*tst): Correct instruction type.
261 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
262 * config/arc/arcHS4x.md: New file.
263 * config/arc/fpu.md: Update instruction type attributes.
264 * config/arc/t-multilib: Regenerate.
266 2018-07-16 Tom de Vries <tdevries@suse.de>
269 * var-tracking.c (vt_initialize): Fix pre_dec handling.
271 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
273 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
276 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
278 PR tree-optimization/86514
279 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
280 conversion to a boolean type from a type with greater precision.
282 2018-07-16 Tom de Vries <tdevries@suse.de>
284 * var-tracking.c (vt_initialize): Print adjusted insn slim if
285 dump_flags request TDF_SLIM.
287 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
289 * fold-const.c (int_const_binop_1): Abstract...
290 (wide_int_binop): ...wide int code here.
291 (poly_int_binop): ...poly int code here.
292 Abstract the rest of int_const_binop_1 into int_const_binop.
293 * fold-const.h (wide_int_binop): New.
294 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
295 Remove useless PLUS/MINUS_EXPR case.
296 (zero_nonzero_bits_from_vr): Move wide int code...
297 (zero_nonzero_bits_from_bounds): ...here.
298 (extract_range_from_binary_expr_1): Move mask optimization code...
299 (range_easy_mask_min_max): ...here.
300 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
301 (range_easy_mask_min_max): New.
303 2018-07-15 Jeff Law <law@redhat.com>
306 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
309 2018-07-14 Jim Wilson <jimw@sifive.com>
311 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
313 2018-07-14 Paul Koning <ni1d@arrl.net>
315 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
317 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
319 * lto-streamer-out.c (copy_function_or_variable): Dump info about
322 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
323 Steve Munroe <munroesj52@gmail.com>
325 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
326 (_mm_andnot_si128): Likewise.
327 (_mm_or_si128): Likewise.
328 (_mm_xor_si128): Likewise.
330 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
333 * builtins.c (expand_builtin_memcmp): Inline the calls first
334 when result_eq is false.
335 (expand_builtin_strcmp): Inline the calls first.
336 (expand_builtin_strncmp): Likewise.
337 (inline_string_cmp): New routine. Expand a string compare
338 call by using a sequence of char comparison.
339 (inline_expand_builtin_string_cmp): New routine. Inline expansion
340 a call to str(n)cmp/memcmp.
341 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
343 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
345 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
347 * config/arm/driver-arm.c: Include arm-native.h.
348 (host_detect_local_cpu): Use auto-generated data tables.
349 (vendors, arm_cpu_table): Delete. Move part information to ...
350 * config/arm/arm-cpus.in: ... here.
351 * config/arm/parsecpu.awk (gen_native): New function.
352 (vendor, part): New CPU fields.
353 (END): Add support for building the native CPU detection tables.
354 * config/arm/t-arm (arm-native.h): Add build rule.
355 (driver-arm.o): Add dependency on arm-native.h.
357 2018-07-13 Richard Biener <rguenther@suse.de>
360 * match.pd (addr1 - addr2): Allow either of the operand to
363 2018-07-13 Tom de Vries <tdevries@suse.de>
365 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
368 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
370 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
371 arrays instead of numbered variables.
373 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
375 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
376 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
378 2018-07-13 Richard Biener <rguenther@suse.de>
381 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
382 instead of get_context_die.
384 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
385 Richard Biener <rguenther@suse.de>
388 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
389 that the loop latch destination where phi is defined.
391 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
393 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
394 (riscv_expand_epilogue): Add assertion to check interrupt mode.
395 (riscv_set_current_function): Extract getting interrupt type to new
397 (riscv_get_interrupt_type): New function.
398 (riscv_merge_decl_attributes): New function, checking interrupt type is
400 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
402 2018-07-12 Paul Koning <ni1d@arrl.net>
404 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
407 2018-07-12 Paul Koning <ni1d@arrl.net>
409 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
410 zero reference, add doloop_end instead.
411 * doc/md.texi (decrement_and_branch_until_zero): Remove.
412 (Looping patterns): Remove decrement_and_branch_until_zero. Add
413 detail for doloop_end.
415 2018-07-12 Martin Sebor <msebor@redhat.com>
418 * attribs.c (decl_attributes): Reject conflicting attributes before
419 calling attribute handlers.
421 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
423 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
425 (gcc::dump_manager::get_dump_file_name): likewise.
426 (dump_begin): Likewise.
427 * dumpfile.h (dump_begin): Update prototype.
428 (gcc::dump_manager::get_dump_file_name,
429 gcc::dump_manager::get_dump_file_name): Update prototype.
431 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
433 * internal-fn.h (vectorizable_internal_fn_p): New function.
434 * tree-vect-slp.c (compatible_calls_p): Likewise.
435 (vect_build_slp_tree_1): Remove nops argument. Handle calls
436 to internal functions.
437 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
439 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
441 * fold-const.h (inverse_conditions_p): Declare.
442 * fold-const.c (inverse_conditions_p): New function.
443 * match.pd: Use inverse_conditions_p. Add folds of view_converts
444 that test the inverse condition of a conditional internal function.
445 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
446 * internal-fn.c (internal_fn_mask_index): Handle conditional
448 (vectorized_internal_fn_supported_p): New function.
449 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
450 (any_pred_load_store): Replace with...
451 (need_to_predicate): ...this new variable.
452 (redundant_ssa_names): New variable.
453 (ifcvt_can_use_mask_load_store): Move initial checks to...
454 (ifcvt_can_predicate): ...this new function. Handle tree codes
455 for which a conditional internal function exists.
456 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
457 instead of ifcvt_can_use_mask_load_store. Update after variable
459 (predicate_load_or_store): New function, split out from
460 predicate_mem_writes.
461 (check_redundant_cond_expr): New function.
462 (value_available_p): Likewise.
463 (predicate_rhs_code): Likewise.
464 (predicate_mem_writes): Rename to...
465 (predicate_statements): ...this. Use predicate_load_or_store
466 and predicate_rhs_code.
467 (combine_blocks, tree_if_conversion): Update after above name changes.
468 (ifcvt_local_dce): Handle redundant_ssa_names.
469 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
470 general conditional functions.
471 * tree-vect-stmts.c (vectorizable_call): Likewise.
473 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
474 Alan Hayward <alan.hayward@arm.com>
475 David Sherwood <david.sherwood@arm.com>
477 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
478 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
479 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
480 plus and minus and convert them into IFN_COND_FMA-based sequences.
481 (convert_mult_to_fma): Handle conditional plus and minus.
483 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
485 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
486 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
487 (cond_fnms_optab): New optabs.
488 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
490 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
491 * internal-fn.h (get_conditional_internal_fn): Declare.
492 (get_unconditional_internal_fn): Likewise.
493 * internal-fn.c (cond_ternary_direct): New macro.
494 (expand_cond_ternary_optab_fn): Likewise.
495 (direct_cond_ternary_optab_supported_p): Likewise.
496 (FOR_EACH_COND_FN_PAIR): Likewise.
497 (get_conditional_internal_fn): New function.
498 (get_unconditional_internal_fn): Likewise.
499 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
500 (gimple_match_op::gimple_match_op): Add a new overload for 5
502 (gimple_match_op::set_op): Likewise.
503 (gimple_resimplify5): Declare.
504 * genmatch.c (decision_tree::gen): Generate simplifications for
506 * gimple-match-head.c (gimple_simplify): Define an overload for
507 5 operands. Handle calls with 5 arguments in the top-level overload.
508 (convert_conditional_op): Handle conversions from unconditional
509 internal functions to conditional ones.
510 (gimple_resimplify5): New function.
511 (build_call_internal): Pass a fifth operand.
512 (maybe_push_res_to_seq): Likewise.
513 (try_conditional_simplification): Try converting conditional
514 internal functions to unconditional internal functions.
515 Handle 3-operand unconditional forms.
516 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
517 Define ternary equivalents of the current rules for binary conditional
519 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
521 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
522 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
523 (optab): Handle them.
524 (SVE_COND_FP_TERNARY): New int iterator.
525 (sve_fmla_op, sve_fmad_op): New int attributes.
526 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
527 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
528 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
530 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
532 * target.def (preferred_else_value): New target hook.
533 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
534 * doc/tm.texi: Regenerate.
535 * targhooks.h (default_preferred_else_value): Declare.
536 * targhooks.c (default_preferred_else_value): New function.
537 * internal-fn.h (conditional_internal_fn_code): Declare.
538 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
539 (get_conditional_internal_fn): Use it.
540 (conditional_internal_fn_code): New function.
541 * gimple-match.h (gimple_match_cond): New struct.
542 (gimple_match_op): Add a cond member function.
543 (gimple_match_op::gimple_match_op): Update all forms to take a
545 * genmatch.c (expr::gen_transform): Use the same condition as res_op
546 for the suboperation, but don't specify a particular else_value.
547 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
548 (visit_nary_op, visit_reference_op_load): Pass
549 gimple_match_cond::UNCOND to the gimple_match_op constructor.
550 * gimple-match-head.c: Include tree-eh.h
551 (convert_conditional_op): New function.
552 (maybe_resimplify_conditional_op): Likewise.
553 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
554 (gimple_resimplify2): Likewise.
555 (gimple_resimplify3): Likewise.
556 (gimple_resimplify4): Likewise.
557 (maybe_push_res_to_seq): Return null for conditional operations.
558 (try_conditional_simplification): New function.
559 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
561 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
563 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
565 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
567 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
569 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
571 (hash_tree): Do not hash DECL_FCONTEXT
572 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
573 Do not stream DECL_FCONTEXT.
574 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
575 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
577 2018-07-12 Richard Biener <rguenther@suse.de>
580 * dwarf2out.c (gen_block_die): Only output blocks when they have
581 at least one !DECL_IGNORED_P variable.
583 2018-07-12 Richard Biener <rguenther@suse.de>
586 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
587 Remove -mieee-fp handling.
589 2018-07-12 Richard Biener <rguenther@suse.de>
591 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
592 left-over from last patch.
594 2018-07-12 Jakub Jelinek <jakub@redhat.com>
596 PR tree-optimization/86492
597 * gimple-ssa-store-merging.c
598 (imm_store_chain_info::coalesce_immediate_stores): Call
599 check_no_overlap even for the merge_overlapping case. Formatting fix.
601 2018-07-12 Richard Biener <rguenther@suse.de>
604 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
605 move possibly trapping operations into the conditional.
607 2018-07-12 Richard Biener <rguenther@suse.de>
609 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
610 (vn_lookup_simplify_result): Remove recursion limit applied
612 (vn_nary_build_or_lookup_1): Adjust.
613 (try_to_simplify): Likewise.
614 * gimple-match-head.c (gimple_resimplify1): Instead apply one
616 (gimple_resimplify2): Likewise.
617 (gimple_resimplify3): Likewise.
618 (gimple_resimplify4): Likewise.
620 2018-07-11 Jakub Jelinek <jakub@redhat.com>
622 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
623 Use __mmask64 type instead of __mmask8 for __M argument.
624 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
625 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
627 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
628 __mmask16 for __M argument.
629 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
630 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
631 to __mmask16 instead of __mmask8.
632 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
633 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
634 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
635 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
636 instead of __mmask16 for __U argument.
637 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
638 __mmask16 instead of __mmask8 for __U argument.
639 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
641 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
643 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
645 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
647 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
649 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
651 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
653 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
655 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
657 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
659 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
660 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
661 return type as well as __M argument type and all casts from __mmask8
663 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
664 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
665 return type as well as __M argument type and all casts from __mmask8
667 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
668 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
669 return type as well as __M argument type and all casts from __mmask8
671 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
672 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
673 return type as well as __M argument type and all casts from __mmask8
675 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
676 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
679 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
681 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
682 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
685 2018-07-11 Paul Koning <ni1d@arrl.net>
687 * doc/md.texi (define_subst): Document how multiple occurrences of
688 the same argument in the replacement pattern are handled.
690 2018-07-11 Paul Koning <ni1d@arrl.net>
692 * doc/extend.texi (Common Variable Attributes): Move "mode" into
694 (Common Type Attributes): Add "mode" attribute.
696 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
698 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
699 stream DECL_ORIGINAL_TYPE.
700 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
701 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
702 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
703 Do not walk original type.
704 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
706 (write_ts_decl_non_common_tree_pointers): Do not stream
708 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
709 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
711 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
713 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
714 thread twice from the same starting edge.
716 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
718 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
719 * gimple.c (gimple_call_nonnull_result_p): ...here...
720 (gimple_call_nonnull_arg): ...and here.
721 * gimple.h (gimple_call_nonnull_result_p): New.
722 (gimple_call_nonnull_arg): New.
724 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
726 * config/arm/arm-cpus.in: Move information from fpu field of each
727 cpu definition to the isa field.
728 * config/arm/parsecpu.awk (fpu): Delete match rule.
729 (gen_comm_data): Don't add bits from the CPU's FPU entry.
731 2018-07-11 Richard Biener <rguenther@suse.de>
734 * dwarf2out.c (init_sections_and_labels): Use
735 output_asm_line_debug_info consistently.
736 (dwarf2out_early_finish): Likewise.
737 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
740 2018-07-11 Richard Biener <rguenther@suse.de>
742 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
743 Rework father_bb setting in a way to avoid propagating constants
744 multiple times on a loop body.
746 2018-07-10 Mark Wielaard <mark@klomp.org>
749 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
751 2018-07-10 Richard Biener <rguenther@suse.de>
753 * hash-map.h (hash_map::iterator::operator*): Return
754 references to key and value.
756 2018-07-10 Jakub Jelinek <jakub@redhat.com>
759 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
760 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
761 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
762 (gimplify_omp_for): For composite loops, move outer
763 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
764 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
765 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
766 TREE_LIST for both the original class iterator and the "last" helper
767 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
768 loop, remember has_decl_expr from outer composite loops for the
769 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
771 2018-07-09 Martin Sebor <msebor@redhat.com>
775 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
776 accessing implicitly initialized array elements.
777 * expr.c (string_constant): Handle string initializers of
778 character arrays within aggregates.
779 * gimple-fold.c (fold_array_ctor_reference): Add argument.
780 Store element offset. As a special case, handle zero size.
781 (fold_nonarray_ctor_reference): Same.
782 (fold_ctor_reference): Add argument. Store subobject offset.
783 * gimple-fold.h (fold_ctor_reference): Add argument.
785 2018-07-09 Paul Koning <ni1d@arrl.net>
787 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
788 (pdp11_insn_cost): New function.
789 (pdp11_md_asm_adjust): New function.
790 (TARGET_INVALID_WITHIN_DOLOOP): Define.
791 (pdp11_rtx_costs): Update to match machine better.
792 (output_addr_const_pdp11): Correct format mismatch warnings.
793 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
794 * config/pdp11/pdp11.md: General change to add base_cost and/or
795 length attributes for use by new pdp11_insn_cost function.
796 (MIN_BRANCH): Correct definition.
798 (doloop_end): Use standard pattern name for looping pattern.
799 (doloop_end_nocc): New.
800 (movsf): Add another constraint alternative.
801 (zero_extendqihi2): Add constraint alternatives for not in place
803 (zero_extendhisi2): Remove.
804 (shift patterns): Add CC handling variants.
808 (define_peephole2): New peephole to recognize mov that sets CC for
811 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
813 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
814 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
817 2018-07-09 Jakub Jelinek <jakub@redhat.com>
820 * real.c (real_nextafter): Return true if result is denormal.
822 2018-07-09 Martin Liska <mliska@suse.cz>
824 * common.opt: Add back wrongly removed attribute.
826 2018-07-09 Richard Biener <rguenther@suse.de>
829 * dwarf2out.c (gen_block_die): For an early generated DIE
830 always output high/low PC attributes.
832 2018-07-09 Tom de Vries <tdevries@suse.de>
834 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
835 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
836 onto VAR_DECL with abstract origin.
838 2018-07-07 Jim Wilson <jimw@sifive.com>
840 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
842 2018-07-07 Tom de Vries <tdevries@suse.de>
844 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
846 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
848 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
850 (combine_bound): Use wide-int overflow calculation instead of
852 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
854 * fold-const.c (int_const_binop_2): Same.
855 (extract_muldiv_1): Same.
856 (fold_div_compare): Same.
857 (fold_abs_const): Same.
859 * poly-int.h (add): Same.
863 * predict.c (predict_iv_comparison): Same.
864 * profile-count.c (slow_safe_scale_64bit): Same.
865 * simplify-rtx.c (simplify_const_binary_operation): Same.
866 * tree-chrec.c (tree_fold_binomial): Same.
867 * tree-data-ref.c (split_constant_offset_1): Same.
868 * tree-if-conv.c (idx_within_array_bound): Same.
869 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
870 * tree-ssa-phiopt.c (minmax_replacement): Same.
871 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
872 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
873 * vr-values.c (vr_values::adjust_range_with_scev): Same.
874 * wide-int.cc (wi::add_large): Same.
875 (wi::mul_internal): Same.
876 (wi::sub_large): Same.
877 (wi::divmod_internal): Same.
878 * wide-int.h: Change overflow type to overflow_type for neg, add,
879 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
880 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
881 mul_internal, divmod_internal.
882 (overflow_type): New enum.
883 (accumulate_overflow): New.
885 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
887 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
888 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
890 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
892 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
893 argument is checked for zero before entering loop, avoid checking again.
895 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
897 * gimplify.h (generic_expr_could_trap_p): Set as global function.
898 * gimplify.h (generic_expr_could_trap_p): Likwise.
899 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
901 2018-07-06 Jakub Jelinek <jakub@redhat.com>
903 PR tree-optimization/86401
904 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
905 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
906 (fold_bit_and_mask): ... here. New helper function for match.pd.
907 * fold-const.h (fold_bit_and_mask): Declare.
908 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
910 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
913 * target.def (translate_mode_attribute): New hook.
914 * targhooks.h (default_translate_mode_attribute): Declare.
915 * targhooks.c (default_translate_mode_attribute): New function.
916 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
917 * doc/tm.texi: Regenerate.
918 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
919 (rs6000_translate_mode_attribute): New function.
921 2018-07-06 Paul Koning <ni1d@arrl.net>
923 * doc/md.texi (define_split): Document DONE and FAIL.
924 (define_peephole2): Ditto.
926 2018-07-05 Jeff Law <law@redhat.com>
928 PR tree-optimization/86010
929 * tree-ssa-dse.c (compute_trims): More aggressively trim at
930 both the head and tail of mem* and str* calls.
932 2018-07-05 Jim Wilson <jimw@sifive.com>
934 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
936 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
938 * config/aarch64/aarch64-simd.md: correct flags text for
939 MIN_EXPR replacement.
941 2018-07-05 James Clarke <jrtc27@jrtc27.com>
943 * configure: Regenerated.
945 2018-07-05 Carl Love <cel@us.ibm.com>
947 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
948 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
949 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
950 VSX_BUILTIN_DOUBLEL_V4SF.
952 2018-07-05 Martin Sebor <msebor@redhat.com>
955 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
956 than its domain to compute its the upper bound of a char array.
958 2018-07-05 Nathan Sidwell <nathan@acm.org>
960 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
961 * doc/cpp.texi: Update comment.
962 * doc/tm.texi: Rebuilt.
963 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
964 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
965 * doc/extend.texi (Backwards Compatibility): Clarify it is system
966 headers affected by extern "C".
967 * system.h: Poison NO_IMPLICIT_EXTERN_C.
968 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
969 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
970 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
971 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
972 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
973 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
974 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
975 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
976 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
977 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
978 config/sparc/sp64-elf.h, config/spu/spu.h,
979 config/stormy16/stormy16.h, config/v850/v850.h,
980 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
981 define NO_IMPLICIT_EXTERN_C.
982 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
984 2018-07-05 Tamar Christina <tamar.christina@arm.com>
987 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
988 instead of GET_MODE_SIZE when comparing Units.
990 2018-07-05 Tamar Christina <tamar.christina@arm.com>
993 * rtlanal.c (set_noop_p): Constrain on mode change,
994 include hard-reg-set.h
996 2018-07-05 Tamar Christina <tamar.christina@arm.com>
998 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
1000 2018-07-05 Jakub Jelinek <jakub@redhat.com>
1003 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
1006 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
1008 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
1010 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
1013 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
1015 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
1017 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
1020 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
1022 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
1024 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
1025 neg_*_op* variables.
1027 2018-07-04 Martin Liska <mliska@suse.cz>
1029 * tree-switch-conversion.c: Define
1030 max_ratio_for_speed and max_ratio_for_size constants.
1032 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
1033 Martin Liska <mliska@suse.cz>
1038 * common.opt: Rename align options with 'str_' prefix.
1039 * common/config/i386/i386-common.c (set_malign_value): New
1041 (ix86_handle_option): Use it to set -falign-* options/
1042 * config/aarch64/aarch64-protos.h (struct tune_params): Change
1043 type from int to string.
1044 * config/aarch64/aarch64.c: Update default values from int
1046 * config/alpha/alpha.c (alpha_override_options_after_change):
1048 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
1049 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
1050 max skip conditionally.
1051 * config/i386/freebsd.h (SUBALIGN_LOG): New.
1052 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
1053 max skip conditionally.
1054 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
1055 max skip conditionally.
1056 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
1057 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
1058 max skip conditionally.
1059 * config/i386/i386.c (struct ptt): Change type from int to
1061 (ix86_default_align): Set default values.
1062 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
1063 max skip conditionally.
1064 * config/i386/iamcu.h (SUBALIGN_LOG): New.
1065 (ASM_OUTPUT_MAX_SKIP_ALIGN):
1066 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
1067 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
1068 max skip conditionally.
1069 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
1070 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
1071 * config/i386/x86-64.h (SUBALIGN_LOG): New.
1072 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
1073 max skip conditionally.
1074 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
1075 * config/ia64/ia64.c (ia64_option_override): Set default values
1076 for alignment options.
1077 * config/m68k/m68k.c: Handle new str_align_* options.
1078 * config/mips/mips.c (mips_set_compression_mode): Change
1080 (mips_option_override): Set default values for options.
1081 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
1083 * config/rs6000/rs6000.c (rs6000_option_override_internal):
1085 * config/rx/rx.c (rx_option_override): Likewise.
1086 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
1087 (LABEL_ALIGN): Use align_labels_log.
1088 (LOOP_ALIGN): Use align_loops_align.
1089 * config/s390/s390.c (s390_asm_output_function_label): Use new
1091 * config/sh/sh.c (sh_override_options_after_change):
1092 Change type of constants.
1093 * config/spu/spu.c (spu_sched_init): Likewise.
1094 * config/sparc/sparc.c (sparc_option_override): Set default
1096 * config/visium/visium.c (visium_option_override): Likewise.
1097 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
1098 emit p2align format with last argument if it's not needed.
1099 * doc/invoke.texi: Document extended format of -falign-*.
1100 * final.c: Use align_labels alignment.
1101 * flags.h (struct target_flag_state): Change type to use
1103 (struct align_flags_tuple): New.
1104 (struct align_flags): Likewise.
1105 (align_loops_log): Redefine macro to use new types.
1106 (align_loops_max_skip): Redefine macro to use new types.
1107 (align_jumps_log): Redefine macro to use new types.
1108 (align_jumps_max_skip): Redefine macro to use new types.
1109 (align_labels_log): Redefine macro to use new types.
1110 (align_labels_max_skip): Redefine macro to use new types.
1111 (align_functions_log): Redefine macro to use new types.
1112 (align_loops): Redefine macro to use new types.
1113 (align_jumps): Redefine macro to use new types.
1114 (align_labels): Redefine macro to use new types.
1115 (align_functions): Redefine macro to use new types.
1116 (align_functions_max_skip): Redefine macro to use new types.
1117 (align_loops_value): New macro.
1118 (align_jumps_value): New macro.
1119 (align_labels_value): New macro.
1120 (align_functions_value): New macro.
1121 * function.c (invoke_set_current_function_hook): Propagate
1122 alignment values from flags to global variables default in
1124 * ipa-icf.c (sem_function::equals_wpa): Use
1125 cl_optimization_option_eq instead of memcmp.
1126 * lto-streamer.h (cl_optimization_stream_out): Support streaming
1128 (cl_optimization_stream_in): Likewise.
1129 * optc-save-gen.awk: Support strings in cl_optimization.
1130 * opth-gen.awk: Likewise.
1131 * opts.c (finish_options): Remove error checking of invalid
1133 (MAX_CODE_ALIGN): Remove.
1134 (MAX_CODE_ALIGN_VALUE): Likewise.
1135 (parse_and_check_align_values): New function.
1136 (check_alignment_argument): Likewise.
1137 (common_handle_option): Use check_alignment_argument.
1138 * opts.h (parse_and_check_align_values): Declare.
1139 * toplev.c (init_alignments): Remove.
1140 (read_log_maxskip): New.
1141 (parse_N_M): Likewise.
1142 (parse_alignment_opts): Likewise.
1143 (backend_init_target): Remove usage of init_alignments.
1144 * toplev.h (parse_alignment_opts): Declare.
1145 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
1147 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
1148 * tree.c (cl_option_hasher::equal): New.
1149 * varasm.c: Use new global macros.
1151 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
1153 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
1154 Use a simpler align directive also if MAXSKIP = ALIGN-1.
1155 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
1156 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
1157 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
1158 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
1159 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
1160 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
1161 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
1162 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
1163 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
1164 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
1165 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
1167 2018-07-04 Martin Liska <mliska@suse.cz>
1168 Jonathan Wakely <jwakely@redhat.com>
1170 * coverage.c: Use correct type.
1171 * doc/invoke.texi: Language correction.
1173 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
1176 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
1177 ENDBRANCH for non-tail call which may return via indirect branch.
1178 * doc/extend.texi: Document indirect_return attribute.
1180 2018-07-03 Martin Sebor <msebor@redhat.com>
1182 PR tree-optimization/86274
1183 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
1185 (format_floating): Correct handling of infinities and NaNs.
1187 2018-07-03 Martin Sebor <msebor@redhat.com>
1189 * print-tree.c (print_real_cst): New function.
1190 (print_node_brief): Call it.
1191 (print_node): Ditto.
1193 2018-07-03 Jeff Law <law@redhat.com>
1195 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
1196 into a single pattern.
1198 * config/h8300/h8300.md (ors code_iterator): New.
1199 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
1200 a single pattern and single splitter.
1201 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
1202 (iorqi3_1, xorqi3_1): Likewise.
1203 (iorqi3, xorqi3 expanders): Similarly.
1205 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
1206 (movmd_internal) into a single pattern using the P mode iterator.
1207 (movmd splitters): Similarly.
1208 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
1209 (movsd splitters): Similarly.
1211 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
1212 ADDB, ADDW and ADDL into a single ADD attribute which selects the
1213 right table based on the size of the operand.
1214 * config/h8300/h8300.md (length_table): Corresponding changes. All
1215 references to "addb", "addw" and "addl" changed to "add".
1216 (btst patterns): Merge two variants into a single pattern.
1217 (tstqi, tsthi): Likewise.
1218 (addhi3_incdec, addsi3_incdec): Likewise.
1219 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
1220 (mulhi3, mulsi3): Likewise.
1221 (udivhi3, udivsi3): Likewise.
1222 (divhi3, divsi3): Likewise.
1223 (andorqi3, andorhi3, andorsi3): Likewise.
1225 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
1228 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
1229 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
1231 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
1233 PR tree-optimization/85694
1234 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
1235 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
1237 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
1238 (<u>avg<mode>3_ceil): New patterns.
1240 2018-07-03 David Malcolm <dmalcolm@redhat.com>
1242 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
1243 scan-tree-dump directive.
1244 * gcc.dg/vect/slp-perm-2.c: Likewise.
1245 * gcc.dg/vect/slp-perm-3.c: Likewise.
1246 * gcc.dg/vect/slp-perm-5.c: Likewise.
1247 * gcc.dg/vect/slp-perm-6.c: Likewise.
1248 * gcc.dg/vect/slp-perm-7.c: Likewise.
1249 * gcc.dg/vect/slp-perm-8.c: Likewise.
1251 2018-07-03 Marek Polacek <polacek@redhat.com>
1254 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
1256 2018-07-03 Richard Biener <rguenther@suse.de>
1259 * tree-ssa-structalias.c (find_func_clobbers): Properly
1260 handle indirect calls.
1262 2018-07-03 Jeff Law <law@redhat.com>
1264 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
1265 (shifts): New code iterator.
1266 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
1267 expander. Fix HImode handling on H8/SX.
1268 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
1269 (subqi3, subhi3, subsi3 expanders): Likewise.
1270 (andqi3, andhi3, andsi3 expanders): Likewise.
1271 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
1272 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
1273 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
1274 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
1275 (zero_extendqihi2, zero_extendqisi2): Likewise.
1276 (extendqihi2, extendqisi2): Likewise.
1277 (rotlqi3, rotlhi3, rotlsi3): Likewise.
1278 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
1279 (rotlqi3_1, rotlhi3_1): Likewise.
1280 (logicalhi3_sn, logicalsi3_sn): Likewise.
1281 (logicalhi3, logicalsi3): Likewise.
1283 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
1285 * tree-vect-patterns.c (vect_recog_rotate_pattern)
1286 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
1287 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
1288 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
1289 type to append_pattern_def_seq instead of creating a stmt_vec_info
1291 (build_mask_conversion): Likewise. Remove vinfo argument.
1292 (vect_add_conversion_to_patterm): Likewise, renaming to...
1293 (vect_add_conversion_to_pattern): ...this.
1294 (vect_recog_mask_conversion_pattern): Update call to
1295 build_mask_conversion. Pass the vector type to
1296 append_pattern_def_seq here too.
1297 (vect_recog_gather_scatter_pattern): Update call to
1298 vect_add_conversion_to_pattern.
1300 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
1302 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
1303 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
1304 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
1305 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
1306 STMT_VINFO_PATTERN_DEF_SEQ to null here.
1307 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
1308 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
1309 append_pattern_def_seq instead of new_pattern_def_seq.
1310 (vect_recog_divmod_pattern): Do both of the above.
1311 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
1314 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
1316 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
1317 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
1318 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
1319 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
1320 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
1321 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
1322 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
1323 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
1324 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
1325 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
1326 parameter with a single stmt_vec_info.
1327 (vect_recog_func_ptr): Likewise.
1328 (vect_recog_gather_scatter_pattern): Likewise, folding in...
1329 (vect_try_gather_scatter_pattern): ...this.
1330 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
1331 the stmt_vec_info of the statement to be matched. Don't clear
1332 STMT_VINFO_RELATED_STMT.
1333 (vect_pattern_recog): Update call accordingly.
1335 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
1337 PR tree-optimization/85694
1338 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
1339 (uavgM3_ceil): Document new optabs.
1340 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
1341 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
1343 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
1344 (savg_ceil_optab): New optabs.
1345 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
1346 (vect_vect_recog_func_ptrs): Add it.
1347 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
1348 constant directly from the associated lhs.
1350 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
1352 * tree-vect-patterns.c (vect_split_statement): New function.
1353 (vect_convert_input): Use it to try to split an existing cast.
1355 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
1357 * poly-int.h (print_hex): New function.
1358 * dumpfile.h (dump_dec, dump_hex): Declare.
1359 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
1360 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
1361 min_input_precision, operation_precision and operation_sign.
1362 * tree-vect-patterns.c (vect_get_range_info): New function.
1363 (vect_same_loop_or_bb_p, vect_single_imm_use)
1364 (vect_operation_fits_smaller_type): Delete.
1365 (vect_look_through_possible_promotion): Add an optional
1366 single_use_p parameter.
1367 (vect_recog_over_widening_pattern): Rewrite to use new
1368 stmt_vec_info infomration. Handle one operation at a time.
1369 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
1370 (vect_truncatable_operation_p, vect_set_operation_type)
1371 (vect_set_min_input_precision): New functions.
1372 (vect_determine_min_output_precision_1): Likewise.
1373 (vect_determine_min_output_precision): Likewise.
1374 (vect_determine_precisions_from_range): Likewise.
1375 (vect_determine_precisions_from_users): Likewise.
1376 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
1377 (vect_vect_recog_func_ptrs): Put over_widening first.
1379 (vect_pattern_recog): Call vect_determine_precisions.
1381 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
1383 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
1384 statements that have been replaced by further pattern statements.
1385 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
1387 2018-07-03 Richard Biener <rguenther@suse.de>
1389 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
1390 always set *dt. Dump vectype in vectype overload.
1391 * dumpfile.h (dump_gimple_expr): New function.
1392 (dump_gimple_expr_loc): Likewise.
1393 * dumpfile.c (dump_gimple_expr): New function.
1394 (dump_gimple_expr_loc): Likewise.
1396 2018-07-02 Jeff Law <law@redhat.com>
1398 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
1399 the H8/300, H8/300H and H8/S variants into a single pattern.
1400 (movhi_h8300, movqi_h8300hs): Similarly.
1401 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
1402 (QHI mode iterator): New.
1404 * config/h8300/h8300.md: Remove trailing whitespace.
1406 2018-07-02 Jim Wilson <jimw@sifive.com>
1408 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
1409 instead of emit_insn for interrupt returns.
1410 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
1411 (riscv_sret, riscv_uret): Likewise.
1413 2018-07-02 David Malcolm <dmalcolm@redhat.com>
1415 * pretty-print.c (selftest::test_pp_format): Move save and restore
1416 of quotes to class auto_fix_quotes, and add an instance.
1417 * selftest.c: Include "intl.h".
1418 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
1419 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
1420 * selftest.h (selftest::auto_fix_quotes): New class.
1422 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
1424 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
1425 (aarch64_sve_prepare_conditional_op): Remove.
1426 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
1427 Allow aarch64_simd_reg_or_zero as select operand; remove
1428 the aarch64_sve_prepare_conditional_op call.
1429 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
1430 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
1431 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
1432 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
1433 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
1434 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
1435 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
1436 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
1437 and a splitters to match all of the *_any patterns.
1438 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
1440 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
1441 (SVE_COND_FP_BINARY_REV): Remove.
1442 (sve_int_op_rev, sve_fp_op_rev): New.
1443 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
1444 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
1445 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
1446 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
1447 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
1448 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
1449 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
1450 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
1451 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
1453 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
1454 Remove match_dup 1 from the inner unspec.
1455 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
1457 * config/aarch64/aarch64.md (movprfx): New attr.
1458 (length): Default movprfx to 8.
1459 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
1460 (*madd<SVE_I>, *msub<SVE_I): Likewise.
1461 (*<su>mul<SVE_I>3_highpart): Likewise.
1462 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
1463 (*v<ASHIFT><SVE_I>3): Likewise.
1464 (*<su><MAXMIN><SVE_I>3): Likewise.
1465 (*<su><MAXMIN><SVE_F>3): Likewise.
1466 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
1467 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
1468 (*div<SVE_F>4): Likewise.
1470 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
1472 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
1475 2018-07-02 Richard Biener <rguenther@suse.de>
1477 PR tree-optimization/86363
1478 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
1479 memset argument refers to a non-variable address.
1481 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
1483 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
1484 {PLUS,MINUS}_EXPR code to...
1485 (adjust_symbolic_bound): ...here,
1486 (combine_bound): ...here,
1487 (set_value_range_with_overflow): ...and here.
1489 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
1491 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
1493 (extract_range_from_abs_expr): ...here.
1495 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
1497 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
1498 -fno-omit-frame-pointer when not optimizing.
1500 2018-07-02 Martin Liska <mliska@suse.cz>
1503 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
1504 (propagate_nothrow): Likewise.
1506 2018-07-02 Martin Liska <mliska@suse.cz>
1509 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
1511 2018-07-02 David Malcolm <dmalcolm@redhat.com>
1513 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
1514 function in r262149, changing "loc" param from source_location to
1515 const dump_location_t &.
1516 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
1517 declaration, as above.
1519 2018-07-01 Paul Koning <ni1d@arrl.net>
1521 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
1522 -munit-asm, -mgnu-asm, -mdec-asm.
1523 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
1524 (pdp11_output_labelref): New.
1525 (pdp11_output_def): New.
1526 (pdp11_output_addr_vec_elt): New.
1527 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
1528 %# and %@ format codes.
1529 (pdp11_option_override): New.
1530 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
1531 (pdp11_output_ident): New.
1532 (pdp11_asm_named_section): New.
1533 (pdp11_asm_init_sections): New.
1534 (pdp11_file_start): New.
1535 (pdp11_file_end): New.
1536 (output_ascii): Use .ascii/.asciz for -mdec-asm.
1537 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
1538 %o, like %c but octal.
1539 (pdp11_option_override): New.
1540 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
1542 (DATA_SECTION_ASM_OP): Ditto.
1543 (READONLY_DATA_SECTION_ASM_OP): New.
1544 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
1545 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
1546 (ASM_OUTPUT_LABELREF): Ditto.
1547 (ASM_OUTPUT_DEF): Ditto.
1548 (ASM_OUTPUT_EXTERNAL): New.
1549 (ASM_OUTPUT_SOURCE_FILENAME): New.
1550 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
1551 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
1552 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
1553 %# and %@ format codes.
1554 * config/pdp11/pdp11.opt (mgnu-asm): New.
1555 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
1556 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
1557 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
1559 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
1561 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
1562 dereferencing path[] beyond its length.
1564 (debug_all_paths): New.
1565 (rewire_first_differing_edge): New.
1566 (adjust_paths_after_duplication): New.
1567 (duplicate_thread_path): Call adjust_paths_after_duplication.
1569 (thread_through_all_blocks): Add new argument to
1570 duplicate_thread_path.
1572 2018-06-30 Jim Wilson <jimw@sifive.com>
1574 * config/riscv/predicates.md (p2m1_shift_operand): New.
1575 (high_mask_shift_operand): New.
1576 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
1577 pattern using p2m1_shift_operand.
1578 (lshsi3_zero_extend_3+2): New combiner pattern using
1579 high_mask_shift_operand.
1581 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
1583 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
1585 (vect_recog_rotate_pattern): ...here.
1586 (vect_convert_input): Try to insert casts of invariants in the
1588 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
1589 preheader to be empty.
1591 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
1593 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
1594 vector type. If given, install it in the new statement's
1596 (vect_element_precision): New function.
1597 (vect_unpromoted_value): New struct.
1598 (vect_unpromoted_value::vect_unpromoted_value): New function.
1599 (vect_unpromoted_value::set_op): Likewise.
1600 (vect_look_through_possible_promotion): Likewise.
1601 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
1602 (vect_widened_op_tree, vect_convert_input): Likewise.
1603 (vect_convert_inputs, vect_convert_output): Likewise.
1604 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
1605 to handle the optional cast of the multiplication result and
1606 vect_widened_op_tree to detect the widened multiplication itself.
1607 Do not require the input and output of promotion casts to have
1608 the same sign, but base the signedness of the operation on the
1609 input rather than the result. If the pattern includes two
1610 promotions, check that those promotions have the same sign.
1611 Do not restrict the MULT_EXPR handling to a double-width result;
1612 handle quadruple-width results and wider. Use vect_convert_inputs
1613 to convert the inputs to the common type.
1614 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
1615 to handle the optional cast of the ABS result. Also allow a sign
1616 change or a sign extension between the ABS and MINUS.
1617 Use vect_widened_op_tree to detect the widened subtraction and use
1618 vect_convert_inputs to convert the inputs to the common type.
1619 (vect_handle_widen_op_by_const): Delete.
1620 (vect_recog_widen_op_pattern): New function.
1621 (vect_recog_widen_mult_pattern): Use it.
1622 (vect_recog_widen_shift_pattern): Likewise.
1623 (vect_recog_widen_sum_pattern): Use
1624 vect_look_through_possible_promotion to handle the promoted
1627 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
1629 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
1630 the containing gimple_seq *.
1631 * gimple-iterator.h (gsi_for_stmt): Declare it.
1632 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
1633 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
1634 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
1635 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
1636 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
1638 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
1640 (vect_mark_pattern_stmts): ...here. Handle cases in which the
1641 statement being replaced is part of an existing pattern
1642 definition sequence, inserting the new pattern statements before
1644 (vect_pattern_recog_1): Don't return a bool. If the statement
1645 is already part of a pattern, instead apply pattern matching
1646 to the pattern definition statements. Don't clear the
1647 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
1648 (vect_pattern_recog): Don't break after the first match;
1649 continue processing the pattern definition statements instead.
1650 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
1652 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
1654 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
1655 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
1656 (vect_recog_widen_sum_pattern): Use it.
1658 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
1660 * tree-vect-loop.c (vectorizable_reduction): Assert that the
1661 phi is not a pattern statement and has not been replaced by
1662 a pattern statement.
1663 * tree-vect-patterns.c (type_conversion_p): Don't check
1664 STMT_VINFO_IN_PATTERN_P.
1665 (vect_recog_vector_vector_shift_pattern): Likewise.
1666 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
1667 the pattern statement rather than the original statement; check
1668 directly for a WIDEN_MULT_EXPR here.
1669 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
1670 vect_is_simple_use to return the pattern statement rather
1671 than the original statement; use is_pattern_stmt_p to check
1672 for such a pattern statement.
1673 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
1674 to return the pattern statement rather than the original statement;
1675 don't do the same transformation here.
1676 (vect_is_simple_use): If the defining statement has been replaced
1677 by a pattern statement, return the pattern statement instead.
1678 Remove the corresponding (local) transformation from the vectype
1681 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
1683 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
1684 end and default to null.
1685 * tree-vect-loop.c (vect_create_epilog_for_reduction)
1686 (vectorizable_reduction): Update calls accordingly, dropping the
1687 gimple ** argument if the passed-back statement isn't needed.
1688 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
1689 (vect_recog_rotate_pattern): Likewise.
1690 (vect_recog_mask_conversion_pattern): Likewise.
1691 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
1692 (vect_mask_constant_operand_p): Likewise.
1693 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
1694 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
1695 (get_group_load_store_type, get_load_store_type): Likewise.
1696 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
1697 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
1698 (vectorizable_conversion, vectorizable_assignment): Likewise.
1699 (vectorizable_shift, vectorizable_operation): Likewise.
1700 (vectorizable_store, vect_is_simple_cond): Likewise.
1701 (vectorizable_condition, vectorizable_comparison): Likewise.
1702 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
1703 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
1704 and move it to the end. Cope with null def_stmt_outs.
1706 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
1708 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
1710 2018-06-29 Jeff Law <law@redhat.com>
1712 * config/v850/v850.c (v850_legitimate_address_p): Handle large
1713 displacements for TARGET_V850E2V3 and newer.
1714 (TARGET_LRA_P): Remove. Defaults to LRA now.
1715 * config/v850/v850.md (sign23byte_load): Remove.
1716 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
1717 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
1719 2018-06-29 Martin Liska <mliska@suse.cz>
1722 * coverage.c (coverage_init): Mangle full path name.
1723 * doc/invoke.texi: Document the change.
1724 * gcov-io.c (mangle_path): New.
1725 * gcov-io.h (mangle_path): Likewise.
1726 * gcov.c (mangle_name): Use mangle_path for path mangling.
1728 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1730 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
1731 if starting source register is not even.
1733 2018-06-29 Martin Liska <mliska@suse.cz>
1735 PR tree-optimization/86263
1736 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
1737 Make edge redirection.
1739 2018-06-29 David Malcolm <dmalcolm@redhat.com>
1741 * dumpfile.c (dump_loc): Add indentation based on scope depth.
1742 (dump_scope_depth): New variable.
1743 (get_dump_scope_depth): New function.
1744 (dump_begin_scope): New function.
1745 (dump_end_scope): New function.
1746 * dumpfile.h (get_dump_scope_depth): New declaration.
1747 (dump_begin_scope): New declaration.
1748 (dump_end_scope): New declaration.
1749 (class auto_dump_scope): New class.
1750 (AUTO_DUMP_SCOPE): New macro.
1751 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
1754 2018-06-29 Richard Biener <rguenther@suse.de>
1756 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
1757 compute_all_dependences succeeds.
1758 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
1759 exceed --param loop-max-datarefs-for-datadeps.
1761 2018-06-29 Jakub Jelinek <jakub@redhat.com>
1763 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
1765 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
1768 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
1769 alternative 0 in preferred_for_speed attribute.
1771 2018-06-28 Paul Koning <ni1d@arrl.net>
1773 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
1774 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
1775 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
1776 * config/pdp11/pdp11.md: Correct "length" attribute calculation
1777 for shift insn patterns.
1779 2018-06-28 David Malcolm <dmalcolm@redhat.com>
1781 * cgraph.c (cgraph_node::get_body): Replace assignments to
1782 "dump_file" with calls to set_dump_file.
1783 * dumpfile.c (alt_dump_file): Make static, and group with...
1784 (alt_flags): ...this definition.
1785 (dumps_are_enabled): New variable.
1786 (refresh_dumps_are_enabled): New function.
1787 (set_dump_file): New function.
1788 (set_alt_dump_file): New function.
1789 (gcc::dump_manager::dump_start): Replace assignments to
1790 "dump_file" and "alt_dump_file" with calls to set_dump_file and
1792 (gcc::dump_manager::dump_finish): Likewise.
1793 * dumpfile.h (alt_dump_file): Delete decl.
1794 (dumps_are_enabled): New variable decl.
1795 (set_dump_file): New function decl.
1796 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
1798 * tree-nested.c (lower_nested_functions): Replace assignments to
1799 "dump_file" with calls to set_dump_file.
1801 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
1803 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
1804 goto_locus of each outgoing edge of each basic block.
1806 2018-06-28 Richard Biener <rguenther@suse.de>
1808 * dwarf2out.c (decl_scope_table): Remove.
1809 (push_decl_scope): Likewise.
1810 (pop_decl_scope): Likewise.
1811 (gen_type_die_for_member): Do not call push/pop_decl_scope.
1812 (gen_struct_or_union_type_die): Likewise.
1813 (gen_tagged_type_die): Likewise.
1814 (dwarf2out_init): Do not initialize decl_scope_table.
1815 (dwarf2out_c_finalize): Do not free it.
1817 2018-06-28 Richard Biener <rguenther@suse.de>
1819 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
1820 deciding whether to not re-use a DIE.
1822 2018-06-28 Richard Biener <rguenther@suse.de>
1824 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
1825 DW_AT_abstract_origin attribute.
1827 2018-06-28 Martin Liska <mliska@suse.cz>
1829 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
1830 Use newly introduced constants.
1831 * tree-switch-conversion.h (struct jump_table_cluster):
1832 Define max_ratio_for_size and max_ratio_for_speed.
1834 2018-06-28 Martin Liska <mliska@suse.cz>
1836 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
1837 Add new checking assert to catch invalid state.
1838 (jump_table_cluster::can_be_handled): Handle single case
1840 (jump_table_cluster::is_beneficial): Bail out for such case.
1841 (bit_test_cluster::find_bit_tests):
1842 Add new checking assert to catch invalid state.
1843 (bit_test_cluster::can_be_handled): Handle single case
1845 (bit_test_cluster::is_beneficial): Bail out for such case.
1846 (switch_decision_tree::analyze_switch_statement):
1849 2018-06-28 Martin Liska <mliska@suse.cz>
1851 * common.opt: Introduce -completion option.
1852 * gcc.c (driver_handle_option): Handle it.
1853 (driver::main): Print completions if completion
1855 * opt-suggestions.c (option_proposer::get_completions):
1857 (option_proposer::suggest_completion): Likewise.
1858 (option_proposer::find_param_completions): Likewise.
1859 (verify_autocompletions): Likewise.
1860 (test_completion_valid_options): Likewise.
1861 (test_completion_valid_params): Likewise.
1862 (in_completion_p): Likewise.
1863 (empty_completion_p): Likewise.
1864 (test_completion_partial_match): Likewise.
1865 (test_completion_garbage): Likewise.
1866 (opt_proposer_c_tests): Likewise.
1867 * opt-suggestions.h: Declare new functions.
1868 * opts.c (common_handle_option): Handle OPT__completion_.
1869 * selftest-run-tests.c (selftest::run_tests): Add
1870 opt_proposer_c_tests.
1871 * selftest.c (assert_str_startswith): New.
1872 * selftest.h (assert_str_startswith): Likewise.
1873 (opt_proposer_c_tests): New.
1874 (ASSERT_STR_STARTSWITH): Likewise.
1876 2018-06-28 Martin Liska <mliska@suse.cz>
1878 * Makefile.in: Add opt-suggestions.o.
1879 * gcc-main.c: Include opt-suggestions.h.
1880 * gcc.c (driver::driver): Likewise.
1881 (driver::~driver): Remove m_option_suggestions.
1882 (driver::build_option_suggestions): Moved to option_proposer.
1883 (driver::suggest_option): Likewise.
1884 (driver::handle_unrecognized_options): Use option_proposer.
1885 * gcc.h (class driver): Add new memver m_option_proposer.
1886 * opt-suggestions.c: New file.
1887 * opt-suggestions.h: New file.
1889 2018-06-28 Martin Liska <mliska@suse.cz>
1891 * vec.h (class auto_string_vec): New (moved from auto_argvec).
1892 (auto_string_vec::~auto_string_vec): Likewise.
1894 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
1896 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
1897 prevent_decl_creation_for_types fields up and add reset_location field.
1898 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
1899 statement if id->reset_location is true.
1900 (copy_edges_for_bb): Do not set goto_locus on the new edges if
1901 id->reset_location is true.
1902 (copy_phis_for_bb): Force input_location on the arguments if
1903 id->reset_location is true.
1904 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
1905 is set on the function to be inlined.
1907 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
1909 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
1911 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
1913 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
1914 registers for Pmode.
1915 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
1916 hard registers for the clobbered pseudo.
1918 2018-06-27 Paul Koning <ni1d@arrl.net>
1920 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
1921 mutually exclusive options.
1922 * config/pdp11/constraints.md (h): New constraint.
1923 (O): Update definition to match shift code generation.
1924 (D): New constraint.
1925 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
1927 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
1929 (output_jump): Change arguments.
1930 (pdp11_fixed_cc_regs): New function.
1931 (pdp11_cc_mode): Ditto.
1932 (pdp11_expand_shift): Ditto.
1933 (pdp11_assemble_shift): Ditto.
1934 (pdp11_small_shift): Ditto.
1935 (pdp11_branch_cost): Remove.
1936 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
1938 (pdp11_register_move_cost): Update for CC registers.
1939 (pdp11_rtx_costs): Add case for LSHIFTRT.
1940 (pdp11_output_jump): Add CCNZ mode conditional branches.
1941 (notice_update_cc_on_set): Remove.
1942 (pdp11_cc_mode): New function.
1943 (simple_memory_operand): Correct pre/post decrement case.
1944 (no_side_effect_operand): New function.
1945 (pdp11_regno_reg_class): Add CC_REGS class.
1946 (pdp11_fixed_cc_regs): New function.
1947 (pdp11_small_shift): New function.
1948 (pdp11_expand_shift): New function to expand shift insns.
1949 (pdp11_assemble_shift): New function to output shifts.
1950 (pdp11_branch_cost): Remove.
1951 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
1952 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
1953 (WCHAR_TYPE): Ditto.
1954 (PTRDIFF_TYPE): Ditto.
1955 (ADJUST_INSN_LENGTH): New macro.
1956 (FIXED_REGISTERS): Add CC registers.
1957 (CALL_USED_REGISTERS): Ditto.
1959 (REG_CLASS_NAMES): Ditto.
1960 (REG_CLASS_CONTENTS): Ditto.
1961 (SELECT_CC_MODE): Use new function.
1962 (TARGET_FLAGS_REGNUM): New macro.
1963 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
1964 (cc0_reg_rtx): Remove.
1965 (CC_STATUS_MDEP): Remove.
1966 (CC_STATUS_MDEFP_INIT): Remove.
1967 (CC_IN_FPU): Remove.
1968 (NOTICE_UPDATE_CC): Remove.
1969 (REGISTER_NAMES): Add CC registers.
1970 (BRANCH_COST): Change to constant 1.
1971 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
1973 * config/pdp11/pdp11.opt (mbcopy): Remove.
1974 (mbcopy-builtin): Remove.
1975 (mbranch-cheap): Remove.
1976 (mbranch-expensive): Remove.
1977 * config/pdp11/predicates.md (expand_shift_operand): Update to
1978 match shift code generation.
1979 (ccnz_operator): New predicate.
1980 * doc/invoke.texi (PDP-11 Options): Remove deleted options
1981 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
1982 Remove non-existent option -mabshi, -mno-abshi. Document mutually
1984 * doc/md.texi (PDP-11): Document new D and h constraints. Update
1985 description of O constraint.
1987 2018-06-27 Jeff Law <law@redhat.com>
1988 Austin Law <austinklaw@gmail.com>
1990 * config/v850/v850.md (addsi3_set_flags): New pattern.
1991 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
1992 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
1993 (zero_extendhisi2_v850_set_flags): Likewise.
1994 (zero_extendqisi2_v850_set_flags): Likewise.
1995 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
1996 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
1997 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
1999 * config/v850/v850-protos.h (notice_update_cc): Remove.
2000 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
2001 (v850_print_operand): Handle 'D' and "d".
2002 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
2003 Add handling of arithmetic/logical operations compared against zero.
2004 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
2005 Do not look at v850_compare_op, instead get mode from last argument.
2006 (v850_gen_compare): Remove
2007 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
2008 after reload for prologue insns.
2009 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
2011 (construct_save_jarl): Likewise.
2012 (TARGET_FLAGS_REGNUM): Define.
2013 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
2014 (NOTICE_UPDATE_CC): Remove.
2015 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
2016 than cc0. Conditionalize on reload_completed.
2017 (cmpsi_insn, setfcc_insn): Likewise.
2018 (tst1 splitter): Turn into define_and_split which sets the flags
2020 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
2021 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
2022 (cstoresf4, cstoredf4): Clobber the flags.
2023 (cmpsi, cmpsf, cmpdf): Remove expanders.
2024 (setf_insn): Remove pattern.
2025 (addsi3): Turn into define_and_split which clobbers the flags after
2026 reload and a suitable pattern (addsi3_clobber_flags) for use after
2028 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
2029 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
2030 (ashrsi3, ashrsi3_v850e2): Likewise.
2031 (bins): Clobber the flags.
2032 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
2033 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
2034 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
2035 (call_value_internal_short, call_value_internal_long): Likewise.
2036 (callt_save_interrupt, callt_return_interrupt): Likewise.
2037 (save_interrupt, return_interrupt): Likewise.
2038 (callt_save_all_interrupt, save_all_interrupt): Likewise.
2039 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
2040 (restore_all_interrupt, _restore_all_interrupt): Likewise.
2041 (All FP comparisons): Only allow after reload has completed.
2043 (divh, divhu): Tweak output template.
2044 (branch_z_normal, branch_z_invert): Remove
2045 (branch_nz_normal, branch_nz_invert): Likewise.
2046 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
2048 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
2049 * config/v850/v850.c (notice_update_cc): Remove.
2050 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
2051 (CC_NO_CARRY): Likewise.
2052 (NOTICE_UPDATE_CC): Define to nothing.
2053 * config/v850/v850.md: Remove block comment on cc0 handling
2054 Remove "cc" attribute from all patterns. Remove cc_status handling
2055 from all patterns. Minor formatting fixes.
2057 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2059 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
2060 (cortex-a76.cortex-a55): Likewise.
2061 * config/aarch64/aarch64-tune.md: Regenerate.
2062 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
2063 cortex-a76.cortex-a55.
2065 2018-06-27 Jeff Law <law@redhat.com>
2067 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
2068 (MULTILIB_DIRNAMES): Similarly.
2070 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
2072 * gimple.h (gimple_return_retbnd): Delete.
2073 (gimple_return_set_retbnd): Likewise.
2074 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
2075 gimple_return_set_retbnd.
2076 * gimple-pretty-print.c (dump_gimple_return): Remove call to
2077 gimple_return_retbnd and adjust.
2078 * tree-inline.h (struct copy_body_data): Remove retbnd field.
2079 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
2080 Explicitly return NULL in a couple more cases. Move assertion
2081 on debug statements and remove unreachable code.
2082 (reset_debug_binding): Do not test id->retbnd.
2083 (expand_call_inline): Do not set it.
2085 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
2087 * configure.ac: Add --disable-gcov option.
2088 * configure: Regenerate.
2089 * Makefile.in: Honour @enable_gcov@.
2090 * doc/install.texi: Document --disable-gcov.
2092 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2094 * config/arm/arm-cpus.in (cortex-a76): New entry.
2095 (cortex-a76.cortex-a55): Likewise.
2096 * config/arm/arm-tables.opt: Regenerate.
2097 * config/arm/arm-tune.md: Likewise.
2098 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
2099 * doc/invoke.texi (ARM Options): Document cortex-a76 and
2100 cortex-a76.cortex-a55.
2102 2018-06-27 Tamar Christina <tamar.christina@arm.com>
2105 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
2107 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
2109 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
2111 (EPILOGUE_USES): Likewise.
2113 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
2115 * tree-inline.c (remap_location): New function extracted from...
2116 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
2117 (copy_phis_for_bb): ...here. Call remap_location.
2118 (copy_cfg_body): Adjust call to copy_edges_for_bb.
2120 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
2122 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
2123 unaligned vsx for 16B memset.
2125 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
2128 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
2129 ieee128_float_type_node to long_double_type_node unless
2130 TARGET_LONG_DOUBLE_128 is set.
2132 2018-06-26 David Malcolm <dmalcolm@redhat.com>
2134 * cfgloop.c (get_loop_location): Convert return type from
2135 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
2136 by implicit construction from rtx_insn *, and using
2137 dump_user_location_t::from_function_decl for the fallback case.
2138 * cfgloop.h (get_loop_location): Convert return type from
2139 location_t to dump_user_location_t.
2140 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
2141 dump_printf_loc to pass in a dump_location_t rather than a
2142 location_t, via the gimple stmt.
2143 * coverage.c (get_coverage_counts): Update calls to
2144 dump_printf_loc to pass in dump_location_t rather than a
2146 * doc/optinfo.texi (Dump types): Convert example of
2147 dump_printf_loc from taking "locus" to taking "insn". Update
2148 description of the "_loc" calls to cover dump_location_t.
2149 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
2151 (dump_user_location_t::dump_user_location_t): New constructors,
2152 from gimple *stmt and rtx_insn *.
2153 (dump_user_location_t::from_function_decl): New function.
2154 (dump_loc): Make static.
2155 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
2156 const dump_location_t &.
2157 (dump_generic_expr_loc): Delete.
2158 (dump_printf_loc): Convert param "loc" from location_t to
2159 const dump_location_t &.
2160 (selftest::test_impl_location): New function.
2161 (selftest::dumpfile_c_tests): New function.
2162 * dumpfile.h: Include "profile-count.h".
2163 (class dump_user_location_t): New class.
2164 (struct dump_impl_location_t): New struct.
2165 (class dump_location_t): New class.
2166 (dump_printf_loc): Convert 2nd param from source_location to
2167 const dump_location_t &.
2168 (dump_generic_expr_loc): Delete.
2169 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
2170 const dump_location_t &.
2171 * gimple-fold.c (fold_gimple_assign): Update call to
2172 dump_printf_loc to pass in a dump_location_t rather than a
2173 location_t, via the gimple stmt.
2174 (gimple_fold_call): Likewise.
2175 * gimple-loop-interchange.cc
2176 (loop_cand::analyze_iloop_reduction_var): Update for change to
2177 check_reduction_path.
2178 (tree_loop_interchange::interchange): Update for change to
2180 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
2181 change in return-type of find_loop_location.
2182 (graphite_regenerate_ast_isl): Likewise.
2183 * graphite-optimize-isl.c (optimize_isl): Likewise.
2184 * graphite.c (graphite_transform_loops): Likewise.
2185 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
2186 pass in a dump_location_t rather than a location_t, via the
2188 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
2189 * ipa.c (walk_polymorphic_call_targets): Likewise.
2190 * loop-unroll.c (report_unroll): Convert "locus" param from
2191 location_t to dump_location_t.
2192 (decide_unrolling): Update for change to get_loop_location's
2194 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
2195 location_t to dump_user_location_t.
2196 (grid_find_single_omp_among_assignments_1): Updates calls to
2197 dump_printf_loc to pass in a dump_location_t rather than a
2198 location_t, via the gimple stmt.
2199 (grid_parallel_clauses_gridifiable): Convert "tloc" from
2200 location_t to dump_location_t. Updates calls to dump_printf_loc
2201 to pass in a dump_location_t rather than a location_t, via the
2203 (grid_inner_loop_gridifiable_p): Likewise.
2204 (grid_dist_follows_simple_pattern): Likewise.
2205 (grid_gfor_follows_tiling_pattern): Likewise.
2206 (grid_target_follows_gridifiable_pattern): Likewise.
2207 (grid_attempt_target_gridification): Convert initialization
2208 of local "grid" from memset to zero-initialization; FIXME: does
2209 this require C++11? Update call to dump_printf_loc to pass in a
2210 optinfo_location rather than a location_t, via the gimple stmt.
2211 * profile.c (read_profile_edge_counts): Updates call to
2212 dump_printf_loc to pass in a dump_location_t rather than a
2214 (compute_branch_probabilities): Likewise.
2215 * selftest-run-tests.c (selftest::run_tests): Call
2217 * selftest.h (dumpfile_c_tests): New decl.
2218 * tree-loop-distribution.c (pass_loop_distribution::execute):
2219 Update for change in return type of find_loop_location.
2220 * tree-parloops.c (parallelize_loops): Likewise.
2221 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
2222 "locus" from location_t to dump_user_location_t.
2223 (canonicalize_loop_induction_variables): Likewise.
2224 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
2225 for change in return type of find_loop_location.
2226 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
2227 to dump_printf_loc to pass in a dump_location_t rather than a
2228 location_t, via the stmt.
2229 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
2231 * tree-vect-loop-manip.c (find_loop_location): Convert return
2232 type from source_location to dump_user_location_t.
2233 (vect_do_peeling): Update for above change.
2234 (vect_loop_versioning): Update for change in type of
2236 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
2237 from location_t to dump_user_location_t.
2238 (vect_estimate_min_profitable_iters): Update for change in type
2240 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
2241 location_t to dump_location_t.
2242 (vect_slp_bb): Update for change in type of vect_location.
2243 * tree-vectorizer.c (vect_location): Convert from source_location
2244 to dump_user_location_t.
2245 (try_vectorize_loop_1): Update for change in vect_location's type.
2246 (vectorize_loops): Likewise.
2247 (increase_alignment): Likewise.
2248 * tree-vectorizer.h (vect_location): Convert from source_location
2249 to dump_user_location_t.
2250 (find_loop_location): Convert return type from source_location to
2251 dump_user_location_t.
2252 (check_reduction_path): Convert 1st param from location_t to
2253 dump_user_location_t.
2254 * value-prof.c (check_counter): Update call to dump_printf_loc to
2255 pass in a dump_user_location_t rather than a location_t; update
2256 call to error_at for change in type of "locus".
2257 (check_ic_target): Update call to dump_printf_loc to
2258 pass in a dump_user_location_t rather than a location_t, via the
2261 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
2263 * config/s390/s390.h (enum processor_flags): Do not use
2264 default tune parameter when -march was specified.
2266 2018-06-26 Jakub Jelinek <jakub@redhat.com>
2269 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
2270 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
2273 2018-06-26 Richard Biener <rguenther@suse.de>
2275 PR tree-optimization/86287
2277 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
2278 (vect_analyze_loop): Initialize n_stmts.
2280 2018-06-26 Richard Biener <rguenther@suse.de>
2283 * fold-const.c (fold_convertible_p): Pointer extension
2286 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
2289 * dwarf2out.c (loc_list_has_views): Adjust comments.
2290 (dw_loc_list): Split single cross-partition range with
2293 2018-06-25 Jeff Law <law@redhat.com>
2295 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
2296 on -mbig-switch by default.
2298 * config/v850/predicates.md (const_float_1_operand): Fix match_code
2300 (const_float_0_operand): Remove unused predicate.
2301 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
2302 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
2303 (recipsf2): New expander. Original pattern now called
2305 (recipdf2, recipdf2_insn): Similarly.
2306 (rsqrtsf2, rsqrtsf2_insn): Similarly
2307 (rsqrtdf2, rsqrtdf2_insn): Similarly
2309 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
2311 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
2312 Simplify logic for FreeBSD (twice).
2314 2018-06-25 Martin Sebor <msebor@redhat.com>
2316 PR tree-optimization/86204
2317 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
2318 a strnlen result if it's less than the length of the string.
2320 2018-06-25 Martin Sebor <msebor@redhat.com>
2322 PR tree-optimization/85700
2323 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
2324 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
2325 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
2327 2018-06-25 Martin Sebor <msebor@redhat.com>
2329 * doc/extend.texi (Zero-length arrays): Update and clarify.
2331 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
2333 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
2334 added IEEE/IBM long double multilib support on PowerPC little
2335 endian Linux systems.
2336 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
2337 (MULTILIB_DEFAULTS): Likewise.
2338 * config/rs6000/rs6000.c (rs6000_option_override_internal):
2340 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
2341 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
2342 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
2344 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
2347 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
2348 (REORDER_45): Likewise.
2350 2018-06-25 Jeff Law <law@redhat.com>
2352 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
2353 dividend to 32 bits. Adjust length.
2354 (udivmodhi4): Cleanup output template. Fix length.
2356 2018-06-25 Carl Love <cel@us.ibm.com>
2358 * config/rs6000/vsx.md: Change word selector to prefered location.
2360 2018-06-25 Richard Biener <rguenther@suse.de>
2362 PR tree-optimization/86304
2363 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
2364 epilogue-if-converted loops as well.
2366 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
2368 * lto-section-out.c (lto_begin_section): Do not print section
2369 name for noaddr and unnumbered dumps.
2371 2018-06-25 Richard Biener <rguenther@suse.de>
2373 * tree-vectorizer.h (struct vec_info_shared): New structure
2374 with parts split out from struct vec_info and loop_nest from
2375 struct _loop_vec_info.
2376 (struct vec_info): Adjust accordingly.
2377 (struct _loop_vec_info): Likewise.
2378 (LOOP_VINFO_LOOP_NEST): Adjust.
2379 (LOOP_VINFO_DATAREFS): Likewise.
2380 (LOOP_VINFO_DDRS): Likewise.
2381 (struct _bb_vec_info): Likewise.
2382 (BB_VINFO_DATAREFS): Likewise.
2383 (BB_VINFO_DDRS): Likewise.
2384 (struct _stmt_vec_info): Add dr_aux member.
2385 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
2386 (DR_MISALIGNMENT_UNINITIALIZED): New.
2387 (set_dr_misalignment): Adjust.
2388 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
2389 (vect_analyze_loop): Adjust prototype.
2390 (vect_analyze_loop_form): Likewise.
2391 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
2392 Compute dependences lazily.
2393 (vect_record_base_alignments): Use shared datarefs/ddrs.
2394 (vect_verify_datarefs_alignment): Likewise.
2395 (vect_analyze_data_refs_alignment): Likewise.
2396 (vect_analyze_data_ref_accesses): Likewise.
2397 (vect_analyze_data_refs): Likewise.
2398 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
2399 constructor parameter for shared part.
2400 (vect_analyze_loop_form): Pass in shared part and adjust.
2401 (vect_analyze_loop_2): Pass in storage for the number of
2402 stmts. Move loop nest finding to the caller. Compute
2404 (vect_analyze_loop): Pass in shared part.
2405 (vect_transform_loop): Verify shared datarefs are unchanged.
2406 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
2407 constructor parameter for shared part.
2408 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
2409 (vect_slp_bb): Verify shared datarefs are unchanged before
2411 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
2413 (new_stmt_vec_info): Initialize DR_AUX misalignment to
2414 DR_MISALIGNMENT_UNINITIALIZED.
2415 * tree-vectorizer.c (vec_info::vec_info): Add constructor
2416 parameter for shared part.
2417 (vec_info::~vec_info): Adjust.
2418 (vec_info_shared::vec_info_shared): New.
2419 (vec_info_shared::~vec_info_shared): Likewise.
2420 (vec_info_shared::save_datarefs): Likewise.
2421 (vec_info_shared::check_datarefs): Likewise.
2422 (try_vectorize_loop_1): Construct shared part live for analyses
2423 of a single loop for multiple vector sizes.
2424 * tree-parloops.c (gather_scalar_reductions): Adjust.
2426 2018-06-25 Richard Biener <rguenther@suse.de>
2428 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
2429 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
2430 (vect_analyze_data_refs): Remove similar code from here and
2431 simplify accordingly.
2433 2018-06-25 Richard Biener <rguenther@suse.de>
2435 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
2436 for reverse storage order accesses rather than asserting
2437 they cannot happen here.
2439 2018-06-25 Tom de Vries <tdevries@suse.de>
2442 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
2443 Use data16 instead of .byte for insn prefix.
2445 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
2448 * parser.c (make_char_string_pack): Pass this literal chars
2449 through cpp_interpret_string.
2450 (cp_parser_userdef_numeric_literal): Check the result of
2451 make_char_string_pack.
2453 2018-06-24 Maya Rashish <coypu@sdf.org>
2455 * ginclude/stddef.h: Simplify conditions around avoiding
2456 re-definition of __size_t.
2458 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
2460 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
2461 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
2463 2018-06-22 Maya Rashish <coypu@sdf.org>
2465 * doc/invoke.texi (mno-fancy-math-387): Update for changes
2466 made to OpenBSD and NetBSD through the years.
2468 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
2470 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
2471 behavior of vec_pack (vector double, vector double) to match
2472 behavior of vec_float2 (vector double, vector double).
2474 2018-06-22 Olivier Hainque <hainque@adacore.com>
2476 * gimplify.c (gimplify_function_tree): Prevent creation
2477 of a trampoline for the address of the current function
2478 passed to entry/exit instrumentation hooks.
2480 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
2483 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
2486 2018-06-22 Martin Liska <mliska@suse.cz>
2488 PR tree-optimization/86263
2489 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
2490 Bail out if is_enabled is false.
2491 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
2493 (jump_table_cluster::is_enabled): New function.
2495 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
2497 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
2498 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
2499 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
2500 (lto_input_ts_binfo_tree_pointers): Likewise.
2501 * tree-streamer-out.c (streamer_write_tree_bitfields,
2502 write_ts_binfo_tree_pointers): Likewise.
2503 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
2505 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
2507 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
2509 2018-06-22 Martin Liska <mliska@suse.cz>
2511 * symbol-summary.h (get): Make it pure and inline move
2512 functionality from ::get function.
2513 (get): Remove and inline into ::get and ::get_create.
2514 (get_create): Move code from ::get function.
2516 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2519 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
2520 -x assembler-with-cpp.
2522 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2524 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
2525 _FILE_OFFSET_BITS=64 for C++.
2527 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
2529 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
2530 conversion insn that shows up when pr85657-3.c is compiled using
2531 IEEE 128-bit long double.
2532 (neg<mode>2_internal): Use the correct mode to check whether the
2533 mode is IBM extended.
2534 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
2535 multiply and divide external functions from being created more
2538 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
2540 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
2542 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
2543 the edge can be forwarded.
2544 (cfg_layout_merge_blocks): Likewise.
2546 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
2548 * except.c (finish_eh_generation): Commit edge insertions only after
2549 the EH edges have been redirected from post-landing to landing pads.
2551 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
2553 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
2554 create_tmp_var_for to create the FRAME decl.
2555 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
2557 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
2559 * tree-inline.c (copy_edges_for_bb): Minor tweak.
2560 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
2561 debug statement when resetting its value.
2562 (expand_call_inline): Copy the locus of the call onto the assignment
2563 of the return value, if any. Use local variable in more cases.
2565 2018-06-21 Martin Liska <mliska@suse.cz>
2567 * ipa-pure-const.c (propagate_nothrow): Use
2568 funct_state_summaries->get.
2569 (dump_malloc_lattice): Likewise.
2570 (propagate_malloc): Likewise.
2572 2018-06-21 Richard Biener <rguenther@suse.de>
2574 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
2575 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
2576 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
2578 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
2579 BLOCK_ABSTRACT_ORIGIN unconditionally.
2581 2018-06-21 David Malcolm <dmalcolm@redhat.com>
2583 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
2585 * ipa-reference.c (ipa_reference_c_finalize): Delete
2586 ipa_ref_opt_sum_summaries and set it to NULL.
2588 2018-06-21 Tom de Vries <tdevries@suse.de>
2590 PR tree-optimization/85859
2591 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
2592 test with comment from bb_no_side_effects_p.
2594 2018-06-21 Richard Biener <rguenther@suse.de>
2596 PR tree-optimization/86232
2597 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
2598 max for constant niter.
2600 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
2602 * config/aarch64/aarch64-simd.md
2603 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
2605 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
2607 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
2608 Make opernads of the unspec commutative.
2610 2018-06-21 Richard Biener <rguenther@suse.de>
2612 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
2613 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
2614 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
2615 (vect_analyze_data_ref_dependence): Re-order checks to deal with
2617 (vect_record_base_alignments): Do not record base alignment
2618 for gathers or scatters.
2619 (vect_compute_data_ref_alignment): Drop return value that is always
2620 true. Bail out early for gathers or scatters.
2621 (vect_enhance_data_refs_alignment): Bail out early for gathers
2623 (vect_find_same_alignment_drs): Likewise.
2624 (vect_analyze_data_refs_alignment): Remove dead code.
2625 (vect_slp_analyze_and_verify_node_alignment): Likewise.
2626 (vect_analyze_data_refs): For possible gathers or scatters do
2627 not create an alternate DR, just check their possible validity
2628 and mark them. Adjust DECL_NONALIASED handling to not rely
2630 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
2631 update inits of gathers or scatters.
2632 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
2633 Also copy gather/scatter flag to pattern vinfo.
2635 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
2637 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
2638 behavior of vec_packsu (vector unsigned long long, vector unsigned
2639 long long) to match behavior of vec_packs with same signature.
2641 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
2642 Thomas Schwinge <thomas@codesourcery.com>
2643 Cesar Philippidis <cesar@codesourcery.com>
2645 * gimplify.c (gimplify_scan_omp_clauses): Add support for
2646 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
2647 (gimplify_adjust_omp_clauses): Likewise.
2648 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
2649 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
2650 (gimplify_omp_target_update): Update handling of acc update and
2652 * omp-low.c (install_var_field): Remove unused parameter
2653 base_pointers_restrict.
2654 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
2655 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
2657 (omp_target_base_pointers_restrict_p): Delete.
2658 (scan_omp_target): Update call to scan_sharing_clauses.
2659 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
2661 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
2662 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
2663 (convert_local_omp_clauses): Likewise.
2664 * tree-pretty-print.c (dump_omp_clause): Likewise.
2665 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
2667 (omp_clause_code_name): Likewise.
2669 2018-06-20 Jakub Jelinek <jakub@redhat.com>
2672 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
2675 PR tree-optimization/86231
2676 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
2677 anti-range don't overwrite *vr0min before using it to compute *vr0max.
2679 2018-06-20 Tom de Vries <tdevries@suse.de>
2681 PR tree-optimization/86097
2682 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
2683 iv type if signedness of iv type is not the same as that of *nit.
2685 2018-06-20 Jakub Jelinek <jakub@redhat.com>
2687 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
2688 EDGE_EH edges, verify they are all EDGE_EH.
2690 2018-06-20 Maya Rashish <coypu@sdf.org>
2692 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
2694 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2696 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
2697 * config/aarch64/aarch64.c (xgene1_tunings): Add
2698 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
2699 (aarch64_mode_valid_for_sched_fusion_p):
2700 Allow 16-byte modes.
2701 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
2702 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
2704 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
2706 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
2707 * config/aarch64/iterators.md (VQ2): New mode iterator.
2709 2018-06-20 Martin Liska <mliska@suse.cz>
2711 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
2712 Change default ratio from 10 to 8.
2714 2018-06-20 Martin Liska <mliska@suse.cz>
2716 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
2718 (bit_test_cluster::find_bit_tests): Likewise.
2719 (switch_decision_tree::analyze_switch_statement): Find clusters.
2720 * tree-switch-conversion.h (struct jump_table_cluster): Document
2723 2018-06-20 Martin Liska <mliska@suse.cz>
2725 * tree-switch-conversion.c (switch_conversion::collect):
2726 Record m_uniq property.
2727 (switch_conversion::expand): Bail out for special conditions.
2728 (group_cluster::~group_cluster): New.
2729 (group_cluster::group_cluster): Likewise.
2730 (group_cluster::dump): Likewise.
2731 (jump_table_cluster::emit): New.
2732 (switch_decision_tree::fix_phi_operands_for_edges): New.
2733 (struct case_node): Remove struct.
2734 (jump_table_cluster::can_be_handled): New.
2735 (case_values_threshold): Moved to header.
2736 (reset_out_edges_aux): Likewise.
2737 (jump_table_cluster::is_beneficial): New.
2738 (bit_test_cluster::can_be_handled): Likewise.
2739 (add_case_node): Remove.
2740 (bit_test_cluster::is_beneficial): New.
2741 (case_bit_test::cmp): New.
2742 (bit_test_cluster::emit): New.
2743 (expand_switch_as_decision_tree_p): Remove.
2744 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
2745 (fix_phi_operands_for_edge): Likewise.
2746 (switch_decision_tree::analyze_switch_statement): New.
2747 (compute_cases_per_edge): Move ...
2748 (switch_decision_tree::compute_cases_per_edge): ... here.
2749 (try_switch_expansion): Likewise.
2750 (switch_decision_tree::try_switch_expansion): Likewise.
2751 (record_phi_operand_mapping): Likewise.
2752 (switch_decision_tree::record_phi_operand_mapping): Likewise.
2753 (emit_case_decision_tree): Likewise.
2754 (switch_decision_tree::emit): Likewise.
2755 (balance_case_nodes): Likewise.
2756 (switch_decision_tree::balance_case_nodes): Likewise.
2757 (dump_case_nodes): Likewise.
2758 (switch_decision_tree::dump_case_nodes): Likewise.
2759 (emit_jump): Likewise.
2760 (switch_decision_tree::emit_jump): Likewise.
2761 (emit_cmp_and_jump_insns): Likewise.
2762 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
2763 (emit_case_nodes): Likewise.
2764 (switch_decision_tree::emit_case_nodes): Likewise.
2765 (conditional_probability): Remove.
2766 * tree-switch-conversion.h (enum cluster_type): New.
2768 (struct cluster): Likewise.
2769 (cluster::cluster): Likewise.
2770 (struct simple_cluster): Likewise.
2771 (simple_cluster::simple_cluster): Likewise.
2772 (struct group_cluster): Likewise.
2773 (struct jump_table_cluster): Likewise.
2774 (struct bit_test_cluster): Likewise.
2775 (struct min_cluster_item): Likewise.
2776 (struct case_tree_node): Likewise.
2777 (case_tree_node::case_tree_node): Likewise.
2778 (jump_table_cluster::case_values_threshold): Likewise.
2779 (struct case_bit_test): Likewise.
2780 (struct switch_decision_tree): Likewise.
2781 (struct switch_conversion): Likewise.
2782 (switch_decision_tree::reset_out_edges_aux): Likewise.
2784 2018-06-20 Martin Liska <mliska@suse.cz>
2786 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
2787 (hoist_edge_and_branch_if_true): Likewise.
2788 (expand_switch_using_bit_tests_p): Likewise.
2789 (struct case_bit_test): Likewise.
2790 (case_bit_test_cmp): Likewise.
2791 (emit_case_bit_tests): Likewise.
2792 (switch_conversion::switch_conversion): New class.
2793 (struct switch_conv_info): Remove old struct.
2794 (collect_switch_conv_info): More to ...
2795 (switch_conversion::collect): ... this.
2796 (check_range): Likewise.
2797 (switch_conversion::check_range): Likewise.
2798 (check_all_empty_except_final): Likewise.
2799 (switch_conversion::check_all_empty_except_final): Likewise.
2800 (check_final_bb): Likewise.
2801 (switch_conversion::check_final_bb): Likewise.
2802 (create_temp_arrays): Likewise.
2803 (switch_conversion::create_temp_arrays): Likewise.
2804 (free_temp_arrays): Likewise.
2805 (gather_default_values): Likewise.
2806 (switch_conversion::gather_default_values): Likewise.
2807 (build_constructors): Likewise.
2808 (switch_conversion::build_constructors): Likewise.
2809 (constructor_contains_same_values_p): Likewise.
2810 (switch_conversion::contains_same_values_p): Likewise.
2811 (array_value_type): Likewise.
2812 (switch_conversion::array_value_type): Likewise.
2813 (build_one_array): Likewise.
2814 (switch_conversion::build_one_array): Likewise.
2815 (build_arrays): Likewise.
2816 (switch_conversion::build_arrays): Likewise.
2817 (gen_def_assigns): Likewise.
2818 (switch_conversion::gen_def_assigns): Likewise.
2819 (prune_bbs): Likewise.
2820 (switch_conversion::prune_bbs): Likewise.
2821 (fix_phi_nodes): Likewise.
2822 (switch_conversion::fix_phi_nodes): Likewise.
2823 (gen_inbound_check): Likewise.
2824 (switch_conversion::gen_inbound_check): Likewise.
2825 (process_switch): Use the newly created class.
2826 (switch_conversion::expand): New.
2827 (switch_conversion::~switch_conversion): New.
2828 * tree-switch-conversion.h: New file.
2830 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
2832 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
2833 tree-vect-patterns.c.
2834 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
2835 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
2836 (vect_recog_sad_pattern): Likewise.
2837 (vect_recog_widen_sum_pattern): Likewise.
2838 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
2839 (vect_recog_widen_shift_pattern): Remove the type_in argument.
2840 (vect_recog_rotate_pattern): Likewise.
2841 (vect_recog_mult_pattern): Likewise.
2842 (vect_recog_vector_vector_shift_pattern): Likewise.
2843 (vect_recog_divmod_pattern): Likewise.
2844 (vect_recog_mixed_size_cond_pattern): Likewise.
2845 (vect_recog_bool_pattern): Likewise.
2846 (vect_recog_mask_conversion_pattern): Likewise.
2847 (vect_try_gather_scatter_pattern): Likewise.
2848 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
2849 (vect_recog_over_widening_pattern): Likewise.
2850 (vect_recog_gather_scatter_pattern): Likewise.
2851 (vect_recog_func_ptr): Move from tree-vectorizer.h
2852 (vect_vect_recog_func_ptrs): Move further down the file.
2853 (vect_recog_func): Likewise. Remove the third argument.
2854 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
2855 (vect_pattern_recog_1): Expect the pattern function to do any
2856 necessary target tests. Also expect it to provide a vector type.
2857 Remove the type_in handling.
2859 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
2861 * tree-vect-patterns.c (vect_pattern_detected): New function.
2862 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
2863 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
2864 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
2865 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
2866 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
2867 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
2868 (vect_recog_mask_conversion_pattern)
2869 (vect_try_gather_scatter_pattern): Likewise.
2871 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
2873 * tree-vect-patterns.c (vect_get_internal_def): New function.
2874 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
2875 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
2876 (search_type_for_mask_1): Use it.
2878 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
2880 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
2881 redundant WIDEN_SUM_EXPR handling.
2882 (vect_recog_sad_pattern): Likewise.
2884 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
2886 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
2887 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
2888 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
2889 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
2890 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
2892 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
2894 * tree-vect-stmts.c (vectorizable_call): Make sure that we
2895 use the stmt_vec_info of the original bb statement for the
2896 new zero assignment, even if the call is part of a pattern.
2898 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
2900 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
2901 that the sequence is attached to the original statement rather
2902 than the pattern statement.
2903 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
2904 PATTERN_DEF_SEQ from the original statement rather than
2905 the main pattern statement.
2906 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
2907 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
2908 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
2910 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
2912 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
2913 definition statements before the early exit for statements that aren't
2915 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
2917 (vect_transform_loop): ...here. Process pattern definition
2918 statements without first checking whether the main pattern
2919 statement is live or relevant.
2921 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
2923 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
2924 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
2926 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
2928 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
2929 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
2930 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
2931 (expand_block_compare): Change select_block_compare_mode call.
2932 (expand_strncmp_align_check): Use new functions, fix comment.
2933 (emit_final_str_compare_gpr): New function.
2934 (expand_strn_compare): Refactor and clean up code.
2935 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
2937 2018-06-19 Tony Reix <tony.reix@atos.com>
2938 Damien Bergamini <damien.bergamini@atos.com>
2939 David Edelsohn <dje.gcc@gmail.com>
2941 * collect2.c (static_obj): New variable.
2942 (static_libs): New variable.
2943 (is_in_list): Uncomment declaration.
2944 (main): Track AIX libraries linked statically.
2945 (is_in_list): Uncomment definition.
2946 (scan_prog_file): Don't add AIX shared libraries initializer
2947 to constructor list if linking statically.
2949 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
2951 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
2953 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
2955 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
2957 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
2960 2018-06-19 Martin Liska <mliska@suse.cz>
2962 * config/i386/i386.c (ix86_can_inline_p): Do not use
2963 ipa_fn_summaries::get_create.
2964 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
2966 (devirtualization_time_bonus): Likewise.
2967 (ipcp_propagate_stage): Likewise.
2968 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
2969 (edge_set_predicate): Likewise.
2970 (evaluate_conditions_for_known_args): Likewise.
2971 (evaluate_properties_for_edge): Likewise.
2972 (ipa_call_summary::reset): Tranform to ...
2973 (ipa_call_summary::~ipa_call_summary): ... this.
2974 (ipa_fn_summary::reset): Transform to ...
2975 (ipa_fn_summary::~ipa_fn_summary): ... this.
2976 (ipa_fn_summary_t::remove): Rename to ...
2977 (ipa_fn_summary_t::remove_callees): ... this.
2978 (ipa_fn_summary_t::duplicate): Use placement new
2979 instead of memory copy.
2980 (ipa_call_summary_t::duplicate): Likewise.
2981 (ipa_call_summary_t::remove): Remove.
2982 (dump_ipa_call_summary): Change get_create to get.
2983 (ipa_dump_fn_summary): Dump only when summary exists.
2984 (analyze_function_body): Use symbol_summary::get instead
2986 (compute_fn_summary): Likewise.
2987 (estimate_edge_devirt_benefit): Likewise.
2988 (estimate_edge_size_and_time): Likewise.
2989 (inline_update_callee_summaries): Likewise.
2990 (remap_edge_change_prob): Likewise.
2991 (remap_edge_summaries): Likewise.
2992 (ipa_merge_fn_summary_after_inlining): Likewise.
2993 (write_ipa_call_summary): Likewise.
2994 (ipa_fn_summary_write): Likewise.
2995 (ipa_free_fn_summary): Likewise.
2996 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
2997 (struct ipa_call_summary): Likewise.
2998 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
3000 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
3001 (estimate_size_after_inlining): Likewise.
3002 (estimate_growth): Likewise.
3003 (growth_likely_positive): Likewise.
3004 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
3005 (inline_call): Likewise.
3006 * ipa-inline.c (caller_growth_limits): Likewise.
3007 (can_inline_edge_p): Likewise.
3008 (can_inline_edge_by_limits_p): Likewise.
3009 (compute_uninlined_call_time): Likewise.
3010 (compute_inlined_call_time): Likewise.
3011 (want_inline_small_function_p): Likewise.
3012 (edge_badness): Likewise.
3013 (update_caller_keys): Likewise.
3014 (update_callee_keys): Likewise.
3015 (inline_small_functions): Likewise.
3016 (inline_to_all_callers_1): Likewise.
3017 (dump_overall_stats): Likewise.
3018 (early_inline_small_functions): Likewise.
3019 (early_inliner): Likewise.
3020 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
3021 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
3022 * ipa-pure-const.c (malloc_candidate_p): Likewise.
3023 * ipa-split.c (execute_split_functions): Likewise.
3024 * symbol-summary.h: Likewise.
3025 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
3027 2018-06-19 Richard Biener <rguenther@suse.de>
3029 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
3030 (vectorize_loops): ... here. Fix dbgcnt handling.
3031 (try_vectorize_loop): Wrap try_vectorize_loop_1.
3033 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
3036 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
3037 ieee128 argument takes up only one (vector) register, not two (floating
3040 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
3042 * gimplify.c (gimplify_init_constructor): Really never clear for an
3043 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
3045 2018-06-19 Richard Biener <rguenther@suse.de>
3047 PR tree-optimization/86179
3048 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
3049 after failed recognition.
3051 2018-06-18 Martin Sebor <msebor@redhat.com>
3054 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
3055 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
3056 Handle integer subtraction.
3057 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
3058 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
3060 2018-06-18 David Malcolm <dmalcolm@redhat.com>
3062 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
3063 param from rtx to rtx_insn *.
3064 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
3066 (frv_ifcvt_modify_insn): Likwise.
3067 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
3068 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
3069 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
3070 as_a <rtx_insn *> cast to local "unprotected_region" once
3071 it's been established that it's not NULL or pc_rtx.
3072 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
3073 param "sethi" from rtx to rtx_insn *.
3074 (nds32_group_float_insns): Likewise for param "insn".
3075 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
3077 (vax_output_int_subtract): Likewise.
3078 * config/vax/vax.c (vax_output_int_add): Likewise for param
3080 (vax_output_int_subtract): Likewise.
3081 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
3082 (emit_pattern_after): Likewise for param "after".
3083 (emit_insn_after): Likewise.
3084 (emit_jump_insn_after): Likewise.
3085 (emit_call_insn_after): Likewise.
3086 (emit_debug_insn_after): Likewise.
3087 (emit_pattern_before): Likewise for param "before".
3088 (emit_insn_before): Likewise.
3089 (emit_jump_insn_before): Likewise.
3090 * final.c (get_insn_template): Likewise for param "insn", removing
3092 * output.h (get_insn_template): Likewise for 2nd param.
3093 * rtl.h (emit_insn_before): Likewise.
3094 (emit_jump_insn_before): Likewise.
3095 (emit_debug_insn_before_noloc): Likewise.
3096 (emit_insn_after): Likewise.
3097 (emit_jump_insn_after): Likewise.
3098 (emit_call_insn_after): Likewise.
3099 (emit_debug_insn_after): Likewise.
3100 (set_insn_deleted): Likewise for param.
3102 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
3105 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
3106 floating point modes, so that IFmode is numerically greater than
3107 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
3108 to declare the ordering. This prevents IFmode from being
3109 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
3110 machine. Include rs6000-modes.h to share the fractional values
3111 between genmodes* and the rest of the compiler.
3115 * config/rs6000/rs6000-modes.h: New file.
3116 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
3117 meaning of rs6000_long_double_size so that 126..128 selects an
3118 appropriate 128-bit floating point type.
3119 (rs6000_option_override_internal): Likewise.
3120 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
3121 (TARGET_LONG_DOUBLE_128): Change the meaning of
3122 rs6000_long_double_size so that 126..128 selects an appropriate
3123 128-bit floating point type.
3124 (LONG_DOUBLE_TYPE_SIZE): Update comment.
3125 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
3126 source and destination to match the standard usage.
3127 (truncifkf2): Likewise.
3128 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
3129 ISA 2.07 to use an explicit clobber, instead of passing in a
3131 (copysign<mode>3_soft): Likewise.
3133 2018-06-18 David Malcolm <dmalcolm@redhat.com>
3135 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
3136 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
3137 (vect_slp_analyze_instance_dependence): Likewise.
3138 (vect_enhance_data_refs_alignment): Likewise.
3139 (vect_analyze_data_refs_alignment): Likewise.
3140 (vect_slp_analyze_and_verify_instance_alignment
3141 (vect_analyze_data_ref_accesses): Likewise.
3142 (vect_prune_runtime_alias_test_list): Likewise.
3143 (vect_analyze_data_refs): Likewise.
3144 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
3145 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
3146 (vect_analyze_scalar_cycles_1): Likewise.
3147 (vect_get_loop_niters): Likewise.
3148 (vect_analyze_loop_form_1): Likewise.
3149 (vect_update_vf_for_slp): Likewise.
3150 (vect_analyze_loop_operations): Likewise.
3151 (vect_analyze_loop): Likewise.
3152 (vectorizable_induction): Likewise.
3153 (vect_transform_loop): Likewise.
3154 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
3155 * tree-vect-slp.c (vect_analyze_slp): Likewise.
3156 (vect_make_slp_decision): Likewise.
3157 (vect_detect_hybrid_slp): Likewise.
3158 (vect_slp_analyze_operations): Likewise.
3159 (vect_slp_bb): Likewise.
3160 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
3161 (vectorizable_bswap): Likewise.
3162 (vectorizable_call): Likewise.
3163 (vectorizable_simd_clone_call): Likewise.
3164 (vectorizable_conversion): Likewise.
3165 (vectorizable_assignment): Likewise.
3166 (vectorizable_shift): Likewise.
3167 (vectorizable_operation): Likewise.
3168 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
3170 2018-06-18 Martin Sebor <msebor@redhat.com>
3172 PR tree-optimization/81384
3173 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
3174 * builtins.c (expand_builtin_strnlen): New function.
3175 (expand_builtin): Call it.
3176 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
3177 * builtins.def (BUILT_IN_STRNLEN): New.
3178 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
3179 Warn for bounds in excess of maximum object size.
3180 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
3181 single-value ranges. Handle strnlen.
3182 (handle_builtin_strlen): Handle strnlen.
3183 (strlen_check_and_optimize_stmt): Same.
3184 * doc/extend.texi (Other Builtins): Document strnlen.
3186 2018-06-18 Maya Rashish <coypu@sdf.org>
3188 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
3189 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
3190 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
3192 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
3194 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
3196 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3198 * tree.c (escaped_string::escape): Replace cast to char * by
3199 const_cast<char *> (unescaped).
3201 2018-06-18 Nick Clifton <nickc@redhat.com>
3204 * tree.c (escaped_string): New class. Converts an unescaped
3205 string into its escaped equivalent.
3206 (warn_deprecated_use): Use the new class to convert the
3207 deprecation message, if present.
3208 (test_escaped_strings): New self test.
3209 (test_c_tests): Add test_escaped_strings.
3210 * doc/extend.texi (deprecated): Add a note that the
3211 deprecation message is affected by the -fmessage-length
3212 option, and that control characters will be escaped.
3213 (#pragma GCC error): Document this pragma.
3214 (#pragma GCC warning): Likewise.
3215 * doc/invoke.texi (-fmessage-length): Document this option's
3216 effect on the #warning and #error preprocessor directives and
3217 the deprecated attribute.
3219 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
3221 * tree.c (decl_value_expr_lookup): Revert latest change.
3222 (decl_value_expr_insert): Likewise.
3224 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
3226 * gimplify.c (nonlocal_vlas): Delete.
3227 (nonlocal_vla_vars): Likewise.
3228 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
3230 (gimplify_body): Do not create and destroy nonlocal_vlas.
3231 * tree-nested.c: Include diagnostic.h.
3232 (use_pointer_in_frame): Tweak.
3233 (lookup_field_for_decl): Add assertion and declare the transformation.
3234 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
3235 internal error when the reference is in a wrong context. Do not
3236 create a debug decl by default.
3237 (note_nonlocal_block_vlas): Delete.
3238 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
3239 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
3240 create a debug decl by default.
3241 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
3242 call is in a wrong context.
3243 (fixup_vla_decls): New function.
3244 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
3245 debug variables were created.
3246 * tree.c (decl_value_expr_lookup): Add checking assertion.
3247 (decl_value_expr_insert): Likewise.
3249 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
3252 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
3253 * tree-scalar-evolution.c (interpret_expr): Likewise.
3254 (expression_expensive_p): Likewise.
3255 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
3256 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
3257 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
3258 (ssa_defined_by_minus_one_stmt_p): New.
3260 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
3263 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
3264 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
3265 * dojump.c (do_jump): Likewise.
3266 * expr.c (expand_expr_real_2): Check operand type's sign.
3267 * fold-const.c (const_unop): Handle ABSU_EXPR.
3268 (fold_abs_const): Likewise.
3269 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
3270 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
3271 (strip_sign_op_1): Likesise.
3272 * match.pd: Add new pattern to generate ABSU_EXPR.
3273 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
3274 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
3275 * tree-eh.c (operation_could_trap_helper_p): Likewise.
3276 * tree-inline.c (estimate_operator_cost): Likewise.
3277 * tree-pretty-print.c (dump_generic_node): Likewise.
3278 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
3279 * tree.def (ABSU_EXPR): New.
3281 2018-06-16 Jakub Jelinek <jakub@redhat.com>
3284 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
3285 documented as preserved for backward compatibility only.
3286 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
3288 PR rtl-optimization/86108
3289 * bb-reorder.c (create_forwarder_block): Renamed to ...
3290 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
3291 jump from new landing pad to the second part.
3292 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
3295 2018-06-15 Jakub Jelinek <jakub@redhat.com>
3298 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
3299 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
3300 Only call store_expr for halves if the mode is the same.
3303 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
3304 Fix up comment formatting.
3306 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
3308 * typed-splay-tree.h (typed_splay_tree::remove): New function.
3309 (typed_splay_tree::closure,
3310 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
3311 (typed_splay_tree::typed_splay_tree,
3312 typed_splay_tree::operator =): Declared private.
3313 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
3314 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
3315 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
3316 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
3317 typed_splay_tree::splay_tree_splay,
3318 typed_splay_tree::splay_tree_foreach_helper,
3319 typed_splay_tree::splay_tree_insert,
3320 typed_splay_tree::splay_tree_remove,
3321 typed_splay_tree::splay_tree_lookup,
3322 typed_splay_tree::splay_tree_predecessor,
3323 typed_splay_tree::splay_tree_successor,
3324 typed_splay_tree::splay_tree_min,
3325 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
3326 (typed_splay_tree::root, typed_splay_tree::comp,
3327 typed_splay_tree::delete_key,
3328 typed_splay_tree::delete_value): New data members.
3329 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
3330 typed_splay_tree::remove.
3332 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
3334 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
3335 -mginv and -mno-ginv to the assembler.
3336 * config/mips/mips.opt (-mcrc): New option.
3338 * doc/invoke.text (-mcrc): Document.
3341 2018-06-15 Nick Clifton <nickc@redhat.com>
3344 * tree.c (escaped_string): New class. Converts an unescaped
3345 string into its escaped equivalent.
3346 (warn_deprecated_use): Use the new class to convert the
3347 deprecation message, if present.
3348 (test_escaped_strings): New self test.
3349 (test_c_tests): Add test_escaped_strings.
3350 * doc/extend.texi (deprecated): Add a note that the
3351 deprecation message is affected by the -fmessage-length
3352 option, and that control characters will be escaped.
3353 (#pragma GCC error): Document this pragma.
3354 (#pragma GCC warning): Likewise.
3355 * doc/invoke.texi (-fmessage-length): Document this option's
3356 effect on the #warning and #error preprocessor directives and
3357 the deprecated attribute.
3359 2018-06-15 Richard Biener <rguenther@suse.de>
3361 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
3362 here, also noting vector size used.
3363 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
3364 size used in MSG_OPTIMIZED_LOCATIONS dump.
3365 (pass_slp_vectorize::execute): Adjust.
3367 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
3370 * config/arc/arc.c (arc_return_address_register): Fix
3373 2018-06-15 Richard Biener <rguenther@suse.de>
3376 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
3377 leave useless conversion stripping to force_gimple_operand_gsi.
3378 (gimplify_build2): Likewise.
3379 (gimplify_build1): Likewise.
3381 2018-06-15 Richard Biener <rguenther@suse.de>
3384 * tree-cfg.c (move_stmt_op): unshare invariant addresses
3385 before adjusting their block.
3387 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
3389 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
3390 multilibs for *-*-rtems*.
3391 * config/riscv/t-rtems: New file.
3393 2018-06-14 Jakub Jelinek <jakub@redhat.com>
3396 * match.pd ((A +- CST1) +- CST2): Punt if last resort
3397 unsigned_type_for returns NULL.
3400 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
3401 subregs of multi-word pseudos unless the float mode has word size.
3403 2018-06-14 Richard Biener <rguenther@suse.de>
3406 * tree-vect-generic.c (build_word_mode_vector_type): Remove
3407 duplicate and harmful type_hash_canon.
3408 * tree.c (type_hash_canon): Assert we didn't find ourselves.
3410 2018-06-14 Richard Biener <rguenther@suse.de>
3413 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
3416 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
3418 * config/rtems.h (STDINT_LONG32): Define.
3420 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
3421 Prachi Godbole <prachi.godbole@imgtec.com>
3423 * config/mips/mips-cpus.def: Define P6600.
3424 * config/mips/mips-tables.opt: Regenerate.
3425 * config/mips/mips.c (mips_ucbranch_type): New enum.
3426 (mips_rtx_cost_data): Add support for P6600.
3427 (mips_issue_rate): Likewise.
3428 (mips_multipass_dfa_lookahead): Likewise.
3429 (mips_avoid_hazard): Likewise.
3430 (mips_reorg_process_insns): Likewise.
3431 (mips_classify_branch_p6600): New function.
3432 * config/mips/mips.h (TUNE_P6600): New define.
3433 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
3434 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
3435 * config/mips/mips.md: Include p6600.md.
3436 (processor): Add p6600.
3437 * config/mips/p6600.md: New file.
3438 * doc/invoke.texi: Add p6600 to supported architectures.
3440 2018-06-13 Martin Sebor <msebor@redhat.com>
3442 PR tree-optimization/86114
3443 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
3445 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
3447 2018-06-13 Richard Biener <rguenther@suse.de>
3449 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
3450 Properly set vector type of the intermediate stmt.
3451 * tree-vect-stmts.c (vectorizable_operation): The destination
3452 var always has vectype_out type.
3454 2018-06-13 Jeff Law <law@redhat.com>
3456 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
3457 integer 0 for argument to print_rtl_with_bb.
3458 (rl78_reorg): Likewise.
3460 2018-06-13 David Malcolm <dmalcolm@redhat.com>
3462 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
3463 from rtx to rtx_insn *.
3464 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
3466 (add_sched_insns_for_speculation): Likewise for local "target",
3467 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
3468 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
3469 from rtx_insn ** to rtx_code_label **.
3470 (reorg_emit_nops): Likewise.
3471 (c6x_reorg): Likewise for local "call_labels".
3472 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
3474 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
3475 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
3476 the loops over LABEL_REFS.
3477 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
3479 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
3480 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
3481 (split_branches): Strengthen local "olabel" from rtx to
3482 rtx_insn *, adding a safe_as_a cast.
3483 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
3485 (add_insn_after): Likewise for first two params.
3486 (add_insn_before): Likewise.
3487 (remove_insn): Likewise for param.
3488 (emit_pattern_before_noloc): Likewise for second and third params.
3489 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
3490 (emit_call_insn_before_noloc): Likewise.
3491 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
3493 (emit_barrier_before): Likewise.
3494 (emit_label_before): Strengthen "label" param from "rtx" to
3495 "rtx_code_label *". Strengthen "before" param from "rtx" to
3497 (emit_insn_after_1): Strengthen "after" param from "rtx" to
3499 (emit_pattern_after_noloc): Likewise.
3500 (emit_insn_after_noloc): Likewise.
3501 (emit_jump_insn_after_noloc): Likewise.
3502 (emit_call_insn_after_noloc): Likewise.
3503 (emit_debug_insn_after_noloc): Likewise.
3504 (emit_barrier_after): Likewise.
3505 (emit_label_after): Likewise for both params.
3506 (emit_pattern_after_setloc): Likewise for "after" param. Convert
3507 "loc" param from "int" to "location_t".
3508 (emit_insn_after_setloc): Likewise.
3509 (emit_jump_insn_after_setloc): Likewise.
3510 (emit_call_insn_after_setloc): Likewise.
3511 (emit_debug_insn_after_setloc): Likewise.
3512 (emit_pattern_before_setloc): Likewise for "before" param. Convert
3513 "loc" param from "int" to "location_t".
3514 (emit_pattern_before): Convert NULL_RTX to NULL.
3515 (emit_insn_before_setloc): Convert "loc" param from "int" to
3517 (emit_jump_insn_before_setloc): Likewise.
3518 (emit_call_insn_before_setloc): Likewise.
3519 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
3520 rtx_insn *. Convert "loc" param from "int" to "location_t".
3521 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
3522 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
3523 Convert 3rd param from "int" to "location_t".
3524 (emit_barrier_before, emit_barrier_after, next_real_insn):
3525 Strengthen param from rtx to rtx_insn *.
3526 (emit_label_before): Strengthen 1st param from "rtx" to
3527 "rtx_code_label *". Strengthen 2nd param from "rtx" to
3529 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
3530 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
3531 Strengthen 2nd param from "rtx" to "rtx_insn *".
3532 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
3533 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
3534 Likewise. Convert 3rd param from "int" to "location_t".
3535 (emit_label_after): Strengthen 1st param from "rtx" to
3537 (next_real_insn, remove_insn): Strengthen param from "rtx" to
3539 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
3540 from "rtx" to "rtx_insn *".
3542 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
3544 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
3545 bodies streamed in with -Q.
3546 * dumpfile.c (dump_files): Add lto-stream-out dump file.
3547 * dumpfile.h (tree_dump_index): Add lto_stream_out.
3548 * gimple-streamer-out.c: Include gimple-pretty-print.h
3549 (output_bb): Dump stmts streamed.
3550 * lto-section-out.c: Include print-tree.h
3551 (lto_begin_section): Dump sections created.
3552 (lto_output_decl_index): Dump decl encoded.
3553 * lto-streamer-out.c: Include print-tree.h
3554 (create_output_block): Dump output block created.
3555 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
3556 (output_function): Dump function output.
3557 (output_constructor): Dump constructor streamed.
3558 (write_global_stream): Output indexes encoded.
3559 (produce_asm_for_decls): Dump streams encoded.
3560 * lto-streamer.c (streamer_dump_file): New global var.
3561 * lto-streamer.h (streamer_dump_file): Declare.
3562 * passes.c (ipa_write_summaries): Initialize streamer dump.
3563 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
3566 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
3569 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
3570 offsets for register save directives. Emit a second batch of save
3571 directives, if need be, when the function accesses prior frames.
3573 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
3575 * config/arc/fpu.md (fmasf4): Force operand to register.
3576 (fnmasf4): Likewise.
3578 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
3580 * config/arc/arc-protos.h (arc_pad_return): Remove.
3581 * config/arc/arc.c (machine_function): Remove force_short_suffix
3583 (arc_print_operand): Adjust printing of '&'.
3584 (arc_verify_short): Remove conditional printing of short suffix.
3585 (arc_final_prescan_insn): Remove reference to size_reason.
3586 (pad_return): New function.
3587 (arc_reorg): Call pad_return.
3588 (arc_pad_return): Remove.
3589 (arc_init_machine_status): Remove reference to force_short_suffix.
3590 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
3591 (attr length): When attribute iscompact is true force to 2
3592 regardless; in the case of maybe check if we want to force the
3593 instruction to have 4 bytes length.
3594 (nopv): Change it to generate 4 byte long nop as well.
3595 (blockage): New pattern.
3596 (simple_return): Remove call to arc_pad_return.
3597 (p_return_i): Likewise.
3599 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
3601 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
3603 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
3605 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
3608 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
3610 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
3611 for ARC700 and ARCv2.
3613 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
3616 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
3617 operands[2] instead of operands[1].
3620 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
3622 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
3623 case, check whether the outer register overlaps an unallocatable
3624 register, not just whether it fits the required class.
3626 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
3628 * poly-int.h (can_div_trunc_p): Add new overload in which all values
3630 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
3631 (memrefs_conflict_p): Likewise.
3632 (init_alias_analysis): Likewise.
3633 * cfgexpand.c (expand_debug_expr): Likewise.
3634 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
3635 * cse.c (fold_rtx): Likewise.
3636 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
3637 * expr.c (emit_block_move_hints): Likewise.
3638 (clear_storage_hints, push_block, emit_push_insn): Likewise.
3639 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
3640 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
3641 (emit_group_store): Likewise.
3642 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
3643 to read the PRE/POST_MODIFY increment.
3644 * calls.c (store_one_arg): Use strip_offset.
3645 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
3647 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
3649 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
3650 (simplify_binary_operation_1): Extend CONST_INT handling to
3652 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
3653 than a HOST_WIDE_INT.
3654 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
3656 (adjust_mems, add_stores): Update accodingly.
3657 (vt_canonicalize_addr): Track polynomial offsets.
3658 (emit_note_insn_var_location): Likewise.
3659 (vt_add_function_parameter): Likewise.
3660 (vt_initialize): Likewise.
3662 2018-06-12 Jeff Law <law@redhat.com>
3664 * config.gcc (alpha*-*-freebsd*): Remove.
3665 * config/alpha/freebsd.h: Remove.
3667 2018-06-12 David Malcolm <dmalcolm@redhat.com>
3670 * spellcheck-tree.c (levenshtein_distance): Rename to...
3671 (get_edit_distance): ...this, and update for underlying renaming.
3672 * spellcheck-tree.h (levenshtein_distance): Rename to...
3673 (get_edit_distance): ...this.
3674 * spellcheck.c (levenshtein_distance): Rename to...
3675 (get_edit_distance): ...this. Convert from Levenshtein distance
3676 to Damerau-Levenshtein distance by supporting transpositions of
3677 adjacent characters. Rename "v1" to "v_next" and "v0" to
3679 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
3680 (selftest::test_edit_distance_unit_test_oneway): ...this, and
3681 update for underlying renaming.
3682 (selftest::levenshtein_distance_unit_test): Rename to...
3683 (selftest::test_get_edit_distance_unit): ...this, and update for
3684 underlying renaming.
3685 (selftest::test_find_closest_string): Add example from PR 69968
3686 where transposition helps
3687 (selftest::test_metric_conditions): Update for renaming.
3688 (selftest::test_metric_conditions): Likewise.
3689 (selftest::spellcheck_c_tests): Likewise.
3690 * spellcheck.h (levenshtein_distance): Rename both overloads to...
3691 (get_edit_distance): ...this.
3692 (best_match::consider): Update for renaming.
3694 2018-06-12 Martin Sebor <msebor@redhat.com>
3696 PR tree-optimization/85259
3697 * builtins.c (compute_objsize): Handle constant offsets.
3698 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
3699 true iff a warning has been issued.
3700 * gimple.h (gimple_nonartificial_location): New function.
3701 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
3702 gimple_nonartificial_location and handle -Wno-system-headers.
3703 (handle_builtin_stxncpy): Same.
3705 2018-06-12 Martin Sebor <msebor@redhat.com>
3708 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
3710 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
3712 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
3713 BUILTIN_VEC_XST entries for pointer to double and long long.
3715 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
3718 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
3720 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
3723 2018-06-12 Martin Liska <mliska@suse.cz>
3725 * doc/options.texi: Document IntegerRange.
3727 2018-06-12 Martin Liska <mliska@suse.cz>
3729 * config/i386/i386.opt: Make MPX-related options as Deprecated.
3730 * opt-functions.awk: Handle Deprecated flag.
3731 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
3733 (read_cmdline_option): Report warning for a deprecated option.
3734 * opts.h (struct cl_option): Add new field cl_deprecated.
3735 (CL_ERR_DEPRECATED): New.
3737 2018-06-12 Martin Liska <mliska@suse.cz>
3739 * doc/options.texi: Document Deprecated option flag.
3741 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
3743 * config/arc/arc-arch.h (arc_extras): New enum.
3744 (arc_cpu_t):Add field extra.
3745 (arc_cpu_types): Consider the extras.
3746 * config/arc/arc-cpus.def: Add extras info.
3747 * config/arc/arc-opts.h (processor_type): Consider extra field.
3748 * config/arc/arc.c (arc_override_options): Handle extra field.
3750 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
3752 * config/arc/arc-arch.h: Update ARC_OPTX macro.
3753 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
3755 * config/arc/arc.c (arc_init): Update pic warning.
3756 (irq_range): Update irq range parsing warnings.
3757 (arc_override_options): Update various warning messages.
3758 (arc_handle_aux_attribute): Likewise.
3760 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
3762 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
3764 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
3766 * doc/sourcebuild.texi: Document usage of line number 0 in verify
3767 compiler messages directives.
3769 2018-06-12 Matthew Fortune <mfortune@gmail.com>
3771 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
3772 * config/mips/mips-tables.opt: Regenerate.
3773 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
3775 * doc/invoke.texi: Document -march=i6500.
3777 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
3779 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
3780 (i6400_gpmul): Add cpu_unit.
3781 (i6400_gpdiv): Likewise.
3782 (i6400_msa_add_d): Update reservations.
3783 (i6400_msa_int_add) Likewise.
3784 (i6400_msa_short_logic3) Likewise.
3785 (i6400_msa_short_logic2) Likewise.
3786 (i6400_msa_short_logic) Likewise.
3787 (i6400_msa_move) Likewise.
3788 (i6400_msa_cmp) Likewise.
3789 (i6400_msa_short_float2) Likewise.
3790 (i6400_msa_div_d) Likewise.
3791 (i6400_msa_long_logic1) Likewise.
3792 (i6400_msa_long_logic2) Likewise.
3793 (i6400_msa_mult) Likewise.
3794 (i6400_msa_long_float2) Likewise.
3795 (i6400_msa_long_float4) Likewise.
3796 (i6400_msa_long_float5) Likewise.
3797 (i6400_msa_long_float8) Likewise.
3798 (i6400_fpu_fadd): Include frint type.
3799 (i6400_fpu_store): New define_insn_reservation.
3800 (i6400_fpu_load): Likewise.
3801 (i6400_fpu_move): Likewise.
3802 (i6400_fpu_fcmp): Likewise.
3803 (i6400_fpu_fmadd): Likewise.
3804 (i6400_int_mult): Include imul3nc type and update reservation.
3805 (i6400_int_div): Include idiv3 type and update reservation.
3806 (i6400_int_load): Update to check type not move_type.
3807 (i6400_int_store): Likewise.
3808 (i6400_int_prefetch): Set zero latency.
3810 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
3812 * gcc.c: Document new %@{...} sequence.
3813 (LINK_COMMAND_SPEC): Use it for the -L switches.
3814 (cpp_unique_options): Use it for the -I switches.
3815 (at_file_argbuf): New global variable.
3816 (in_at_file): Likewise.
3817 (alloc_args): Create at_file_argbuf.
3818 (clear_args): Truncate at_file_argbuf.
3819 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
3820 (open_at_file): New function.
3821 (close_at_file): Likewise.
3822 (create_at_file): Delete.
3823 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
3826 (validate_switches_from_spec): Deal with %@{...} sequence.
3827 (validate_switches): Likewise.
3828 (driver::finalize): Call clear_args.
3830 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
3832 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
3834 2018-06-11 Martin Sebor <msebor@redhat.com>
3836 * doc/invoke.texi (-Wall): List -Wc++17-compat.
3837 (Wno-class-memaccess): Add @opindex.
3838 (Wno-templates, Wno-multiple-inheritance): Same.
3839 (Wno-virtual-inheritance, Wno-namespaces): Same.
3840 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
3841 (Wno-format-overflow, Wno-format-truncation): Same.
3842 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
3843 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
3844 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
3845 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
3846 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
3847 (Wno-misspelled-isr): Same.
3849 2018-06-11 Martin Sebor <msebor@redhat.com>
3851 * PR tree-optimization/86083
3852 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
3854 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
3856 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
3858 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
3861 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
3862 on the correct operand.
3863 (*movdi_internal64): Ditto.
3865 2018-06-11 Martin Liska <mliska@suse.cz>
3867 PR tree-optimization/86089
3868 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
3870 2018-06-11 Julia Koval <julia.koval@intel.com>
3872 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
3873 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
3874 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
3876 2018-06-11 Olivier Hainque <hainque@adacore.com>
3878 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
3879 for Ada with strict dwarf2.
3881 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
3884 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
3887 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
3889 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
3891 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
3893 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
3896 2018-06-08 Martin Liska <mliska@suse.cz>
3898 * tree-cfg.h (debug_function): Fix argument type to match
3901 2018-06-08 Martin Liska <mliska@suse.cz>
3903 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
3904 Remove usage of MPX-related (and removed) fields.
3905 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
3907 2018-06-08 David Malcolm <dmalcolm@redhat.com>
3909 * cfg.c (debug): Use TDF_NONE rather than 0.
3910 * cfghooks.c (debug): Likewise.
3911 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
3912 (struct dump_option_value_info): Convert to...
3913 (struct kv_pair): ...this template type.
3914 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
3916 (optinfo_verbosity_options): Likewise.
3917 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
3919 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
3920 than int for "optgroup_flags" param.
3921 (dump_generic_expr_loc): Use dump_flags_t rather than int for
3923 (dump_dec): Likewise.
3924 (dump_finish): Use TDF_NONE rather than 0.
3925 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
3926 rather than int for "optgroup_flags" param. Use TDF_NONE rather
3927 than 0. Update for change to option_ptr.
3928 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
3929 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
3930 0. Update for changes to optinfo_verbosity_options and
3932 (opt_info_switch_p): Convert optgroup_flags from int to
3934 (dump_basic_block): Use dump_flags_t rather than int
3935 for "dump_kind" param.
3936 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
3937 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
3938 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
3939 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
3940 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
3941 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
3942 TDF_NONE): Convert from macros to...
3943 (enum dump_flag): ...this new enum.
3944 (dump_flags_t): Update to use enum.
3945 (operator|, operator&, operator~, operator|=, operator&=):
3946 Implement for dump_flags_t.
3947 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
3948 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
3949 Convert from macros to...
3950 (enum optgroup_flag): ...this new enum.
3951 (optgroup_flags_t): New typedef.
3952 (operator|, operator|=): Implement for optgroup_flags_t.
3953 (struct dump_file_info): Convert field "alt_flags" to
3954 dump_flags_t. Convert field "optgroup_flags" to
3956 (dump_basic_block): Use dump_flags_t rather than int for param.
3957 (dump_generic_expr_loc): Likewise.
3958 (dump_dec): Likewise.
3959 (dump_register): Convert param "optgroup_flags" to
3961 (opt_info_enable_passes): Likewise.
3962 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
3964 * gimple-pretty-print.c (debug): Likewise.
3965 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
3966 (merged_store_group::apply_stores): Likewise.
3967 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
3968 * gimple.c (verify_gimple_pp): Likewise.
3969 * graphite-poly.c (print_pbb_body): Likewise.
3970 * passes.c (pass_manager::register_one_dump_file): Convert
3971 local "optgroup_flags" to optgroup_flags_t.
3972 * print-tree.c (print_node): Use TDF_NONE rather than 0.
3974 (debug_body): Likewise.
3975 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
3976 to optgroup_flags_t.
3977 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
3979 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
3980 (convert_mult_to_fma): Likewise.
3981 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
3982 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
3983 * tree-vect-data-refs.c (dump_lower_bound): Convert param
3984 "dump_kind" to dump_flags_t.
3986 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
3988 * config/rs6000/rs6000.c (min, max): Delete.
3990 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
3992 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
3993 -mabi=spe and -mabi=no-spe.
3995 2018-06-08 Martin Liska <mliska@suse.cz>
3997 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
3998 where we expect an existing summary.
4000 2018-06-08 Martin Liska <mliska@suse.cz>
4002 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
4003 * ipa-inline.h (estimate_edge_growth): Likewise.
4005 2018-06-08 Martin Liska <mliska@suse.cz>
4007 * cgraph.c (function_version_hasher::hash): Use
4008 cgraph_node::get_uid ().
4009 (function_version_hasher::equal):
4010 * cgraph.h (cgraph_node::get_uid): New method.
4011 * ipa-inline.c (update_caller_keys): Use
4012 cgraph_node::get_uid ().
4013 (update_callee_keys): Likewise.
4014 * ipa-utils.c (searchc): Likewise.
4015 (ipa_reduced_postorder): Likewise.
4016 * lto-cgraph.c (input_node): Likewise.
4017 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
4018 * symbol-summary.h (symtab_insertion): Likewise.
4019 (symtab_removal): Likewise.
4020 (symtab_duplication): Likewise.
4021 * tree-pretty-print.c (dump_function_header): Likewise.
4022 * tree-sra.c (convert_callers_for_node): Likewise.
4024 2018-06-08 Martin Liska <mliska@suse.cz>
4026 * cgraph.c (symbol_table::create_edge): Always assign a new
4028 (symbol_table::free_edge): Do not recycle numbers.
4029 * cgraph.h (cgraph_edge::get): New method.
4030 * symbol-summary.h (symtab_removal): Use it.
4031 (symtab_duplication): Likewise.
4032 (call_summary::hashable_uid): Remove.
4034 2018-06-08 Martin Liska <mliska@suse.cz>
4036 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
4037 (initialize_growth_caches): Remove.
4038 (free_growth_caches): Likewise.
4039 (do_estimate_edge_time): Use edge_growth_cache.
4040 (do_estimate_edge_size): Likewise.
4041 (do_estimate_edge_hints): Likewise.
4042 * ipa-inline.c (reset_edge_caches): Likewise.
4043 (recursive_inlining): Likewise.
4044 (inline_small_functions): Likewise.
4045 * ipa-inline.h (initialize_growth_caches): Remove.
4046 (estimate_edge_size): Likewise.
4047 (estimate_edge_time): Likewise.
4048 (estimate_edge_hints): Likewise.
4049 (reset_edge_growth_cache): Likewise.
4050 * symbol-summary.h (call_summary::remove): New method.
4052 2018-06-08 Martin Liska <mliska@suse.cz>
4054 * ipa-cp.c (class edge_clone_summary): New summary.
4055 (grow_edge_clone_vectors): Remove.
4056 (ipcp_edge_duplication_hook): Remove.
4057 (class edge_clone_summary_t): New call_summary class.
4058 (ipcp_edge_removal_hook): Remove.
4059 (edge_clone_summary_t::duplicate): New function.
4060 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
4061 (create_specialized_node): Likewise.
4062 (ipcp_driver): Initialize edge_clone_summaries and do not
4065 2018-06-08 Martin Liska <mliska@suse.cz>
4067 * symbol-summary.h (get): New function.
4068 (call_summary::m_initialize_when_cloning): New class member.
4070 2018-06-08 Martin Liska <mliska@suse.cz>
4072 * cgraph.c (cgraph_node::remove): Do not recycle uid.
4073 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
4074 (symbol_table::allocate_cgraph_symbol): Do not set uid.
4075 * passes.c (uid_hash_t): Record removed_nodes by their uids.
4076 (remove_cgraph_node_from_order): Use the removed_nodes set.
4077 (do_per_function_toporder): Likwise.
4078 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
4079 instead of summary_uid.
4080 (symtab_removal): Likewise.
4081 (symtab_duplication): Likewise.
4083 2018-06-08 Martin Liska <mliska@suse.cz>
4085 * ipa-cp.c (ipcp_store_bits_results): Use
4086 ipcp_transformation_sum.
4087 (ipcp_store_vr_results): Likewise.
4088 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
4090 (ipcp_transformation_initialize): ... this.
4091 (ipa_set_node_agg_value_chain):
4092 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
4093 (write_ipcp_transformation_info): Likewise.
4094 (read_ipcp_transformation_info): Likewise.
4095 (ipcp_update_bits): Likewise.
4096 (ipcp_update_vr): Likewise.
4097 (ipcp_transform_function): Likewise.
4098 * ipa-prop.h: Rename ipcp_transformation_summary to
4099 ipcp_transformation.
4100 (class ipcp_transformation_t): New function summary.
4101 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
4102 (ipa_get_agg_replacements_for_node): Likewise.
4104 2018-06-08 Martin Liska <mliska@suse.cz>
4106 * ipa-pure-const.c (struct funct_state_d): Do it class instead
4108 (class funct_state_summary_t): New function_summary class.
4109 (has_function_state): Remove.
4110 (get_function_state): Likewise.
4111 (set_function_state): Likewise.
4112 (add_new_function): Likewise.
4113 (funct_state_summary_t::insert): New function.
4114 (duplicate_node_data): Remove.
4115 (remove_node_data): Remove.
4116 (funct_state_summary_t::duplicate): New function.
4117 (register_hooks): Create new funct_state_summaries.
4118 (pure_const_generate_summary): Use it.
4119 (pure_const_write_summary): Likewise.
4120 (pure_const_read_summary): Likewise.
4121 (propagate_pure_const): Likewise.
4122 (propagate_nothrow): Likewise.
4123 (dump_malloc_lattice): Likewise.
4124 (propagate_malloc): Likewise.
4125 (execute): Do not register hooks, just remove summary
4127 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
4130 2018-06-08 Martin Liska <mliska@suse.cz>
4132 * ipa-reference.c (remove_node_data): Remove.
4133 (duplicate_node_data): Likewise.
4134 (class ipa_ref_var_info_summary_t): New class.
4135 (class ipa_ref_opt_summary_t): Likewise.
4136 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
4137 (get_reference_optimization_summary): Use
4138 ipa_ref_opt_sum_summaries.
4139 (set_reference_vars_info): Remove.
4140 (set_reference_optimization_summary): Likewise.
4141 (ipa_init): Create summaries.
4142 (init_function_info): Use function summary.
4143 (ipa_ref_opt_summary_t::duplicate): New function.
4144 (ipa_ref_opt_summary_t::remove): New function.
4145 (get_read_write_all_from_node): Fix GNU coding style.
4146 (propagate): Use function summary.
4147 (write_node_summary_p): Fix GNU coding style.
4148 (stream_out_bitmap): Likewise.
4149 (ipa_reference_read_optimization_summary): Use function summary.
4150 (ipa_reference_c_finalize): Do not release hooks.
4152 2018-06-08 Martin Liska <mliska@suse.cz>
4154 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
4155 (analyze_function_body): Extract multiple calls of get_create.
4156 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
4157 * ipa-inline.c (recursive_inlining): Use ::get method.
4158 * ipa-inline.h (estimate_edge_growth): Likewise.
4160 2018-06-08 Martin Liska <mliska@suse.cz>
4162 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
4164 (hsa_function_summary::hsa_function_summary): Use the new enum
4166 (hsa_gpu_implementation_p): Use hsa_summaries::get.
4167 * hsa-gen.c (hsa_get_host_function): Likewise.
4168 (get_brig_function_name): Likewise.
4169 * ipa-hsa.c (process_hsa_functions): Likewise.
4170 (ipa_hsa_write_summary): Likewise.
4171 * symbol-summary.h (symtab_duplication): Use ::get function/
4172 (get): New function.
4174 2018-06-08 Martin Liska <mliska@suse.cz>
4176 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
4178 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
4179 (hsa_register_kernel): Likewise.
4180 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
4181 * hsa-gen.c (hsa_get_host_function): Likewise.
4182 (get_brig_function_name): Likewise.
4183 (generate_hsa): Likewise.
4184 (pass_gen_hsail::execute): Likewise.
4185 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
4186 (devirtualization_time_bonus): Likewise.
4187 (ipcp_propagate_stage): Likewise.
4188 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
4189 (edge_set_predicate): Likewise.
4190 (evaluate_conditions_for_known_args): Likewise.
4191 (evaluate_properties_for_edge): Likewise.
4192 (ipa_fn_summary::reset): Likewise.
4193 (ipa_fn_summary_t::duplicate): Likewise.
4194 (dump_ipa_call_summary): Likewise.
4195 (ipa_dump_fn_summary): Likewise.
4196 (analyze_function_body): Likewise.
4197 (compute_fn_summary): Likewise.
4198 (estimate_edge_devirt_benefit): Likewise.
4199 (estimate_edge_size_and_time): Likewise.
4200 (estimate_calls_size_and_time): Likewise.
4201 (estimate_node_size_and_time): Likewise.
4202 (inline_update_callee_summaries): Likewise.
4203 (remap_edge_change_prob): Likewise.
4204 (remap_edge_summaries): Likewise.
4205 (ipa_merge_fn_summary_after_inlining): Likewise.
4206 (ipa_update_overall_fn_summary): Likewise.
4207 (read_ipa_call_summary): Likewise.
4208 (inline_read_section): Likewise.
4209 (write_ipa_call_summary): Likewise.
4210 (ipa_fn_summary_write): Likewise.
4211 (ipa_free_fn_summary): Likewise.
4212 * ipa-hsa.c (process_hsa_functions): Likewise.
4213 (ipa_hsa_write_summary): Likewise.
4214 (ipa_hsa_read_section): Likewise.
4215 * ipa-icf.c (sem_function::merge): Likewise.
4216 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
4217 (do_estimate_edge_time): Likewise.
4218 (estimate_size_after_inlining): Likewise.
4219 (estimate_growth): Likewise.
4220 (growth_likely_positive): Likewise.
4221 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
4222 (inline_call): Likewise.
4223 * ipa-inline.c (caller_growth_limits): Likewise.
4224 (can_inline_edge_p): Likewise.
4225 (can_inline_edge_by_limits_p): Likewise.
4226 (compute_uninlined_call_time): Likewise.
4227 (compute_inlined_call_time): Likewise.
4228 (want_inline_small_function_p): Likewise.
4229 (edge_badness): Likewise.
4230 (update_caller_keys): Likewise.
4231 (update_callee_keys): Likewise.
4232 (recursive_inlining): Likewise.
4233 (inline_small_functions): Likewise.
4234 (inline_to_all_callers_1): Likewise.
4235 (dump_overall_stats): Likewise.
4236 (early_inline_small_functions): Likewise.
4237 (early_inliner): Likewise.
4238 * ipa-inline.h (estimate_edge_growth): Likewise.
4239 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
4240 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
4241 * ipa-prop.h (IPA_NODE_REF): Likewise.
4242 (IPA_EDGE_REF): Likewise.
4243 * ipa-pure-const.c (malloc_candidate_p): Likewise.
4244 (propagate_malloc): Likewise.
4245 * ipa-split.c (execute_split_functions): Likewise.
4246 * symbol-summary.h: Rename get to get_create.
4248 (get_create): Likewise.
4249 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
4251 2018-06-08 Martin Liska <mliska@suse.cz>
4253 * symbol-summary.h (release): Move definition out of class
4255 (symtab_removal): Likewise.
4256 (symtab_duplication): Likewise.
4258 2018-06-08 Martin Liska <mliska@suse.cz>
4260 * symbol-summary.h (function_summary): Move constructor
4261 implementation out of class declaration.
4262 (release): Likewise.
4263 (symtab_insertion): Likewise.
4264 (symtab_removal): Likewise.
4265 (symtab_duplication): Likewise.
4268 2018-06-08 Martin Liska <mliska@suse.cz>
4270 * Makefile.in: Remove support for MPX (macros, related functions,
4271 fields in cgraph_node, ...).
4272 * builtin-types.def (BT_BND): Likewise.
4273 (BT_FN_BND_CONST_PTR): Likewise.
4274 (BT_FN_CONST_PTR_BND): Likewise.
4275 (BT_FN_VOID_PTR_BND): Likewise.
4276 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
4277 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
4278 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
4279 (expand_builtin_mempcpy_with_bounds): Likewise.
4280 (expand_builtin_memset_with_bounds): Likewise.
4281 (expand_builtin_memset_args): Likewise.
4282 (std_expand_builtin_va_start): Likewise.
4283 (expand_builtin): Likewise.
4284 (expand_builtin_with_bounds): Likewise.
4285 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
4286 (DEF_LIB_BUILTIN_CHKP): Likewise.
4287 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
4288 (DEF_CHKP_BUILTIN): Likewise.
4289 (BUILT_IN_MEMCPY): Likewise.
4290 (BUILT_IN_MEMMOVE): Likewise.
4291 (BUILT_IN_MEMPCPY): Likewise.
4292 (BUILT_IN_MEMSET): Likewise.
4293 (BUILT_IN_STPCPY): Likewise.
4294 (BUILT_IN_STRCAT): Likewise.
4295 (BUILT_IN_STRCHR): Likewise.
4296 (BUILT_IN_STRCPY): Likewise.
4297 (BUILT_IN_STRLEN): Likewise.
4298 (BUILT_IN_MEMCPY_CHK): Likewise.
4299 (BUILT_IN_MEMMOVE_CHK): Likewise.
4300 (BUILT_IN_MEMPCPY_CHK): Likewise.
4301 (BUILT_IN_MEMSET_CHK): Likewise.
4302 (BUILT_IN_STPCPY_CHK): Likewise.
4303 (BUILT_IN_STRCAT_CHK): Likewise.
4304 (BUILT_IN_STRCPY_CHK): Likewise.
4305 * calls.c (store_bounds): Likewise.
4306 (emit_call_1): Likewise.
4307 (special_function_p): Likewise.
4308 (maybe_warn_nonstring_arg): Likewise.
4309 (initialize_argument_information): Likewise.
4310 (finalize_must_preallocate): Likewise.
4311 (compute_argument_addresses): Likewise.
4312 (expand_call): Likewise.
4313 * cfgexpand.c (expand_call_stmt): Likewise.
4314 (expand_return): Likewise.
4315 (expand_gimple_stmt_1): Likewise.
4316 (pass_expand::execute): Likewise.
4317 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
4318 (cgraph_node::remove): Likewise.
4319 (cgraph_node::dump): Likewise.
4320 (cgraph_node::verify_node): Likewise.
4321 * cgraph.h (chkp_function_instrumented_p): Likewise.
4322 (symtab_node::get_alias_target): Likewise.
4323 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
4324 (cgraph_local_p): Likewise.
4325 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
4326 (cgraph_edge::rebuild_references): Likewise.
4327 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
4328 (walk_polymorphic_call_targets): Likewise.
4329 (cgraph_node::expand_thunk): Likewise.
4330 (symbol_table::output_weakrefs): Likewise.
4331 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
4332 (ix86_handle_option): Likewise.
4333 * config/i386/constraints.md: Likewise.
4334 * config/i386/i386-builtin-types.def (BND): Likewise.
4338 * config/i386/i386-builtin.def (BDESC_END): Likewise.
4339 (BDESC_FIRST): Likewise.
4341 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
4342 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
4343 * config/i386/i386.c (enum reg_class): Likewise.
4344 (ix86_target_string): Likewise.
4345 (ix86_option_override_internal): Likewise.
4346 (ix86_conditional_register_usage): Likewise.
4347 (ix86_valid_target_attribute_inner_p): Likewise.
4348 (ix86_set_indirect_branch_type): Likewise.
4349 (ix86_set_current_function): Likewise.
4350 (ix86_function_arg_regno_p): Likewise.
4351 (init_cumulative_args): Likewise.
4352 (ix86_function_arg_advance): Likewise.
4353 (ix86_function_arg): Likewise.
4354 (ix86_pass_by_reference): Likewise.
4355 (ix86_function_value_regno_p): Likewise.
4356 (ix86_function_value_1): Likewise.
4357 (ix86_function_value_bounds): Likewise.
4358 (ix86_return_in_memory): Likewise.
4359 (ix86_setup_incoming_vararg_bounds): Likewise.
4360 (ix86_va_start): Likewise.
4361 (indirect_thunk_need_prefix): Likewise.
4362 (print_reg): Likewise.
4363 (ix86_print_operand): Likewise.
4364 (ix86_expand_call): Likewise.
4365 (ix86_output_function_return): Likewise.
4366 (reg_encoded_number): Likewise.
4367 (BDESC_VERIFYS): Likewise.
4368 (ix86_init_mpx_builtins): Likewise.
4369 (ix86_init_builtins): Likewise.
4370 (ix86_emit_cmove): Likewise.
4371 (ix86_emit_move_max): Likewise.
4372 (ix86_expand_builtin): Likewise.
4373 (ix86_builtin_mpx_function): Likewise.
4374 (ix86_get_arg_address_for_bt): Likewise.
4375 (ix86_load_bounds): Likewise.
4376 (ix86_store_bounds): Likewise.
4377 (ix86_load_returned_bounds): Likewise.
4378 (ix86_store_returned_bounds): Likewise.
4379 (ix86_class_likely_spilled_p): Likewise.
4380 (ix86_hard_regno_mode_ok): Likewise.
4381 (x86_order_regs_for_local_alloc): Likewise.
4382 (ix86_mitigate_rop): Likewise.
4383 (ix86_bnd_prefixed_insn_p): Likewise.
4384 (ix86_mpx_bound_mode): Likewise.
4385 (ix86_make_bounds_constant): Likewise.
4386 (ix86_initialize_bounds): Likewise.
4387 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
4388 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
4389 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
4390 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
4391 (TARGET_CHKP_BOUND_MODE): Likewise.
4392 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
4393 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
4394 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
4395 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
4396 * config/i386/i386.h (TARGET_MPX): Likewise.
4397 (TARGET_MPX_P): Likewise.
4398 (VALID_BND_REG_MODE): Likewise.
4399 (FIRST_BND_REG): Likewise.
4400 (LAST_BND_REG): Likewise.
4401 (enum reg_class): Likewise.
4402 (BND_REG_P): Likewise.
4403 (BND_REGNO_P): Likewise.
4404 (BNDmode): Likewise.
4405 (ADJUST_INSN_LENGTH): Likewise.
4406 * config/i386/i386.md: Likewise.
4407 * config/i386/i386.opt: Likewise.
4408 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
4409 (defined): Likewise.
4410 (LINK_MPX): Likewise.
4411 (MPX_SPEC): Likewise.
4412 (LIBMPX_SPEC): Likewise.
4413 (LIBMPXWRAPPERS_SPEC): Likewise.
4414 (CHKP_SPEC): Likewise.
4415 * config/i386/predicates.md: Likewise.
4416 * dbxout.c (dbxout_type): Likewise.
4417 * doc/extend.texi: Likewise.
4418 * doc/invoke.texi: Likewise.
4419 * doc/md.texi: Likewise.
4420 * doc/tm.texi: Likewise.
4421 * doc/tm.texi.in: Likewise.
4422 * dwarf2out.c (is_base_type): Likewise.
4423 (gen_formal_types_die): Likewise.
4424 (gen_subprogram_die): Likewise.
4425 (gen_type_die_with_usage): Likewise.
4426 (gen_decl_die): Likewise.
4427 (dwarf2out_late_global_decl): Likewise.
4428 * expr.c (expand_assignment): Likewise.
4429 (emit_storent_insn): Likewise.
4430 (store_expr_with_bounds): Likewise.
4431 (store_expr): Likewise.
4432 (expand_expr_real_1): Likewise.
4433 * expr.h (store_expr_with_bounds): Likewise.
4434 * function.c (use_register_for_decl): Likewise.
4435 (struct bounds_parm_data): Likewise.
4436 (assign_parms_augmented_arg_list): Likewise.
4437 (assign_parm_find_entry_rtl): Likewise.
4438 (assign_parm_is_stack_parm): Likewise.
4439 (assign_parm_load_bounds): Likewise.
4440 (assign_bounds): Likewise.
4441 (assign_parms): Likewise.
4442 (expand_function_start): Likewise.
4443 * gcc.c (CHKP_SPEC): Likewise.
4444 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
4445 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
4446 (wrestrict_dom_walker::check_call): Likewise.
4447 * gimple.c (gimple_build_call_from_tree): Likewise.
4448 * gimple.h (enum gf_mask): Likewise.
4449 (gimple_call_with_bounds_p): Likewise.
4450 (gimple_call_set_with_bounds): Likewise.
4451 * gimplify.c (gimplify_init_constructor): Likewise.
4452 * ipa-cp.c (initialize_node_lattices): Likewise.
4453 (propagate_constants_across_call): Likewise.
4454 (find_more_scalar_values_for_callers_subset): Likewise.
4455 * ipa-hsa.c (process_hsa_functions): Likewise.
4456 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
4457 * ipa-icf.c (sem_function::merge): Likewise.
4458 * ipa-inline.c (early_inliner): Likewise.
4459 * ipa-pure-const.c (warn_function_noreturn): Likewise.
4460 (warn_function_cold): Likewise.
4461 (propagate_pure_const): Likewise.
4462 * ipa-ref.h (enum GTY): Likewise.
4463 * ipa-split.c (find_retbnd): Likewise.
4464 (consider_split): Likewise.
4465 (split_function): Likewise.
4466 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
4467 * ipa.c (walk_polymorphic_call_targets): Likewise.
4468 (symbol_table::remove_unreachable_nodes): Likewise.
4469 (process_references): Likewise.
4470 (cgraph_build_static_cdtor_1): Likewise.
4471 * lto-cgraph.c (lto_output_node): Likewise.
4472 (output_refs): Likewise.
4473 (compute_ltrans_boundary): Likewise.
4474 (input_overwrite_node): Likewise.
4475 (input_node): Likewise.
4476 (input_cgraph_1): Likewise.
4477 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
4478 * passes.c (pass_manager::execute_early_local_passes): Likewise.
4479 (class pass_chkp_instrumentation_passes): Likewise.
4480 (make_pass_chkp_instrumentation_passes): Likewise.
4481 * passes.def: Likewise.
4482 * rtl.h (struct GTY): Likewise.
4483 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
4484 * stor-layout.c (layout_type): Likewise.
4485 * symtab.c: Likewise.
4486 * target.def: Likewise.
4487 * targhooks.c (default_chkp_bound_type): Likewise.
4488 (default_chkp_bound_mode): Likewise.
4489 (default_builtin_chkp_function): Likewise.
4490 (default_chkp_function_value_bounds): Likewise.
4491 (default_chkp_make_bounds_constant): Likewise.
4492 (default_chkp_initialize_bounds): Likewise.
4493 * targhooks.h (default_chkp_bound_type): Likewise.
4494 (default_chkp_bound_mode): Likewise.
4495 (default_builtin_chkp_function): Likewise.
4496 (default_chkp_function_value_bounds): Likewise.
4497 (default_chkp_make_bounds_constant): Likewise.
4498 (default_chkp_initialize_bounds): Likewise.
4499 * toplev.c (compile_file): Likewise.
4500 (process_options): Likewise.
4501 * tree-core.h (DEF_BUILTIN): Likewise.
4502 (DEF_BUILTIN_CHKP): Likewise.
4503 * tree-inline.c (declare_return_variable): Likewise.
4504 (remap_gimple_stmt): Likewise.
4505 (copy_bb): Likewise.
4506 (initialize_inlined_parameters): Likewise.
4507 (expand_call_inline): Likewise.
4508 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
4509 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
4510 (make_pass_ipa_chkp_produce_thunks): Likewise.
4511 (make_pass_chkp): Likewise.
4512 (make_pass_chkp_opt): Likewise.
4513 (make_pass_chkp_instrumentation_passes): Likewise.
4514 * tree-pretty-print.c (dump_generic_node): Likewise.
4515 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
4516 * tree-ssa-dce.c (propagate_necessity): Likewise.
4517 (eliminate_unnecessary_stmts): Likewise.
4518 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
4519 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
4520 * tree-ssa-sccvn.h: Likewise.
4521 * tree-ssa-strlen.c (get_string_length): Likewise.
4522 (valid_builtin_call): Likewise.
4523 (adjust_last_stmt): Likewise.
4524 (handle_builtin_strchr): Likewise.
4525 (handle_builtin_strcpy): Likewise.
4526 (handle_builtin_stxncpy): Likewise.
4527 (handle_builtin_memcpy): Likewise.
4528 (handle_builtin_strcat): Likewise.
4529 (strlen_check_and_optimize_stmt): Likewise.
4530 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
4531 * tree-streamer-in.c: Likewise.
4532 * tree-streamer.c (record_common_node): Likewise.
4533 * tree.c (tree_code_size): Likewise.
4534 (wide_int_to_tree_1): Likewise.
4535 (type_contains_placeholder_1): Likewise.
4536 (build_common_tree_nodes): Likewise.
4537 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
4538 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
4539 (POINTER_BOUNDS_P): Likewise.
4540 (BOUNDED_TYPE_P): Likewise.
4541 (BOUNDED_P): Likewise.
4542 (CALL_WITH_BOUNDS_P): Likewise.
4543 (pointer_bounds_type_node): Likewise.
4544 * value-prof.c (gimple_ic): Likewise.
4545 * var-tracking.c (vt_add_function_parameters): Likewise.
4546 * varasm.c (make_decl_rtl): Likewise.
4547 (assemble_start_function): Likewise.
4548 (output_constant): Likewise.
4549 (maybe_assemble_visibility): Likewise.
4550 * varpool.c (ctor_for_folding): Likewise.
4551 * chkp-builtins.def: Remove.
4552 * ipa-chkp.c: Remove.
4553 * ipa-chkp.h: Remove.
4554 * rtl-chkp.c: Remove.
4555 * rtl-chkp.h: Remove.
4556 * tree-chkp-opt.c: Remove.
4557 * tree-chkp.c: Remove.
4558 * tree-chkp.h: Remove.
4560 2018-06-07 Carl Love <cel@us.ibm.com>
4562 * gcc/config/rs6000/vsx.md (vextract_fp_from_shorth,
4563 vextract_fp_from_shortl): Add BE support.
4565 2018-06-07 Paul Koning <ni1d@arrl.net>
4567 * compare-elim.c (try_merge_compare): Don't merge compare if
4568 address contains a side effect.
4569 (try_eliminate_compare): Likewise.
4571 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
4573 * config.gcc: Support "tremont".
4574 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
4575 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4577 * config/i386/i386.c (m_TREMONT): Define.
4578 (processor_target_table): Add "tremont".
4579 (PTA_TREMONT): Define.
4580 (ix86_lea_outperforms): Add TARGET_TREMONT.
4581 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
4582 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
4583 and M_INTEL_GOLDMONT_PLUS.
4584 (fold_builtin_cpu): Add "tremont".
4585 (ix86_add_stmt_cost): Add TARGET_TREMONT.
4586 (ix86_option_override_internal): Add "tremont".
4587 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
4588 (processor_type): Add PROCESSOR_TREMONT.
4589 * config/i386/x86-tune.def: Add m_TREMONT.
4590 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
4592 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
4594 * gcc/config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
4595 symbol defined for msp430i* devices to be lower case.
4597 gcc/testsuite/gcc.target/msp430/
4598 * msp430i-device-symbol.c: New test.
4599 * msp430f-device-symbol.c: New test.
4600 * msp430.h: New test header file.
4602 2018-06-07 Richard Biener <rguenther@suse.de>
4604 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
4605 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
4606 Properly wrap signed arithmetic if overflow wraps.
4608 2018-06-07 Jakub Jelinek <jakub@redhat.com>
4610 PR tree-optimization/69615
4611 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
4612 of a cast from a same precision integral SSA_NAME in a bb dominated
4613 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
4614 cast to utype if rhs2 has already a compatible type.
4616 2018-06-07 Richard Biener <rguenther@suse.de>
4618 PR tree-optimization/85935
4619 * graphite-scop-detection.c (find_params_in_bb): Analyze
4620 condition operands with respect to the correct loop. Assert
4621 the analysis doesn't fail.
4623 2018-06-04 Carl Love <cel@us.ibm.com>
4625 * gcc/config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
4626 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
4627 as it is slightly cheaper.
4628 (first_match_or_eos_index_<mode>):
4629 Calculate index using natural element order.
4630 (first_match_index_<mode>):
4631 Calculate index using natural element order.
4632 (first_match_or_eos_index_<mode>):
4633 Calculate index using natural order.
4634 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
4635 for BE and LE modes.
4636 * gcc/config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
4637 P9V_BUILTIN_VCLZLSBB_V16QI.
4638 * gcc/config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
4641 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
4643 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
4644 indentation and line wrap for many prototypes. Add missing
4645 @smallexample directives around block of prototypes for vec_xl and
4648 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
4650 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
4651 track if we pass or return IEEE 128-bit floating point.
4652 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
4653 C++ mangling that is compatible with GCC 8.1.
4654 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
4655 (init_cumulative_args): Note if we pass or return IEEE 128-bit
4656 floating point types.
4657 (rs6000_function_arg_advance_1): Likewise.
4658 (rs6000_mangle_type): Optionally generate mangled names that match
4659 what GCC 8.1 generated for IEEE 128-bit floating point types.
4660 (rs6000_globalize_decl_name): If we have an external function that
4661 passes or returns IEEE 128-bit types, generate a weak reference
4662 from the mangled name used in GCC 8.1 to the current mangled
4664 (rs6000_init_builtins): Make __ibm128 use the long double type if
4665 long double is IBM extended double. Make __float128 use the long
4666 double type if long double is IEEE 128-bit.
4669 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
4670 macro for __ibm128 built-in functions.
4671 (PACK_IF): Add __ibm128 pack/unpack functions.
4672 (UNPACK_IF): Likewise.
4673 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
4674 enable long double built-in functions if long double is IEEE
4675 128-bit floating point.
4676 (rs6000_invalid_builtin): Update long double built-in function
4678 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
4679 functions, adjust the built-in function to use the long double
4680 built-in function if __ibm128 and long double are the same type.
4681 * doc/extend.texi (PowerPC builtins): Update documention for
4682 __builtin_{,un}pack_longdouble. Add documentation for
4683 __builtin_{,un}pack_ibm128.
4685 2018-06-06 Jim Wilson <jimw@sifive.com>
4687 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
4688 (struct machine_function): New field interrupt_mode.
4689 (riscv_handle_type_attribute): New function. Add forward declaration.
4690 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
4691 (riscv_expand_epilogue): Check interrupt_mode field.
4692 (riscv_set_current_function): Check interrupt attribute args and
4693 set interrupt_mode field.
4694 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
4695 (riscv_sret, riscv_uret): New.
4696 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
4697 new arguments to interrupt attribute.
4699 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
4702 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
4703 Don't handle -mcpu=power8 if -mpower9-vector is also used.
4705 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
4707 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
4708 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
4709 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
4710 several redundant entries.
4712 2018-06-06 David Malcolm <dmalcolm@redhat.com>
4714 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
4715 type from "rtx" to "rtx_insn *".
4716 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
4717 for local "call_insn", removing cast.
4718 (ix86_expand_call): Likewise, introducing a "call_insn" local.
4720 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
4722 PR tree-optimization/86066
4723 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
4724 for BIT_INSERT_EXPR stores.
4726 2018-06-06 Richard Biener <rguenther@suse.de>
4728 PR tree-optimization/86062
4729 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
4730 component refs ontop
4731 of to be offsetted base.
4733 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
4735 * gcc/config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
4736 to be static and remove check on interrupt attribute name.
4738 gcc/testsuite/gcc.target/msp430/
4739 * function-attributes-4.c: New test.
4740 * static-interrupts.c: New test.
4742 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
4744 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
4745 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
4747 2018-06-05 Steve Ellcey <sellcey@cavium.com>
4750 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
4752 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
4753 Remove second argument, change how error is called.
4754 (aarch64_layout_arg): Remove second argument from
4755 aarch64_err_no_fpadvsimd call.
4756 (aarch64_init_cumulative_args): Ditto.
4757 (aarch64_gimplify_va_arg_expr): Ditto.
4758 * config/aarch64/aarch64.md (mov<mode>): Ditto.
4760 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
4762 * config/i386/i386.md (simple_return_indirect_internal): New expander.
4763 (*simple_return_indirect_internal<mode>): Rename from
4764 simple_return_indirect_internal. Use W mode iterator.
4765 (rstorssp): New expander.
4766 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
4767 (clrssbsy): New expander.
4768 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
4770 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
4772 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
4774 (cmse_check_pointed_object): Likewise.
4776 2018-06-05 Martin Liska <mliska@suse.cz>
4778 PR gcov-profile/47618
4779 * doc/invoke.texi: Document how -fprofile-dir format
4782 2018-06-05 Richard Biener <rguenther@suse.de>
4784 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
4785 removal pretend DOM info isn't available so we do not update
4786 it and only remove edges, not dominated blocks. Actually free
4787 DOM info in case we removed something. Remove unreachable blocks.
4788 (mfb_keep_latches): Work with either DOM info or marked backedges.
4789 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
4790 first. Mark backedges if DOM info isn't available.
4791 (Re-)compute DOM info after cleanup_control_flow_pre.
4793 2018-06-05 Richard Biener <rguenther@suse.de>
4795 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
4796 (locus_discrim_hasher::hash): Adjust.
4797 (locus_discrim_hasher::equal): Likewise.
4798 (next_discriminator_for_locus): Work on line directly.
4799 (same_line_p): Pass in expanded locus1 as well.
4800 (assign_discriminators): Avoid redundant location expansions.
4802 2018-06-05 Richard Biener <rguenther@suse.de>
4804 PR tree-optimization/86046
4805 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
4806 if required after clearing TREE_ADDRESSABLE.
4808 2018-06-05 Richard Biener <rguenther@suse.de>
4810 PR tree-optimization/86047
4811 * tree-ssa-loop.c (for_each_index): Glob handling of all
4812 decls and constants and really handle all of them.
4814 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4817 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
4818 qualifier_void_pointer and qualifier_const_void_pointer.
4819 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
4820 (arm_init_builtins): Handle the above.
4821 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
4822 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
4823 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
4826 2018-06-05 Martin Liska <mliska@suse.cz>
4828 * auto-profile.c (read_autofdo_file): Do not use
4829 gcov_ctr_summary struct.
4830 (afdo_callsite_hot_enough_for_early_inline): Likewise.
4831 * coverage.c (struct counts_entry): Likewise.
4832 (read_counts_file): Read just single summary entry.
4833 (get_coverage_counts): Use gcov_summary struct.
4834 * coverage.h (get_coverage_counts): Likewise.
4835 * gcov-dump.c (dump_working_sets): Likewise.
4836 (tag_summary): Dump just single summary.
4837 * gcov-io.c (gcov_write_summary): Write just histogram
4839 (gcov_read_summary): Read just single summary.
4840 (compute_working_sets): Use gcov_summary struct.
4841 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
4842 of GCOV_COUNTERS_SUMMABLE.
4843 (GCOV_COUNTERS_SUMMABLE): Remove.
4844 (GCOV_FIRST_VALUE_COUNTER): Replace with
4845 GCOV_COUNTER_V_INTERVAL.
4846 (struct gcov_ctr_summary): Remove.
4847 (struct gcov_summary): Directly use fields of former
4849 (compute_working_sets): Use gcov_summary struct.
4850 * gcov.c (read_count_file): Do not use ctrs fields.
4851 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
4853 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
4855 * profile.c: Likewise.
4856 * profile.h: Likewise.
4858 2018-06-05 Martin Liska <mliska@suse.cz>
4860 PR gcov-profile/84846
4861 * gcov.c (output_lines): Print working directory only
4862 in intermediate format.
4864 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
4866 * config/s390/s390-builtin-types.def: Add void function type.
4867 * config/s390/s390-builtins.def: Use the function type for the
4870 2018-06-04 Jim Wilson <jimw@sifive.com>
4872 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
4874 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
4875 and maybe_eh_return. Change regno to unsigned int. Use new args to
4876 handle EH_RETURN_DATA_REGNO registers properly.
4877 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
4878 (riscv_expand_epilogue): Update comment. Change argument name and
4879 type. Update code to use new name and type. Pass new args to
4880 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
4882 * config/riscv/riscv.md (NORMAL_RETURN): New.
4883 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
4884 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
4885 (eh_return): Call gen_eh_return_internal and emit barrier.
4886 (eh_return_internal): Call riscv_expand_epilogue.
4888 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
4890 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
4891 bit_insertion field and declare can_be_merged_into method.
4892 (merged_store_group::can_be_merged_into): New method.
4893 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
4894 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
4895 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
4897 2018-06-04 Richard Biener <rguenther@suse.de>
4899 PR tree-optimization/85955
4900 * builtins.c (fold_builtin_sincos): Convert pointers to
4901 destination to appropriate type before dereferencing.
4903 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
4905 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
4907 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
4909 * expr.c (expand_expr_real_1): Force the operand into memory if
4910 its TYPE_MODE is BLKmode and if there is no integer mode for
4911 the number of bits being extracted.
4913 2018-06-04 Jakub Jelinek <jakub@redhat.com>
4917 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
4918 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
4920 2018-06-04 Richard Biener <rguenther@suse.de>
4922 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
4923 (cleanup_tree_cfg_noloop): ... single caller. Do
4924 start_recording_case_labels later.
4926 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
4928 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
4929 to _IMMINTRIN_H_INCLUDED.
4930 * config/i386/pconfigintrin.h: Ditto.
4931 * config/i386/waitpkgintrin.h: Ditto.
4932 * config/i386/immintrin.h: Add includes for sgxintrin.h,
4933 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
4934 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
4935 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
4936 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
4937 waitpkgintrin.h and cldemoteintrin.h.
4939 2018-06-04 Richard Biener <rguenther@suse.de>
4941 PR tree-optimization/86038
4942 * tracer.c (find_best_successor): Check probability for
4943 being initialized, bail out if not.
4945 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
4948 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
4949 of bits to ignore when comparing architectures.
4951 2018-06-04 Jakub Jelinek <jakub@redhat.com>
4953 PR tree-optimization/69615
4954 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
4955 maximum or minimum of the type, try to merge it also as if
4956 range1 is + [-, x - 1] or + [x + 1, -].
4959 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
4961 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
4963 PR tree-optimization/86034
4964 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
4965 the unsigned bitfield type in a bit insertion sequence if it does not
4966 have a larger precision than the bitfield size.
4967 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
4969 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
4971 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
4973 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
4975 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
4976 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
4977 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
4978 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
4980 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
4982 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
4983 Disable -fdelete-null-pointer-checks for ELF toolchain.
4985 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
4986 Kito Cheng <kito.cheng@gmail.com>
4988 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
4989 (nds32le-*-*, nds32be-*-*): Integrate checking process.
4990 (nds32*-*-*): Add glibc and uclibc conditions.
4991 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
4992 (TARGET_EXCEPT_UNWIND_INFO): Define.
4993 * config/nds32/elf.h: New file.
4994 * config/nds32/linux.h: New file.
4995 * config/nds32/nds32-elf.opt: New file.
4996 * config/nds32/nds32-linux.opt: New file.
4997 * config/nds32/nds32-fp-as-gp.c
4998 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
4999 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
5001 (nds32_asm_file_end): Ditto.
5002 (nds32_print_operand): Ditto.
5003 (nds32_insert_attributes): Ditto.
5004 (nds32_init_libfuncs): New function.
5005 (TARGET_HAVE_TLS): Define.
5006 (TARGET_INIT_LIBFUNCS): Define.
5007 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
5009 (TARGET_ELF): Apply different mcmodel setting.
5010 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
5011 been migrated into elf.h and linux.h files.
5012 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
5013 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
5014 (mcmodel): The content has been migrated into nds32-elf.opt and
5015 nds32-linux.opt files.
5016 * config/nds32/t-elf: New file.
5017 * config/nds32/t-linux: New file.
5019 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
5020 Shiva Chen <shiva0217@gmail.com>
5022 * config/nds32/constants.md (unspec_volatile_element): Add
5023 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
5024 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
5026 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
5027 (make_pass_nds32_fp_as_gp): Declare.
5028 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
5030 (nds32_asm_function_end_prologue): Remove unused asm output.
5031 (nds32_asm_function_begin_epilogue): Remove unused asm output.
5032 (nds32_asm_file_start): Output necessary fp_as_gp information.
5033 (nds32_option_override): Adjust register usage.
5034 (nds32_expand_prologue): Consider fp_as_gp situation.
5035 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
5036 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
5039 (simple_return): Ditto.
5040 (omit_fp_begin): Output special directive for fp_as_gp.
5041 (omit_fp_end): Output special directive for fp_as_gp.
5042 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
5043 mforbid-fp-as-gp): New options.
5045 2018-06-01 Mark Wielaard <mark@klomp.org>
5047 * dwarf2out.c (dwarf2out_finish): Remove generation of
5048 DW_AT_loclists_base.
5050 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
5052 * gimple-ssa-store-merging.c: Include gimple-fold.h.
5053 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
5054 (struct merged_store_group): Add bit_insertion field.
5055 (dump_char_array): Use standard hexadecimal format.
5056 (merged_store_group::merged_store_group): Set bit_insertion to false.
5057 (merged_store_group::apply_stores): Use optimal buffer size. Deal
5058 with BIT_INSERT_EXPR stores. Move up code updating the mask and
5059 also print the mask in the dump file.
5060 (pass_store_merging::gate): Minor tweak.
5061 (imm_store_chain_info::coalesce_immediate): Fix wrong association
5062 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
5063 stores with INTEGER_CST stores.
5064 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
5065 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
5066 and use it throughout. Generate bit insertion sequences if need be.
5067 (pass_store_merging::process_store): Remove redundant condition.
5068 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
5070 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
5072 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
5073 the 128-bit floating point types. Fix function comment.
5075 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5077 * config/aarch64/aarch64-simd.md
5078 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
5080 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
5083 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
5085 PR tree-optimization/85989
5086 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
5088 (backprop::intersect_uses): Check it when deciding whether this
5089 is a backedge reference.
5090 (backprop::process_block): Add each phi to m_visited_phis
5091 after visiting it, then clear it at the end.
5093 2018-06-01 Richard Biener <rguenther@suse.de>
5095 * tree-vectorizer.h (vect_dr_stmt): New function.
5096 (vect_get_load_cost): Adjust.
5097 (vect_get_store_cost): Likewise.
5098 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
5099 Use vect_dr_stmt instead of DR_SMTT.
5100 (vect_record_base_alignments): Likewise.
5101 (vect_calculate_target_alignment): Likewise.
5102 (vect_compute_data_ref_alignment): Likewise and make static.
5103 (vect_update_misalignment_for_peel): Likewise.
5104 (vect_verify_datarefs_alignment): Likewise.
5105 (vector_alignment_reachable_p): Likewise.
5106 (vect_get_data_access_cost): Likewise. Pass down
5107 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
5108 (vect_get_peeling_costs_all_drs): Likewise.
5109 (vect_peeling_hash_get_lowest_cost): Likewise.
5110 (vect_enhance_data_refs_alignment): Likewise.
5111 (vect_find_same_alignment_drs): Likewise.
5112 (vect_analyze_data_refs_alignment): Likewise.
5113 (vect_analyze_group_access_1): Likewise.
5114 (vect_analyze_group_access): Likewise.
5115 (vect_analyze_data_ref_access): Likewise.
5116 (vect_analyze_data_ref_accesses): Likewise.
5117 (vect_vfa_segment_size): Likewise.
5118 (vect_small_gap_p): Likewise.
5119 (vectorizable_with_step_bound_p): Likewise.
5120 (vect_prune_runtime_alias_test_list): Likewise.
5121 (vect_analyze_data_refs): Likewise.
5122 (vect_supportable_dr_alignment): Likewise.
5123 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
5124 (vect_gen_prolog_loop_niters): Likewise.
5125 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
5126 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
5128 (vect_recog_mask_conversion_pattern): Likewise.
5129 (vect_try_gather_scatter_pattern): Likewise.
5130 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
5131 to vect_get_store_cost.
5132 (vect_get_store_cost): Get stmt_info instead of DR.
5133 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
5134 (vect_get_load_cost): Get stmt_info instead of DR.
5136 2018-06-01 Richard Biener <rguenther@suse.de>
5139 * gimple-fold.c (var_decl_component_p): Also allow offsetted
5140 vars wrapped in MEM_REFs.
5142 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
5144 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
5145 Fix subreg tests so that we only return a choice between
5146 GENERAL_REGS and FP_REGS if the original classes included both.
5148 2018-06-01 Richard Biener <rguenther@suse.de>
5151 * tree-ssa-structalias.c (get_function_part_constraint):
5152 Handle NULL fi->decl.
5153 (find_func_aliases_for_call): Properly handle indirect
5154 fi from direct call.
5155 (find_func_clobbers): Likewise.
5156 (ipa_pta_execute): Likewise.
5157 (create_variable_info_for): For functions that are ifunc_resolver
5158 resolve to a varinfo that contains the result of the resolver call.
5159 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
5162 2018-05-31 Michael Collison <michael.collison@arm.com>
5164 * config/aarch64/aarch64.md:
5165 (*fix_to_zero_extenddfdi2): New pattern.
5166 * gcc.target/aarch64/fix_extend1.c: New testcase.
5168 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
5172 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
5173 and BUILT_IN_STRNCMP_EQ.
5174 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
5175 BUILT_IN_STRNCMP_EQ.
5176 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
5177 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
5178 (gimple_fold_builtin): Likewise.
5179 * tree-ssa-strlen.c (compute_string_length): New function.
5180 (determine_min_obsize): New function.
5181 (handle_builtin_string_cmp): New function to handle calls to
5182 string compare functions.
5183 (strlen_optimize_stmt): Add handling to builtin string compare
5185 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
5186 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
5187 * tree.c (build_common_builtin_nodes): Add new defines of
5188 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
5190 2018-05-31 Jakub Jelinek <jakub@redhat.com>
5193 * bb-reorder.c (pass_partition_blocks::gate): Return false for
5194 functions with naked attribute.
5196 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
5198 * config/i386/sse.md (avx_vec_concat<mode>):
5199 Substitute concat_tg_mode mode attribute with xtg_mode.
5200 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
5201 (concat_tg_mode): Remove mode attribute.
5203 2018-05-31 Martin Sebor <msebor@redhat.com>
5206 * calls.c (alloc_max_size): Correct a logic error/typo.
5207 Treat excessive arguments as infinite. Warn for invalid arguments.
5208 * doc/invoke.texi (-Walloc-size-larger-than): Update.
5210 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
5213 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
5214 and movx for Haswell.
5216 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
5217 Cesar Philippidis <cesar@codesourcery.com>
5220 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
5221 when emitting error on private/firstprivate reductions.
5222 * omp-low.c (lower_omp_target): Avoid reference-type processing
5223 on pointers for firstprivate clause.
5225 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
5227 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
5230 * config/aarch64/aarch64-simd.md
5231 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
5232 (aarch64_ld1_x3_<mode>): Likewise
5233 (aarch64_st1x2<VALLDIF:mode>): Likewise
5234 (aarch64_st1_x2_<mode>): Likewise
5235 (aarch64_st1x3<VALLDIF:mode>): Likewise
5236 (aarch64_st1_x3_<mode>): Likewise
5237 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
5238 (vld1_s8_x3): Likewise.
5239 (vld1_u16_x3): Likewise.
5240 (vld1_s16_x3): Likewise.
5241 (vld1_u32_x3): Likewise.
5242 (vld1_s32_x3): Likewise.
5243 (vld1_u64_x3): Likewise.
5244 (vld1_s64_x3): Likewise.
5245 (vld1_f16_x3): Likewise.
5246 (vld1_f32_x3): Likewise.
5247 (vld1_f64_x3): Likewise.
5248 (vld1_p8_x3): Likewise.
5249 (vld1_p16_x3): Likewise.
5250 (vld1_p64_x3): Likewise.
5251 (vld1q_u8_x3): Likewise.
5252 (vld1q_s8_x3): Likewise.
5253 (vld1q_u16_x3): Likewise.
5254 (vld1q_s16_x3): Likewise.
5255 (vld1q_u32_x3): Likewise.
5256 (vld1q_s32_x3): Likewise.
5257 (vld1q_u64_x3): Likewise.
5258 (vld1q_s64_x3): Likewise.
5259 (vld1q_f16_x3): Likewise.
5260 (vld1q_f32_x3): Likewise.
5261 (vld1q_f64_x3): Likewise.
5262 (vld1q_p8_x3): Likewise.
5263 (vld1q_p16_x3): Likewise.
5264 (vld1q_p64_x3): Likewise.
5265 (vst1_s64_x2): Likewise.
5266 (vst1_u64_x2): Likewise.
5267 (vst1_f64_x2): Likewise.
5268 (vst1_s8_x2): Likewise.
5269 (vst1_p8_x2): Likewise.
5270 (vst1_s16_x2): Likewise.
5271 (vst1_p16_x2): Likewise.
5272 (vst1_s32_x2): Likewise.
5273 (vst1_u8_x2): Likewise.
5274 (vst1_u16_x2): Likewise.
5275 (vst1_u32_x2): Likewise.
5276 (vst1_f16_x2): Likewise.
5277 (vst1_f32_x2): Likewise.
5278 (vst1_p64_x2): Likewise.
5279 (vst1q_s8_x2): Likewise.
5280 (vst1q_p8_x2): Likewise.
5281 (vst1q_s16_x2): Likewise.
5282 (vst1q_p16_x2): Likewise.
5283 (vst1q_s32_x2): Likewise.
5284 (vst1q_s64_x2): Likewise.
5285 (vst1q_u8_x2): Likewise.
5286 (vst1q_u16_x2): Likewise.
5287 (vst1q_u32_x2): Likewise.
5288 (vst1q_u64_x2): Likewise.
5289 (vst1q_f16_x2): Likewise.
5290 (vst1q_f32_x2): Likewise.
5291 (vst1q_f64_x2): Likewise.
5292 (vst1q_p64_x2): Likewise.
5293 (vst1_s64_x3): Likewise.
5294 (vst1_u64_x3): Likewise.
5295 (vst1_f64_x3): Likewise.
5296 (vst1_s8_x3): Likewise.
5297 (vst1_p8_x3): Likewise.
5298 (vst1_s16_x3): Likewise.
5299 (vst1_p16_x3): Likewise.
5300 (vst1_s32_x3): Likewise.
5301 (vst1_u8_x3): Likewise.
5302 (vst1_u16_x3): Likewise.
5303 (vst1_u32_x3): Likewise.
5304 (vst1_f16_x3): Likewise.
5305 (vst1_f32_x3): Likewise.
5306 (vst1_p64_x3): Likewise.
5307 (vst1q_s8_x3): Likewise.
5308 (vst1q_p8_x3): Likewise.
5309 (vst1q_s16_x3): Likewise.
5310 (vst1q_p16_x3): Likewise.
5311 (vst1q_s32_x3): Likewise.
5312 (vst1q_s64_x3): Likewise.
5313 (vst1q_u8_x3): Likewise.
5314 (vst1q_u16_x3): Likewise.
5315 (vst1q_u32_x3): Likewise.
5316 (vst1q_u64_x3): Likewise.
5317 (vst1q_f16_x3): Likewise.
5318 (vst1q_f32_x3): Likewise.
5319 (vst1q_f64_x3): Likewise.
5320 (vst1q_p64_x3): Likewise.
5322 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
5324 * config/msp430/msp430.c (msp430_output_labelref): Prepend
5325 user_label_prefix to name.
5327 * tree-core.h: Update comment about the format of NAME string
5328 passed to handler in attribute_spec.
5330 * config/msp430/msp430.md: Remove erroneous subreg expression from
5331 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
5332 zero_extend{q,h}isi2.
5334 2018-05-30 Borislav Petkov <bp@suse.de>
5336 * doc/extend.texi: Document some architecture specific
5337 constraints and sort entries.
5339 2018-05-30 Martin Sebor <msebor@redhat.com>
5342 * builtins.c (expand_builtin_stpcpy_1): New function.
5343 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
5344 only if the former succeeds.
5346 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
5348 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
5351 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
5353 * doc/invoke.texi (-flinker-output): Document
5355 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
5357 * passes.c (ipa_write_summaries): Only modify statements if body
5359 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
5360 incrementally linking.
5361 (ipa_passes): Likewise.
5362 * lto-cgraph.c (lto_output_node): When incrementally linking do not
5363 pass down resolution info.
5364 * common.opt (flag_incremental_link): Update info.
5365 * gcc.c (plugin specs): Turn flinker-output=* to
5366 -plugin-opt=-linker-output-known
5367 * toplev.c (compile_file): Also cut compilation when doing incremental
5369 * flag-types. (enum lto_partition_model): Add
5370 LTO_LINKER_OUTPUT_NOLTOREL.
5371 (invoke.texi): Add -flinker-output docs.
5372 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
5373 link same way as WPA; do not stream in dead initializers.
5375 * dwarf2out.c (dwarf2out_die_ref_for_decl,
5376 darf2out_register_external_decl): Support incremental link.
5378 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
5380 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
5382 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
5384 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
5385 it down to simple_object_copy_lto_debug_sections.
5386 (run_gcc): Determine incremental LTO link time and configure
5387 lto1 into non-wpa mode, disable renaming of debug sections.
5389 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
5391 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
5392 descriptions of various incorrectly documented functions.
5394 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
5397 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
5398 address check not strict.
5400 2018-05-30 Richard Biener <rguenther@suse.de>
5402 PR tree-optimization/85964
5403 * tracer.c (better_p): Drop initialized count check, we only
5404 call the function with initialized counts now.
5405 (find_best_successor): Do find a best edge if one
5406 has uninitialized count.
5407 (find_best_predecessor): Likewise. Do BB frequency check only
5408 if count is initialized.
5410 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
5412 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
5413 (aarch64_ldrstr_offset_compare): New.
5414 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
5415 load/store orderings.
5416 (aarch64_gen_adjusted_ldpstp): Likewise.
5418 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
5420 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
5421 Check for subset of GENERAL_REGS and FP_REGS.
5422 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
5425 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
5427 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
5428 and wi::to_poly_offset. Add the current offset and then check
5429 whether the sum fits, rather than using an unchecked addition of
5430 a checked term. Check for a shwi rather than a uhwi.
5431 * expr.c (get_bit_range): Use tree_to_poly_uint64.
5432 (store_constructor): Use poly_int_tree_p.
5433 (expand_expr_real_1): Likewise.
5434 * function.c (assign_temp): Likewise.
5435 * fold-const.c (const_binop): Use poly_int_tree_p and
5437 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
5439 * ipa-icf-gimple.c (func_checker::compare_operand): Use
5440 to_poly_offset for MEM offsets.
5441 * ipa-icf.c (sem_variable::equals): Likewise.
5442 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
5443 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
5444 wi::to_poly_offset for BIT_FIELD_REF offsets.
5445 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
5447 * var-tracking.c (emit_note_insn_var_location): Use
5448 tree_to_poly_uint64.
5450 2018-05-29 Jim Wilson <jimw@sifive.com>
5452 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
5454 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
5457 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
5458 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
5460 (sse4_1_round<mode>2): Use nonimmediate_operand
5461 for operand 1 predicate.
5463 2018-05-29 Martin Sebor <msebor@redhat.com>
5464 Richard Biener <rguenther@suse.de>
5467 * calls.c (get_size_range): Call determine_value_range instead
5468 of get_value_range..
5469 * tree-vrp.h (determine_value_range): Declared new function.
5470 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
5472 2018-05-29 Richard Biener <rguenther@suse.de>
5474 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
5475 sure to use non-pattern stmts for get_earlier_stmt arguments.
5476 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
5477 called on pattern stmts.
5478 (get_later_stmt): Likewise.
5480 2018-05-29 Martin Liska <mliska@suse.cz>
5482 PR gcov-profile/85759
5483 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
5486 2018-05-29 Jakub Jelinek <jakub@redhat.com>
5488 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
5490 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
5494 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
5495 VEC_PACK_FLOAT_EXPR): New tree codes.
5496 * tree-pretty-print.c (op_code_prio): Handle
5497 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
5498 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
5499 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
5500 * tree-inline.c (estimate_operator_cost): Likewise.
5501 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
5502 * fold-const.c (const_binop): Likewise.
5503 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
5504 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
5505 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
5506 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
5507 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
5508 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
5509 * expr.c (expand_expr_real_2): Likewise.
5510 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
5511 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
5512 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
5514 * optabs.c (expand_widen_pattern_expr): For
5515 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
5516 sign from result type rather than operand's type.
5517 (expand_binop_directly): For vec_packu_float_optab and
5518 vec_packs_float_optab allow result type to be different from operand's
5520 * optabs-tree.c (optab_for_tree_code): Handle
5521 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
5522 VEC_PACK_FLOAT_EXPR. Formatting fixes.
5523 * tree-vect-generic.c (expand_vector_operations_1): Handle
5524 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
5525 VEC_PACK_FLOAT_EXPR.
5526 * tree-vect-stmts.c (supportable_widening_operation): Handle
5528 (supportable_narrowing_operation): Handle FLOAT_EXPR.
5529 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
5530 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
5531 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
5532 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
5534 (vec_pack<floatprefix>_float_<mode>): New expander.
5535 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
5537 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
5538 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
5539 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
5540 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
5541 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
5543 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
5544 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
5545 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
5546 VEC_PACK_FLOAT_EXPR): Document.
5548 2018-05-29 Richard Biener <rguenther@suse.de>
5550 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
5552 (stmt_vec_info_vec): Make pointer.
5553 (init_stmt_vec_info_vec): Remove.
5554 (free_stmt_vec_info_vec): Likewise.
5555 (set_stmt_vec_info_vec): New function.
5556 (free_stmt_vec_infos): Likewise.
5557 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
5558 (set_vinfo_for_stmt): Likewise.
5559 (get_earlier_stmt): Likewise.
5560 (get_later_stmt): Likewise.
5561 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
5562 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
5563 (vec_info::~vec_info): Free stmt_vec_infos.
5564 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
5565 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
5566 (pass_slp_vectorize::execute): Likewise.
5567 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
5568 (free_stmt_vec_info_vec): Likewise.
5569 (set_stmt_vec_info_vec): New function.
5570 (free_stmt_vec_infos): Likewise.
5571 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
5572 the global stmt_vec_info_vec.
5573 * tree-parloops.c (gather_scalar_reductions): Use
5574 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
5577 2018-05-29 Richard Biener <rguenther@suse.de>
5579 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
5581 2018-05-29 Martin Liska <mliska@suse.cz>
5582 David Malcolm <dmalcolm@redhat.com>
5584 * vec.c (test_reverse): New.
5585 (vec_c_tests): Add new test.
5586 * vec.h (vl_ptr>::reverse): New function.
5588 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
5590 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
5592 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
5595 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
5597 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
5599 2018-05-28 Richard Biener <rguenther@suse.de>
5601 PR tree-optimization/85933
5602 * tree-vect-data-refs.c (vect_record_base_alignments): Only
5603 look at stmts marked as vectorizable.
5605 2018-05-28 Richard Biener <rguenther@suse.de>
5607 PR tree-optimization/85934
5608 * tree-vect-generic.c (expand_vector_operations_1): Hoist
5609 vector boolean check before scalar optimization.
5611 2018-05-28 Jakub Jelinek <jakub@redhat.com>
5613 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
5616 2018-05-28 Mark Wielaard <mark@klomp.org>
5618 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
5619 if it is an expression containing a minus sign.
5621 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
5623 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
5625 2018-05-27 Paul Koning <ni1d@arrl.net>
5627 * config/pdp11/pdp11.md (truncsihi2): Remove.
5629 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
5630 Chung-Ju Wu <jasonwucj@gmail.com>
5632 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
5634 (unaligned_store_dw): Ditto.
5635 * config/nds32/nds32-memory-manipulation.c
5636 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
5637 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
5638 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
5639 (emit_setmem_word_loop): Rename to ...
5640 (emit_setmem_doubleword_loop): ... this.
5641 (nds32_gen_dup_4_byte_to_word_value): New function.
5642 (nds32_gen_dup_8_byte_to_double_word_value): New function.
5643 (nds32_expand_setmem_loop): Refine implementation.
5644 (nds32_expand_setmem_loop_v3m): Ditto.
5645 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
5648 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
5650 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
5652 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
5654 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
5655 (nds32_init_machine_status): Initialize machine->attr_naked_p and
5656 machine->attr_no_prologue_p.
5657 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
5658 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
5659 (nds32_expand_epilogue): Consider attr_naked_p.
5660 (nds32_expand_epilogue_v3pop): Likewise.
5661 (nds32_can_use_return_insn): Likewise.
5662 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
5663 attr_no_prologue_p fields.
5664 * config/nds32/nds32.opt (mret-in-naked-func): New option.
5666 2018-05-27 Jakub Jelinek <jakub@redhat.com>
5669 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
5671 * config/i386/sse.md
5672 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
5674 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
5676 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
5678 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
5680 (*<floatsuffix>floatv2div2sf2): Rename to ...
5681 (*float<floatunssuffix>v2div2sf2): ... this.
5682 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
5683 (float<floatunssuffix>v2div2sf2_mask): ... this.
5684 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
5685 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
5686 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
5688 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
5690 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
5692 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
5694 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
5696 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
5698 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
5699 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
5700 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
5701 gen_ufix_truncv8dfv8si2.
5702 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
5703 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
5704 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
5705 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
5706 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
5707 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
5708 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
5709 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
5711 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
5715 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
5717 2018-05-25 Jim Wilson <jimw@sifive.com>
5719 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
5720 * config/riscv/riscv.c (struct machine_function): Add
5721 interrupt_handler_p and attribute_checked_p fields.
5722 (riscv_attribute_table): Add interrupt.
5723 (riscv_interrupt_type_p): New.
5724 (riscv_save_reg_p): Save extra regs for interrupt handler.
5725 (riscv_use_save_libcall): Return false for interrupt handler.
5726 (riscv_first_stack_step): Add forward declaration.
5727 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
5728 for interrupt handler with large frame. Use it for saved reg list.
5729 (riscv_expand_prologue): Move flag_stack_usage_info support to
5730 eliminate duplication.
5731 (riscv_expand_epilogue): Generate mret for interrupt handler.
5732 (riscv_epilogue_uses): New.
5733 (riscv_can_use_return_insn): Return false for interrupt handler.
5734 (riscv_function_ok_for_sibcall): Likewise.
5735 (riscv_set_current_function): Add interrupt handler support.
5736 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
5737 * config/riscv/riscv.md (UNSPECV_MRET): New.
5739 (riscv_frflags, riscv_fsflags): Use tab after opcode.
5741 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
5743 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
5745 PR tree-optimization/85712
5746 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
5747 this candidate has already been replaced in-situ by a copy.
5749 2018-05-25 Jason Merrill <jason@redhat.com>
5751 PR c++/80485 - inline function non-zero address.
5752 * symtab.c (nonzero_address): Check DECL_COMDAT.
5754 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
5757 * config/alpha/alpha.md (ashlsi3): New insn pattern.
5758 (*ashlsi_se): Rename from *ashldi_se. Define as sign
5759 extension of SImode operation. Use const123_operand predicate.
5760 (*saddsi_1): Remove.
5761 (*saddl_se_1): Ditto.
5763 (*ssubl_se_1): Ditto.
5764 * config/alpha/predicates.md (const123_operand): New predicate.
5765 * config/alpha/constraints.md (P): Use IN_RANGE.
5767 2018-05-25 Richard Biener <rguenther@suse.de>
5769 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
5771 (ref_maybe_used_by_stmt_p): Likewise.
5772 (stmt_may_clobber_ref_p): Likewise.
5773 (stmt_may_clobber_ref_p_1): Likewise.
5774 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
5776 (ref_maybe_used_by_stmt_p): Likewise.
5777 (stmt_may_clobber_ref_p): Likewise.
5778 (stmt_may_clobber_ref_p_1): Likewise.
5779 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
5780 the alias oracle to disambiguate DRs with stmts DR analysis
5782 (vect_analyze_data_refs): Do not give up on not analyzable
5783 DRs for BB vectorization. Remove code truncating the dataref
5786 2018-05-25 Jakub Jelinek <jakub@redhat.com>
5789 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
5790 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
5791 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
5793 2018-05-25 Richard Biener <rguenther@suse.de>
5795 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
5796 function, combining stmt data ref gathering and fatal analysis
5798 (vect_analyze_data_refs): Remove now redudnant code and simplify.
5799 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
5800 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
5801 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
5802 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
5804 2018-05-25 Bin Cheng <bin.cheng@arm.com>
5806 PR tree-optimization/85720
5807 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
5808 SCC if all partitions are builtins.
5809 (version_loop_by_alias_check): New parameter. Generate cancelable
5810 runtime alias check if all partitions are builtins.
5811 (distribute_loop): Update call to above function.
5813 2018-05-25 Bin Cheng <bin.cheng@arm.com>
5815 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
5816 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
5817 (parm_default_def_partition_arg): Ditto.
5818 (set_parm_default_def_partition): Ditto.
5819 (get_parm_default_def_partitions): Ditto and make it static.
5820 (get_undefined_value_partitions): Ditto and make it static.
5821 (remove_ssa_form): Refactor call to init_var_map here.
5822 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
5823 computation for loop region.
5824 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
5825 (register_default_def): Delete.
5826 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
5827 (parm_default_def_partition_arg): Ditto.
5828 (set_parm_default_def_partition): Ditto.
5829 (get_parm_default_def_partitions): Ditto and make it static.
5830 (get_undefined_value_partitions): Ditto and make it static.
5831 (coalesce_with_default, coalesce_with_default): Update comment.
5832 (create_coalesce_list_for_region): New func factored out from
5833 create_outofssa_var_map.
5834 (populate_coalesce_list_for_outofssa): New func factored out from
5835 create_outofssa_var_map and coalesce_ssa_name.
5836 (create_outofssa_var_map): Delete.
5837 (coalesce_ssa_name): Refactor to support live range computation.
5838 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
5839 (get_parm_default_def_partitions): Delete.
5840 (get_undefined_value_partitions): Ditto.
5841 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
5842 computation for loop region.
5843 (new_tree_live_info, loe_visit_block): Ditto.
5844 (live_worklist, set_var_live_on_entry): Ditto.
5845 (calculate_live_on_exit, verify_live_on_entry): Ditto.
5846 * tree-ssa-live.h (struct _var_map): New fields.
5847 (init_var_map): Change decl.
5848 (region_contains_p): New.
5850 2018-05-25 Bin Cheng <bin.cheng@arm.com>
5852 * tree-ssa-live.h (live_merge_and_clear): Delete.
5854 2018-05-25 Richard Biener <rguenther@suse.de>
5857 * tree-dump.c (dequeue_and_dump): Remove access to removed
5858 operand 2 of a SWITCH_EXPR.
5860 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
5862 * doc/sourcebuild.texi (vect_double_cond_arith): Include
5863 multiplication and division.
5864 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
5865 (cond_udiv@var{m}, cond_umod@var{m}): Document.
5866 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
5867 (cond_udiv_optab, cond_umod_optab): New optabs.
5868 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
5869 (IFN_COND_RDIV): New internal functions.
5870 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
5871 TRUNC_MOD_EXPR and RDIV_EXPR.
5872 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
5873 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
5875 (SVE_INT_BINARY): Include mult.
5876 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
5877 (optab, sve_int_op): Handle mult.
5878 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
5880 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
5881 for SVE_INT_BINARY_SD.
5883 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
5885 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
5886 (optab, sve_int_op): Handle div and udiv.
5887 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
5888 for SVE_INT_BINARY_SD.
5889 (*<optab><mode>3): New insn for the same.
5891 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
5893 * tree-vect-patterns.c: Include predict.h.
5894 (vect_recog_divmod_pattern): Restrict check for division support
5895 to when optimizing for size.
5897 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
5899 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
5900 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
5901 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
5902 (gimple_match_op::set_op): Likewise.
5903 (gimple_resimplify4): Declare.
5904 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
5905 (expr::gen_transform): Likewise.
5906 (decision_tree::gen): Generate a simplification routine for 4 operands.
5907 * gimple-match-head.c (gimple_simplify): Add an overload for
5908 4 operands. In the top-level function, handle up to 4 call
5909 arguments and call gimple_resimplify4.
5910 (gimple_resimplify4): New function.
5911 (build_call_internal): Pass a fourth operand.
5912 (maybe_push_to_seq): Likewise.
5913 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
5914 Fold VEC_COND_EXPRs of an operation and a default value into
5915 an IFN_COND_* function if possible.
5916 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
5918 (SVE_COND_FP_BINARY): Include them.
5919 (optab, sve_fp_op): Handle them.
5920 (SVE_INT_BINARY_REV): New code iterator.
5921 (SVE_COND_FP_BINARY_REV): New int iterator.
5922 (commutative): New int attribute.
5923 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
5925 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
5927 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
5928 (*cond_<optab><mode>): New patterns for reversed operands.
5930 2018-05-25 Richard Biener <rguenther@suse.de>
5932 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
5933 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
5934 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
5935 (STMT_VINFO_GROUPED_ACCESS): Adjust.
5936 * tree-vect-data-refs.c (everywhere): Adjust users.
5937 * tree-vect-loop.c (everywhere): Likewise.
5938 * tree-vect-slp.c (everywhere): Likewise.
5939 * tree-vect-stmts.c (everywhere): Likewise.
5940 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
5942 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5944 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
5946 (gcc_cv_as_section_exclude): ... this.
5947 Try Solaris as #exclude syntax.
5948 * configure: Regenerate.
5949 * config.in: Regenerate.
5950 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
5952 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
5953 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
5955 * varasm.c (default_elf_asm_named_section): Don't check if
5956 HAVE_GAS_SECTION_EXCLUDE is defined.
5958 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
5960 * doc/md.texi: Update the documentation of the cond_* optabs
5961 to mention the new final operand. Fix GET_MODE_NUNITS call.
5962 Describe the scalar case too.
5963 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
5964 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
5966 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
5967 (get_conditional_internal_fn): Update comment.
5968 * tree-vect-loop.c (vectorizable_reduction): Pass the original
5969 accumulator value as a final argument to conditional functions.
5970 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
5971 a define_expand and add an "else" operand. Assert for now that
5972 the else operand is equal to operand 2. Use SVE_INT_BINARY and
5973 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
5974 (*cond_<optab><mode>): New patterns.
5975 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
5976 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
5977 (UNSPEC_COND_EOR): Delete.
5978 (optab): Remove associated mappings.
5979 (SVE_INT_BINARY): New code iterator.
5980 (sve_int_op): Remove int attribute and add "minus" to the code
5982 (SVE_COND_INT_OP): Delete.
5983 (SVE_COND_FP_OP): Rename to...
5984 (SVE_COND_FP_BINARY): ...this.
5986 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
5988 * optabs.c (can_reuse_operands_p): New function.
5989 (maybe_legitimize_operands): Try to reuse the results for
5992 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
5994 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
5995 Add {q} suffix to insn mnemonic.
5997 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
5999 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
6000 (msp430_warn_func_return): New.
6002 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
6004 * fold-const.c (tree_nonzero_bits): New function.
6005 * fold-const.h (tree_nonzero_bits): Likewise.
6006 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
6007 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
6009 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
6013 * varasm.c (assemble_alias): Check ifunc_resolver only on
6016 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
6019 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
6020 when memory input operand is handled.
6022 2018-05-24 Luis Machado <luis.machado@linaro.org>
6024 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
6026 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
6028 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
6030 * match.pd: Delay FMA folds until after vectorization.
6032 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
6035 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
6036 address check not strict.
6038 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
6040 * gimple-match.h (gimple_match_op): New class.
6041 (mprts_hook): Replace parameters with a gimple_match_op *.
6042 (maybe_build_generic_op): Likewise.
6043 (gimple_simplified_result_is_gimple_val): Replace parameters with
6044 a const gimple_match_op *.
6045 (gimple_simplify): Replace code_helper * and tree * parameters with
6046 a gimple_match_op * parameter.
6047 (gimple_resimplify1): Replace code_helper *, tree and tree *
6048 parameters with a gimple_match_op * parameter.
6049 (gimple_resimplify2): Likewise.
6050 (gimple_resimplify3): Likewise.
6051 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
6052 parameters with a gimple_match_op * parameter.
6053 * gimple-match-head.c (gimple_simplify): Change prototypes of
6054 auto-generated functions to take a gimple_match_op * instead of
6055 separate code_helper * and tree * parameters. Make the same
6056 change in the top-level overload and update calls to the
6057 gimple_resimplify routines. Update calls to the auto-generated
6058 functions and to maybe_push_res_to_seq in the publicly-facing
6059 operation-specific gimple_simplify overloads.
6060 (gimple_match_op::MAX_NUM_OPS): Define.
6061 (gimple_resimplify1): Replace rcode and ops with a single res_op
6062 parameter. Update call to gimple_simplify.
6063 (gimple_resimplify2): Likewise.
6064 (gimple_resimplify3): Likewise.
6065 (mprts_hook): Replace parameters with a gimple_match_op *.
6066 (maybe_build_generic_op): Likewise.
6067 (build_call_internal): Replace type, nargs and ops with
6068 a gimple_match_op *.
6069 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
6070 with a single gimple_match_op *. Update calls to mprts_hook,
6071 build_call_internal and gimple_simplified_result_is_gimple_val.
6072 Factor out code that is common to the tree_code and combined_fn cases.
6073 * genmatch.c (expr::gen_transform): Replace tem_code and
6074 tem_ops with a gimple_match_op called tem_op. Update calls
6075 to the gimple_resimplify functions and maybe_push_res_to_seq.
6076 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
6077 res_ops. Update call to the gimple_resimplify functions.
6078 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
6079 (decision_tree::gen): Make the functions take a gimple_match_op *
6080 called res_op instead of separate res_code and res_ops parameters.
6081 Update call accordingly.
6082 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
6083 and ops with a single res_op parameter. Update calls to
6084 maybe_build_generic_op and maybe_push_res_to_seq.
6085 (fold_stmt_1): Update calls to gimple_simplify and
6086 replace_stmt_with_simplification.
6087 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
6088 and gimple_simplified_result_is_gimple_val.
6089 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
6091 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
6092 with a gimple_match_op *.
6093 (vn_nary_build_or_lookup): Likewise. Update call to
6094 vn_nary_build_or_lookup_1.
6095 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
6096 gimple_match_op *. Update calls to the gimple_resimplify routines
6097 and to gimple_simplified_result_is_gimple_val.
6098 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
6099 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
6100 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
6101 (visit_nary_op): Likewise.
6102 (visit_reference_op_load): Likewise.
6104 2018-05-23 Luis Machado <luis.machado@linaro.org>
6106 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
6107 modifier for printing the step amount.
6109 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6112 * gcc/tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
6115 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
6117 * doc/sourcebuild.texi (Endianness): New subsubsection.
6119 2018-05-23 Luis Machado <luis.machado@linaro.org>
6121 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
6122 <prefetch_dynamic_strides>: New const bool field.
6123 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
6124 prefetch_dynamic_strides.
6125 (exynosm1_prefetch_tune): Likewise.
6126 (thunderxt88_prefetch_tune): Likewise.
6127 (thunderx_prefetch_tune): Likewise.
6128 (thunderx2t99_prefetch_tune): Likewise.
6129 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
6131 (aarch64_override_options_internal): Update to set
6132 PARAM_PREFETCH_DYNAMIC_STRIDES.
6133 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
6134 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
6135 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
6136 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
6137 prefetch-dynamic-strides setting.
6139 2018-05-23 Luis Machado <luis.machado@linaro.org>
6141 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
6142 <minimum_stride>: New const int field.
6143 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
6144 minimum_stride field defaulting to -1.
6145 (exynosm1_prefetch_tune): Likewise.
6146 (thunderxt88_prefetch_tune): Likewise.
6147 (thunderx_prefetch_tune): Likewise.
6148 (thunderx2t99_prefetch_tune): Likewise.
6149 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
6150 <default_opt_level>: Set to 3.
6151 (aarch64_override_options_internal): Update to set
6152 PARAM_PREFETCH_MINIMUM_STRIDE.
6153 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
6154 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
6155 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
6156 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
6157 stride is constant and is below the minimum stride threshold.
6159 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6161 * config/arm/arm-cpus.in (mode26): Delete.
6162 (armv4): Delete mode26 reference.
6163 * config/arm/arm.c (arm_configure_build_target): Delete use of
6166 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
6168 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
6170 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
6171 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
6173 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
6174 (floatunsdidf2): Ditto.
6176 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
6178 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
6179 (fixuns_trunc<mode>si2_avx512f): Ditto.
6180 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
6181 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
6182 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
6184 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
6186 PR rtl-optimization/79985
6187 * df-scan.c (df_insn_refs_collect): Remove special case for
6188 global registers and asm statements.
6190 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
6192 * extend.texi (Global Register Variables): Rewrite the bullet list.
6193 Note that the register is available for allocation. Note that access
6194 via inline asm must use constraints. Add note about async-signal
6195 handlers. Remove paragraph about automagic register selection.
6197 2018-05-23 Richard Biener <rguenther@suse.de>
6199 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
6200 of fixed offset from memset VN.
6202 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
6204 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
6206 (alloc_cand_and_find_basis): Initialize first_interp field.
6207 (slsr_process_mul): Modify first_interp field.
6208 (slsr_process_add): Likewise.
6209 (slsr_process_cast): Modify first_interp field for each new
6211 (slsr_process_copy): Likewise.
6212 (dump_candidate): Dump first_interp field.
6213 (replace_mult_candidate): Process all interpretations, not just
6215 (replace_rhs_if_not_dup): Likewise.
6216 (replace_one_candidate): Likewise.
6218 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
6220 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
6222 (aarch64_needs_frame_chain): New function.
6223 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
6225 2018-05-23 Sudakshina Das <sudi.das@arm.com>
6228 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
6229 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
6230 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
6231 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
6232 as true for strict-align.
6233 (aarch64_can_inline_p): Perform checks even when callee has no
6234 attributes to check for strict alignment.
6235 * doc/extend.texi (AArch64 Function Attributes): Document
6237 * doc/invoke.texi: (AArch64 Options): Likewise.
6239 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
6241 PR tree-optimization/85853
6242 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
6243 the handling of the root of the node to...
6244 (vect_slp_analyze_node_operations_1): ...this new function,
6245 and run the whole thing with the child nodes' def types
6246 set according to their SLP node's def type.
6248 2018-05-23 Richard Biener <rguenther@suse.de>
6251 * tree-data-ref.c (create_runtime_alias_checks): Defer
6252 and ignore overflow warnings.
6254 2018-05-23 Yury Gribov <tetra2005@gmail.com>
6256 PR tree-optimization/85822
6257 * tree-vrp.c (is_masked_range_test): Fix handling of negative
6260 2018-05-23 Richard Biener <rguenther@suse.de>
6262 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
6263 memset constants via native_interpret_expr.
6265 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
6268 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
6270 (cgraph_node::create_alias): Likewise.
6271 (cgraph_node::get_availability): Check ifunc_resolver instead
6272 of looking up ifunc attribute.
6273 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
6274 * varasm.c (do_assemble_alias): Likewise.
6275 (assemble_alias): Likewise.
6276 (default_binds_local_p_3): Likewise.
6277 * cgraph.h (cgraph_node): Add ifunc_resolver.
6278 (cgraph_node::only_called_directly_or_aliased_p): Return false
6280 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
6282 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
6283 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
6284 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
6285 instead of looking up ifunc attribute.
6287 2018-05-22 Luis Machado <luis.machado@linaro.org>
6289 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
6291 2018-05-22 Martin Sebor <msebor@redhat.com>
6294 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
6295 only when expasion succeeds.
6296 (expand_builtin_strcmp): Same.
6297 (expand_builtin_strncmp): Same.
6299 2018-05-22 Martin Sebor <msebor@redhat.com>
6301 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
6303 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
6304 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6306 * config/aarch64/aarch64-ldpstp.md: Replace uses of
6307 aarch64_mem_pair_operand with memory_operand and delete operand swapping
6309 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
6310 Add check for legitimate_address.
6311 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
6312 (aarch64_swap_ldrstr_operands): New.
6313 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
6316 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
6317 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6319 * config/aarch64/aarch64.md: New patterns to generate stp
6321 (store_pair_sw, store_pair_dw): New patterns to generate stp for
6322 single words and double words.
6323 (load_pair_sw, load_pair_dw): Likewise.
6324 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
6326 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
6328 * config/aarch64/aarch64-ldpstp.md: Modify peephole
6329 for different mode ldpstp and add peephole for merged zero stores.
6331 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
6333 (aarch64_gen_store_pair): Rename calls to match new patterns.
6334 (aarch64_gen_load_pair): Rename calls to match new patterns.
6335 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
6336 (load_pair<DREG:mode><DREG2:mode>): ... This.
6337 (store_pair<mode>): Rename to...
6338 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
6339 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
6341 (V_INT_EQUIV): Handle SImode.
6342 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
6345 2018-05-22 Martin Sebor <msebor@redhat.com>
6348 * calls.c (maybe_warn_nonstring_arg): Use string length to set
6349 or ajust the presumed bound on an operation to avoid unnecessary
6352 2018-05-22 Martin Sebor <msebor@redhat.com>
6354 PR tree-optimization/85826
6355 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
6356 assuming that a DECL necesarily has a constant size.
6358 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
6361 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
6363 2018-05-22 Richard Biener <rguenther@suse.de>
6365 PR tree-optimization/85834
6366 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
6367 non-constant and non-zero memset arguments.
6369 2018-05-22 Martin Liska <mliska@suse.cz>
6372 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
6374 2018-05-22 Richard Biener <rguenther@suse.de>
6376 PR tree-optimization/85863
6377 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
6378 comparisons when vectype is specified.
6379 (vectorizable_condition): Do not specify vectype for
6380 vect_is_simple_cond when SLP vectorizing.
6382 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
6385 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
6386 define __ibm128 as long double.
6387 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
6388 as a distinct type when IEEE 128-bit support is enabled.
6389 (init_float128_ieee): Fix up conversions between IFmode and IEEE
6390 128-bit types to use the correct functions.
6391 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
6392 convert between 128-bit floating point types that have different
6393 modes but the same representation, instead of using gen_lowpart to
6395 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
6397 (IFKF_reg): New attributes to give the register constraints for
6399 (extend<mode>tf2_internal): New insns to mark an explicit
6400 conversion between 128-bit floating point types that have a
6401 different mode but share the same representation.
6403 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
6405 PR tree-optimization/85814
6406 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
6407 a null return from get_strinfo when unsharing the next
6408 strinfo in the chain.
6410 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
6413 * varasm.c (weak_finish): Clean up weak_decls.
6415 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6417 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
6418 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
6419 UNSPEC_UADALP values.
6420 * config/aarch64/iterators.md (ABAL): New int iterator.
6423 (sur): Add mappings for the above.
6424 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
6426 (aarch64_<sur>abal<mode>_4): Likewise.
6427 (aarch64_<sur>adalp<mode>_3): Likewise.
6428 (<sur>sadv16qi): New define_expand.
6430 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
6432 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
6433 (*movdf_internal): Ditto.
6434 (*rcpsf2_sse): Ditto.
6435 (*rsqrtsf2_sse): Ditto.
6436 (*sqrt<mode>2_sse): Ditto.
6438 2018-05-21 Tamar Christina <tamar.christina@arm.com>
6440 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
6442 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
6443 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
6444 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
6445 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
6447 * config/aarch64/arm_neon.h: Likewise.
6448 * config/aarch64/iterators.md (VQ_I): New.
6450 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
6452 * config.gcc: Add arc/t-multilib-linux to tmake_file for
6454 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
6457 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
6459 * config/nds32/constraints.md (S): New constraint.
6460 * config/nds32/nds32.md (call_internal): Use constraint S.
6461 (call_value_internal): Likewise.
6462 (sibcall_internal): Likewise.
6463 (sibcall_value_internal): Likewise.
6465 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
6466 Chung-Ju Wu <jasonwucj@gmail.com>
6468 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
6471 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
6472 Chung-Ju Wu <jasonwucj@gmail.com>
6474 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
6475 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
6476 (nds32_rtx_costs_impl): Simplify.
6477 (nds32_address_cost_impl): Simplify.
6478 (nds32_init_rtx_costs): New function.
6479 (nds32_rtx_costs_speed_prefer): Likewise.
6480 (nds32_rtx_costs_size_prefer): Likewise.
6481 (nds32_address_cost_speed_prefer): Likewise.
6482 (nds32_address_cost_speed_fwprop): Likewise.
6483 (nds32_address_cost_size_prefer): Likewise.
6484 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
6485 * config/nds32/nds32.c (nds32_option_override): Use
6486 nds32_init_rtx_costs function.
6488 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
6490 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
6491 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
6492 (TARGET_PIPELINE_N8): Likewise.
6493 (TARGET_PIPELINE_N10): Likewise.
6494 (TARGET_PIPELINE_N13): Likewise.
6495 (TARGET_PIPELINE_GRAYWOLF): Likewise.
6497 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
6499 * config/nds32/nds32-fpu.md: Update copyright year.
6501 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
6503 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
6505 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
6507 * config/nds32/nds32.c
6508 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
6509 * config/nds32/nds32.opt (minline-asm-r15): New option.
6511 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
6513 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
6515 * config/nds32/nds32.md (abssi2): New pattern.
6517 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
6519 * config/i386/i386.md (rex64namesuffix): New mode attribute.
6520 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
6521 Merge insn pattern from sse_cvtsi2ss<round_name> and
6522 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
6523 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
6524 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
6525 using SWI48 mode iterator.
6526 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
6527 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
6528 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
6529 pattern from sse_cvttss2si<round_saeonly_name>
6530 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
6531 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
6532 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
6533 using SWI48 mode iterator.
6534 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
6535 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
6536 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
6537 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
6538 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
6539 using SWI48 mode iterator.
6540 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
6541 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
6542 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
6543 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
6544 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
6545 SWI48 mode iterator.
6546 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
6547 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
6548 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
6549 pattern from sse_cvttsd2si<round_saeonly_name>
6550 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
6552 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
6554 * config/nds32/nds32-md-auxiliary.c
6555 (nds32_valid_smw_lwm_base_p): Refine.
6556 (nds32_output_smw_single_word): Refine.
6557 (nds32_output_smw_double_word): New.
6558 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
6560 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
6562 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
6563 (nds32_output_stack_pop): Refine.
6564 (nds32_expand_unaligned_load): Refine.
6565 (nds32_expand_unaligned_store): Refine.
6567 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
6568 Chung-Ju Wu <jasonwucj@gmail.com>
6570 * config/nds32/constants.md: Add TP_REGNUM constant.
6571 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
6572 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
6574 * config/nds32/nds32-doubleword.md: Consider flag_pic.
6575 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
6576 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
6577 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
6578 and PIC code generation.
6579 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
6581 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
6583 * config/nds32/nds32.md: Support TLS and PIC.
6584 * config/nds32/nds32.c: Support TLS and PIC.
6585 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
6586 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
6589 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
6591 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
6592 mode with E_ prefix.
6594 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
6595 Chung-Ju Wu <jasonwucj@gmail.com>
6597 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
6598 * config/nds32/nds32-md-auxiliary.c
6599 (symbolic_reference_mentioned_p): New.
6600 (nds32_legitimize_ict_address): New.
6601 (nds32_expand_ict_move): New.
6602 (nds32_indirect_call_referenced_p): New.
6603 (nds32_symbol_binds_local_p): Delete.
6604 (nds32_long_call_p): Modify.
6605 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
6606 * config/nds32/nds32-protos.h
6607 (symbolic_reference_mentioned_p): Declare.
6608 (nds32_legitimize_ict_address): Declare.
6609 (nds32_expand_ict_move): Declare.
6610 (nds32_indirect_call_referenced_p): Declare.
6611 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
6612 (nds32_relax_group): Use nds32_ict_const_p as condition.
6613 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
6614 (nds32_asm_file_start): Output ict_model directive in asm code.
6615 (nds32_legitimate_address_p): Consider indirect call.
6616 (nds32_print_operand): Consider indirect call.
6617 (nds32_print_operand_address): Consider indirect call.
6618 (nds32_insert_attributes): Handle "indirect_call" attribute.
6619 (TARGET_LEGITIMATE_ADDRESS_P): Define.
6620 (TARGET_LEGITIMATE_CONSTANT_P): Define.
6621 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
6622 (TARGET_DELEGITIMIZE_ADDRESS): Define.
6623 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
6624 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
6625 (TARGET_ICT_MODEL_SMALL): Define.
6626 (TARGET_ICT_MODEL_LARGE): Define.
6627 * config/nds32/nds32.md (movsi): Consider ict model.
6628 (call, call_value): Consider ict model.
6629 (sibcall, sibcall_value): Consider ict model.
6630 * config/nds32/nds32.opt (mict-model): New option.
6631 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
6634 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
6635 Monk Chiang <sh.chiang04@gmail.com>
6636 Jim Wilson <jimw@sifive.com>
6638 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
6639 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
6640 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
6641 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
6642 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
6643 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
6644 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
6645 compute save_libcall_adjustment properly.
6646 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
6647 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
6648 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
6649 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
6650 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
6651 (ABI_SPEC): Handle mabi=ilp32e.
6652 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
6653 (RVE): Add RVE mask.
6654 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
6655 <-march>: Add rv32e as an example.
6657 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
6660 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
6661 (intra_create_variable_infos): Handle C++ constructors.
6663 2018-05-18 Martin Liska <mliska@suse.cz>
6665 * passes.def: Remove a redundant pass.
6667 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
6670 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
6672 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6674 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
6676 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
6678 (armv2, armv2a, armv3, armv3m): Delete architectures.
6679 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
6680 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
6681 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
6683 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
6684 (*mulsidi3adddi): Likewise.
6685 (mulsidi3): Likewise.
6686 (*mulsidi3_nov6): Likewise.
6687 (umulsidi3): Likewise.
6688 (umulsidi3_nov6): Likewise.
6689 (umaddsidi4): Likewise.
6690 (*umulsidi3adddi): Likewise.
6691 (smulsi3_highpart): Likewise.
6692 (*smulsi3_highpart_nov6): Likewise.
6693 (umulsi3_highpart): Likewise.
6694 (*umulsi3_highpart_nov6): Likewise.
6695 * config/arm/arm.h (arm_arch3m): Delete.
6696 * config/arm/arm.c (arm_arch3m): Delete.
6697 (arm_option_override_internal): Update armv3-related comment.
6698 (arm_configure_build_target): Delete use of isa_bit_mode32.
6699 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
6700 (arm_rtx_costs_internal): Delete check of arm_arch3m.
6701 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
6703 (mulusa3): Likewise.
6704 * config/arm/arm-protos.h (arm_arch3m): Delete.
6705 * config/arm/arm-tables.opt: Regenerate.
6706 * config/arm/arm-tune.md: Likewise.
6707 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
6708 deleted architectures.
6710 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6712 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
6713 (armv5t, armv5te): New features.
6714 (ARMv5, ARMv5e): Delete fgroups.
6715 (ARMv5t, ARMv5te): Adjust for above changes.
6717 (armv5, armv5e): Delete arches.
6718 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
6720 (*call_reg_arm): Likewise.
6721 (*call_value_reg_armv5): Likewise.
6722 (*call_value_reg_arm): Likewise.
6723 (*call_symbol): Likewise.
6724 (*call_value_symbol): Likewise.
6725 (*sibcall_insn): Likewise.
6726 (*sibcall_value_insn): Likewise.
6728 (prefetch): Likewise.
6729 (define_split and define_peephole2 dependent on arm_arch5):
6731 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
6733 (TARGET_ARM_QBIT): Likewise.
6734 (TARGET_DSP_MULTIPLY): Likewise.
6735 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
6736 (arm_arch5, arm_arch5e): Delete.
6737 (arm_arch5t, arm_arch5te): Declare.
6738 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
6739 (arm_arch5t): Declare.
6740 (arm_option_reconfigure_globals): Update for the above.
6741 (arm_options_perform_arch_sanity_checks): Update comment, replace
6742 use of arm_arch5 with arm_arch5t.
6743 (use_return_insn): Likewise.
6744 (arm_emit_call_insn): Likewise.
6745 (output_return_instruction): Likewise.
6746 (arm_final_prescan_insn): Likewise.
6747 (arm_coproc_builtin_available): Likewise.
6748 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
6749 arm_arch5e with arm_arch5t and arm_arch5te.
6750 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
6751 (arm_arch5t, arm_arch5te): Declare.
6752 * config/arm/arm-tables.opt: Regenerate.
6753 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
6754 * config/arm/t-multilib: Likewise.
6755 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
6756 instead of arm_arch5.
6757 (*call_reg_thumb1): Likewise.
6758 (*call_value_reg_thumb1_v5): Likewise.
6759 (*call_value_reg_thumb1): Likewise.
6760 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
6762 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
6764 2018-05-18 Martin Liska <mliska@suse.cz>
6766 PR gcov-profile/84846
6767 * doc/gcov.texi: Document -t option of gcov tool.
6769 2018-05-18 Martin Liska <mliska@suse.cz>
6771 PR gcov-profile/84846
6772 * gcov.c (print_usage): Add new -t option.
6773 (process_args): Handle the option.
6774 (generate_results): Use stdout as output when requested by
6777 2018-05-18 Martin Liska <mliska@suse.cz>
6779 PR gcov-profile/84846
6780 * coverage.c (coverage_init): Write PWD to .gcno file.
6781 * doc/gcov.texi: Document how working directory is printed.
6782 * gcov-dump.c (dump_gcov_file): Print PWD.
6783 * gcov.c (output_intermediate_file): Likewise.
6784 (read_graph_file): Read PWD string.
6785 (output_lines): Print PWD.
6787 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6790 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
6791 for retval and return false if all args to phi are zero.
6793 2018-05-18 Richard Biener <rguenther@suse.de>
6795 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
6797 (evrp_dom_walker::before_dom_children): Call it.
6799 2018-05-18 Richard Biener <rguenther@suse.de>
6801 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
6802 results when processing array refs with variable index.
6804 2018-05-18 Toon Moene <toon@moene.org>
6806 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
6807 directly after that of -floop-interchange. Indicate that both
6808 options are enabled by default when specifying -O3.
6810 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6812 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
6813 iterator. Delete separate integer-mode vec_set<mode> expander.
6814 (aarch64_simd_vec_setv2di): Delete.
6815 (vec_setv2di): Delete.
6816 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
6817 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
6818 the "w, r" alternative.
6820 2018-05-18 Martin Liska <mliska@suse.cz>
6822 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
6823 * tree-pass.h (make_pass_lower_switch_O0): New function.
6824 * tree-switch-conversion.c (node_has_low_bound): Remove.
6825 (node_has_high_bound): Likewise.
6826 (node_is_bounded): Likewise.
6827 (class pass_lower_switch): Make it a template type and create
6829 (pass_lower_switch::execute): Add template argument.
6830 (make_pass_lower_switch): New function.
6831 (make_pass_lower_switch_O0): New function.
6832 (do_jump_if_equal): Remove.
6833 (emit_case_nodes): Simplify to just handle all 3 cases and leave
6834 all the hard work to tree optimization passes.
6836 2018-05-18 Martin Liska <mliska@suse.cz>
6838 * dbgcnt.c (limit_low): Renamed from limit.
6839 (limit_high): New variable.
6840 (dbg_cnt_is_enabled): Check for upper limit.
6841 (dbg_cnt): Adjust dumping.
6842 (dbg_cnt_set_limit_by_index): Add new argument for high
6844 (dbg_cnt_set_limit_by_name): Likewise.
6845 (dbg_cnt_process_single_pair): Parse new format.
6846 (dbg_cnt_process_opt): Use strtok.
6847 (dbg_cnt_list_all_counters): Remove 'value' and add
6849 * doc/invoke.texi: Document changes.
6851 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
6853 * doc/sourcebuild.texi (scalar_all_fma): Document.
6854 * tree.def (FMA_EXPR): Delete.
6855 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
6856 * internal-fn.c (ternary_direct): New macro.
6857 (expand_ternary_optab_fn): Likewise.
6858 (direct_ternary_optab_supported_p): Likewise.
6859 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
6860 * builtins.c (fold_builtin_fma): Delete.
6861 (fold_builtin_3): Don't call it.
6862 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
6863 * expr.c (expand_expr_real_2): Likewise.
6864 * fold-const.c (operand_equal_p): Likewise.
6865 (fold_ternary_loc): Likewise.
6866 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
6867 * gimple.c (DEFTREECODE): Likewise.
6868 * gimplify.c (gimplify_expr): Likewise.
6869 * optabs-tree.c (optab_for_tree_code): Likewise.
6870 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
6871 * tree-eh.c (operation_could_trap_p): Likewise.
6872 (stmt_could_throw_1_p): Likewise.
6873 * tree-inline.c (estimate_operator_cost): Likewise.
6874 * tree-pretty-print.c (dump_generic_node): Likewise.
6875 (op_code_prio): Likewise.
6876 * tree-ssa-loop-im.c (stmt_cost): Likewise.
6877 * tree-ssa-operands.c (get_expr_operands): Likewise.
6878 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
6879 * fold-const-call.h (fold_fma): Delete.
6880 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
6881 CFN_FNMA and CFN_FNMS.
6883 * genmatch.c (combined_fn): New enum.
6884 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
6885 (commutative_op): New function.
6886 (commutate): Use it. Handle more than 2 operands.
6887 (dt_operand::gen_gimple_expr): Use commutative_op.
6888 (parser::parse_expr): Allow :c to be used with non-binary
6889 operators if the commutative operand is known.
6890 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
6891 CFN_FMS, CFN_FNMA and CFN_FNMS.
6892 (backprop::process_assign_use): Remove FMA_EXPR handling.
6893 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
6894 (gen_hsa_fma): New function.
6895 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
6896 IFN_FNMA and IFN_FNMS.
6897 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
6898 * gimple-fold.h (follow_all_ssa_edges): Declare.
6899 * gimple-fold.c (follow_all_ssa_edges): New function.
6900 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
6901 gimple_build interface and use follow_all_ssa_edges to fold the result.
6902 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
6903 instead of checking for optabs directly.
6904 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
6905 rather than FMA_EXPRs.
6906 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
6907 call to IFN_FMA instead of an FMA_EXPR.
6909 2018-05-17 Jim Wilson <jimw@sifive.com>
6911 * expr.c (do_tablejump): When converting index to Pmode, if we have a
6912 sign extended promoted subreg, and the range does not have the sign bit
6913 set, then do a sign extend.
6915 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
6916 test, check for sign extended subreg and/or constant operands, and
6917 do a sign extend in that case.
6919 2018-05-17 Steve Ellcey <sellcey@cavium.com>
6921 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
6922 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
6924 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
6925 Change logics_shift_reg to logics_shift_imm.
6926 (thunderx2t99_fp_loadpair_basic): Delete.
6927 (thunderx2t99_fp_storepair_basic): Delete.
6928 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
6929 (thunderx2t99_asimd_polynomial): Delete.
6930 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
6931 and neon_fp_mul_d_scalar_q.
6932 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
6933 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
6934 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
6935 (thunderx2t99_asimd_lut): Add missing tbl types.
6936 (thunderx2t99_asimd_ext): Delete.
6937 (thunderx2t99_asimd_load1_1_mult): Delete.
6938 (thunderx2t99_asimd_load1_2_mult): Delete.
6939 (thunderx2t99_asimd_load1_ldp): New.
6940 (thunderx2t99_asimd_load1): New.
6941 (thunderx2t99_asimd_load2): Add missing *load2* types.
6942 (thunderx2t99_asimd_load3): New.
6943 (thunderx2t99_asimd_load4): New.
6944 (thunderx2t99_asimd_store1_1_mult): Delete.
6945 (thunderx2t99_asimd_store1_2_mult): Delete.
6946 (thunderx2t99_asimd_store2_mult): Delete.
6947 (thunderx2t99_asimd_store2_onelane): Delete.
6948 (thunderx2t99_asimd_store_stp): New.
6949 (thunderx2t99_asimd_store1): New.
6950 (thunderx2t99_asimd_store2): New.
6951 (thunderx2t99_asimd_store3): New.
6952 (thunderx2t99_asimd_store4): New.
6954 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
6956 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
6957 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
6959 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
6960 Segher Boessenkool <segher@kernel.crashing.org>
6963 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
6966 2018-05-17 Richard Biener <rguenther@suse.de>
6968 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
6969 for pruning loop and prune defs feeding only already visited PHIs.
6971 2018-05-17 Richard Biener <rguenther@suse.de>
6973 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
6975 2018-05-17 Bin Cheng <bin.cheng@arm.com>
6976 Richard Biener <rguenther@suse.de>
6978 PR tree-optimization/85793
6979 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
6980 for VMAT_ELEMENTWISE.
6982 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
6984 * internal-fn.h (lookup_internal_fn): Declare
6985 * internal-fn.c (lookup_internal_fn): New function.
6986 * gimple.c (gimple_build_call_from_tree): Handle calls to
6988 * gimple-pretty-print.c (dump_gimple_call): Print "." before
6989 internal function names.
6990 * tree-pretty-print.c (dump_generic_node): Likewise.
6991 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
6993 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
6995 * gimple-fold.h (gimple_build): Make the function forms take
6996 combined_fn rather than built_in_function.
6997 (gimple_simplify): Likewise.
6998 * gimple-match-head.c (gimple_simplify): Likewise.
6999 * gimple-fold.c (gimple_build): Likewise.
7000 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
7001 rather than gimple_build_call_internal.
7002 (get_initial_defs_for_reduction): Likewise.
7003 (vect_create_epilog_for_reduction): Likewise.
7004 (vectorizable_live_operation): Likewise.
7006 2018-05-17 Martin Liska <mliska@suse.cz>
7008 * gimple-ssa-sprintf.c (format_directive): Do not use
7009 space in between 'G_' and '('.
7011 2018-05-17 Jakub Jelinek <jakub@redhat.com>
7014 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
7015 even if the mask is not all ones.
7018 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
7020 (ix86_gimple_fold_builtin): Likewise.
7023 * config/i386/i386.c: Include tree-vector-builder.h.
7024 (ix86_vector_shift_count): New function.
7025 (ix86_fold_builtin): Fold shift builtins by scalar count.
7026 (ix86_gimple_fold_builtin): Likewise.
7028 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
7029 _mm512_setzero): New intrinsics.
7031 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
7032 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7034 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
7035 code generation for cases where splatting a value is not useful.
7036 * simplify-rtx.c (simplify_ternary_operation): Simplify
7037 vec_merge across a vec_duplicate and a paradoxical subreg forming
7038 a vector mode to a vec_concat.
7040 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
7042 * config.gcc: Support "goldmont-plus".
7043 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
7045 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
7046 PROCESSOR_GOLDMONT_PLUS.
7047 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
7048 (processor_target_table): Add "goldmont-plus".
7049 (PTA_GOLDMONT_PLUS): Define.
7050 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
7051 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
7052 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
7053 (fold_builtin_cpu): Add "goldmont-plus".
7054 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
7055 (ix86_option_override_internal): Add "goldmont-plus".
7056 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
7057 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
7058 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
7059 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
7061 2018-05-17 Richard Biener <rguenther@suse.de>
7063 PR tree-optimization/85757
7064 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
7065 remove defs that only feed that PHI from further processing.
7067 2018-05-16 Jim Wilson <jimw@sifive.com>
7069 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
7071 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
7073 2018-05-16 Mark Wielaard <mark@klomp.org>
7075 * dwarf2out.c (count_index_strings): New function.
7076 (output_indirect_strings): Call count_index_strings and generate
7077 header for dwarf_version >= 5.
7079 2018-05-16 Mark Wielaard <mark@klomp.org>
7081 * dwarf2out.c (dwarf_FORM): New function.
7082 (set_indirect_string): Use dwarf_FORM.
7083 (reset_indirect_string): Likewise.
7084 (size_of_die): Likewise.
7085 (value_format): Likewise.
7086 (output_die): Likewise.
7087 (add_skeleton_AT_string): Likewise.
7088 (output_macinfo_op): Likewise.
7089 (index_string): Likewise.
7090 (output_index_string_offset): Likewise.
7091 (output_index_string): Likewise.
7092 (count_index_strings): Likewise.
7094 2018-05-16 Carl Love <cel@us.ibm.com>
7096 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
7097 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
7099 2018-05-16 Martin Jambor <mjambor@suse.cz>
7101 * ipa-prop.c (ipa_free_all_edge_args): Remove.
7102 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
7104 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
7106 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
7107 (fnma<mode>4): Likewise.
7108 (fms<mode>4): Likewise.
7109 (fnms<mode>4): Likewise.
7110 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
7111 (aarch64_fnma<mode>4): Likewise.
7112 (aarch64_fms<mode>4): Likewise.
7113 (aarch64_fnms<mode>4): Likewise.
7114 (aarch64_fnmadd<mode>4): Likewise.
7116 2018-05-16 Jason Merrill <jason@redhat.com>
7118 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
7120 2018-05-16 Richard Biener <rguenther@suse.de>
7122 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
7123 (dump_stmt_cost): Declare.
7124 (add_stmt_cost): Dump cost we add.
7125 (add_stmt_costs): New function.
7126 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
7128 (vect_analyze_stmt): Adjust prototype.
7129 (vectorizable_condition): Likewise.
7130 (vectorizable_live_operation): Likewise.
7131 (vectorizable_reduction): Likewise.
7132 (vectorizable_induction): Likewise.
7133 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
7134 cost vector to pass to vectorizable_ and record afterwards.
7135 (vect_model_reduction_cost): Take cost vector argument and adjust.
7136 (vect_model_induction_cost): Likewise.
7137 (vectorizable_reduction): Likewise.
7138 (vectorizable_induction): Likewise.
7139 (vectorizable_live_operation): Likewise.
7140 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
7141 SLP_TREE_NUMBER_OF_VEC_STMTS.
7142 (vect_analyze_slp_cost_1): Remove.
7143 (vect_analyze_slp_cost): Likewise.
7144 (vect_slp_analyze_node_operations): Take visited args and
7145 a target cost vector. Avoid processing already visited stmt sets.
7146 (vect_slp_analyze_operations): Use a local cost vector to gather
7147 costs and register those of non-discarded instances.
7148 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
7149 (vect_schedule_slp_instance): Remove copying of
7150 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
7152 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
7153 adding cost. Record cost entry location.
7154 (vect_prologue_cost_for_slp_op): Function to compute cost of
7155 a constant or invariant generated for SLP vect in the prologue,
7156 split out from vect_analyze_slp_cost_1.
7157 (vect_model_simple_cost): Make static. Adjust for SLP costing.
7158 (vect_model_promotion_demotion_cost): Likewise.
7159 (vect_model_store_cost): Likewise, make static.
7160 (vect_model_load_cost): Likewise.
7161 (vectorizable_bswap): Add cost vector arg and adjust.
7162 (vectorizable_call): Likewise.
7163 (vectorizable_simd_clone_call): Likewise.
7164 (vectorizable_conversion): Likewise.
7165 (vectorizable_assignment): Likewise.
7166 (vectorizable_shift): Likewise.
7167 (vectorizable_operation): Likewise.
7168 (vectorizable_store): Likewise.
7169 (vectorizable_load): Likewise.
7170 (vectorizable_condition): Likewise.
7171 (vectorizable_comparison): Likewise.
7172 (can_vectorize_live_stmts): Likewise.
7173 (vect_analyze_stmt): Likewise.
7174 (vect_transform_stmt): Adjust calls to vectorizable_*.
7175 * tree-vectorizer.c: Include gimple-pretty-print.h.
7176 (dump_stmt_cost): New function.
7178 2018-05-16 Richard Biener <rguenther@suse.de>
7180 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
7181 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
7182 * tree-ssa-dse.c: Include tree-ssa-loop.h.
7183 (check_name): New callback.
7184 (dse_classify_store): Track cycles via a visited bitmap of PHI
7185 defs and simplify handling of in-loop and across loop dead stores
7186 and properly fail for loop-variant refs. Handle byte-tracking with
7187 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
7190 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
7192 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
7193 (vect_get_mask_type_for_stmt): Likewise.
7194 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
7196 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
7197 to determine the statement's vector type and the vector type that
7198 should be used for calculating nunits. Deal with cases in which
7199 the type has to be deferred.
7200 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
7201 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
7202 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
7203 (vect_determine_vf_for_stmt): New functions, split out from...
7204 (vect_determine_vectorization_factor): ...here.
7205 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
7206 (vect_get_mask_type_for_stmt): New functions, split out from
7207 vect_determine_vectorization_factor.
7209 2018-05-16 Richard Biener <rguenther@suse.de>
7211 * tree-cfg.c (verify_gimple_assign_ternary): Properly
7212 verify the [VEC_]COND_EXPR embedded comparison.
7214 2018-05-15 Martin Sebor <msebor@redhat.com>
7216 PR tree-optimization/85753
7217 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
7218 RECORD_TYPE in addition to ARRAY_TYPE.
7220 2018-05-15 Martin Sebor <msebor@redhat.com>
7223 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
7225 2018-05-15 Richard Biener <rguenther@suse.de>
7227 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
7228 add by_clobber_p one. Change algorithm to collect all defs
7229 representing uses we need to walk and try reducing them to
7230 a single one before failing.
7231 (dse_dom_walker::dse_optimize_stmt): Adjust.
7233 2018-05-13 Mark Wielaard <mark@klomp.org>
7235 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
7236 (size_of_loc_descr): Likewise.
7237 (output_loc_operands): Likewise.
7238 (output_loc_operands_raw): Likewise.
7239 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
7240 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
7241 (hash_loc_operands): Likewise.
7242 (compare_loc_operands): Likewise.
7244 2018-05-14 Mark Wielaard <mark@klomp.org>
7246 * dwarf2out.c (count_index_addrs): New function.
7247 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
7249 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7251 PR tree-optimization/83648
7252 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
7253 return value as malloc candidate.
7255 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7258 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
7259 param as true in call to suggest_attribute.
7261 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
7263 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
7264 -mreadonly-in-sdata.
7266 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7268 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
7270 (aarch64_crypto_aesd_fused): Likewise.
7272 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
7274 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
7275 (movsi_aarch64): Likewise.
7276 (load_pairsi): Likewise.
7277 (load_pairdi): Likewise.
7278 (store_pairsi): Likewise.
7279 (store_pairdi): Likewise.
7280 (load_pairsf): Likewise.
7281 (load_pairdf): Likewise.
7282 (store_pairsf): Likewise.
7283 (store_pairdf): Likewise.
7284 (zero_extend): Likewise.
7285 (trunc): Swap alternatives.
7286 (fcvt_target): Add '?' to prefer w over r.
7288 2018-05-14 Jakub Jelinek <jakub@redhat.com>
7291 * config/i386/i386.md: Disallow non-commutative arithmetics in
7292 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
7293 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
7294 in the peephole2 before it.
7296 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
7298 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
7299 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
7300 (ix86_handle_option): Handle -mcldemote.
7301 * config.gcc: New header.
7302 * config/i386/cldemoteintrin.h: New file.
7303 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
7304 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
7306 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
7307 OPTION_MASK_ISA_CLDEMOTE.
7308 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
7309 (ix86_valid_target_attribute_inner_p): Ditto.
7310 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
7311 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
7312 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
7313 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
7314 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
7316 * config/i386/i386.opt: Add -mcldemote.
7317 * config/i386/x86intrin.h: New header.
7318 * doc/invoke.texi: Add -mcldemote.
7320 2018-05-14 Richard Biener <rguenther@suse.de>
7322 * doc/match-and-simplify.texi: Adjust :s documentation.
7324 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
7326 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
7327 intended memcpy size.
7328 (REORDER_45): Likewise.
7330 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
7332 * sort.cc: New file.
7333 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
7334 * vec.c (qsort_chk): Use gcc_qsort.
7335 * Makefile.in (OBJS-libcommon): Add sort.o.
7336 (build/sort.o): New target. Use it...
7337 (BUILD_RTL): ... here, and...
7338 (build/gencfn-macros): ... here, and...
7339 (build/genmatch): ... here.
7341 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
7342 Chung-Ju Wu <jasonwucj@gmail.com>
7344 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
7345 * config/nds32/nds32-graywolf.md: New file.
7346 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
7347 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
7349 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
7350 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
7351 * config/nds32/nds32.md (pipeline_model): Add graywolf.
7352 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
7353 * config/nds32/pipelines.md: Include n15 settings.
7355 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
7356 Chung-Ju Wu <jasonwucj@gmail.com>
7358 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
7359 * config/nds32/nds32-n13.md: New file.
7360 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
7361 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
7363 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
7364 * config/nds32/nds32.md (pipeline_model): Add n13.
7365 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
7366 * config/nds32/pipelines.md: Include n13 settings.
7368 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
7369 Chung-Ju Wu <jasonwucj@gmail.com>
7371 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
7372 * config/nds32/nds32-n10.md: New file.
7373 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
7374 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
7376 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
7377 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
7378 * config/nds32/nds32.md (pipeline_model): Add n10.
7379 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
7380 * config/nds32/pipelines.md: Include n10 settings.
7382 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
7383 Kito Cheng <kito.cheng@gmail.com>
7384 Chung-Ju Wu <jasonwucj@gmail.com>
7386 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
7387 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
7388 Add enum values for DSP extension instructions.
7389 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
7391 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
7392 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
7394 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
7395 * config/nds32/nds32-dspext.md: New file for DSP implementation.
7396 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
7397 * config/nds32/nds32-intrinsic.md: Likewise.
7398 * config/nds32/nds32_intrinsic.h: Likewise.
7399 * config/nds32/nds32-md-auxiliary.c: Likewise.
7400 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
7401 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
7402 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
7403 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
7404 * config/nds32/nds32-protos.h: New declarations for DSP extension.
7405 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
7406 TYPE_DMAC in switch statement.
7407 * config/nds32/nds32.c: New checking and implementation for DSP
7408 extension instructions.
7409 * config/nds32/nds32.h: Likewise.
7410 * config/nds32/nds32.md: Likewise.
7411 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
7412 * config/nds32/predicates.md: Implement new predicates for DSP
7415 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
7417 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
7418 Reformat alternatives and attributes so it is easier to identify
7419 which constraints/attributes go with which instruction.
7420 (mov<mode>_hardfloat32, FMOVE64): Likewise.
7421 (mov<mode>_softfloat32, FMOVE64): Likewise.
7422 (mov<mode>_hardfloat64, FMOVE64): Likewise.
7423 (mov<mode>_softfloat64, FMOVE64): Likewise.
7425 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
7427 * doc/extend.texi (PowerPC Built-in Functions): Rename this
7429 (Basic PowerPC Built-in Functions): The new name of the
7430 subsection previously known as "PowerPC Built-in Functions".
7431 (Basic PowerPC Built-in Functions Available on all Configurations):
7433 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
7434 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
7435 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
7436 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
7438 2018-05-11 Martin Jambor <mjambor@suse.cz>
7441 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
7444 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
7447 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
7449 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
7451 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
7452 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
7453 (ix86_handle_option): Handle -mwaitpkg.
7454 * config.gcc: New header.
7455 * config/i386/cpuid.h (bit_WAITPKG): New bit.
7456 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
7457 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
7459 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
7460 OPTION_MASK_ISA_WAITPKG.
7461 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
7462 (ix86_option_override_internal): Add PTA_WAITPKG.
7463 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
7464 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
7465 IX86_BUILTIN_TPAUSE.
7466 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
7467 __builtin_ia32_umwait and __builtin_ia32_tpause.
7468 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
7469 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
7470 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
7471 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
7472 UNSPECV_TPAUSE): New.
7473 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
7474 * config/i386/i386.opt: Add -mwaitpkg.
7475 * config/i386/waitpkgintrin.h: New file.
7476 * config/i386/x86intrin.h: New header.
7477 * doc/invoke.texi: Add -mwaitpkg.
7479 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
7482 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
7484 (cortex-m0): Use armv6s-m isa.
7485 (cortex-m0plus): Likewise.
7486 (cortex-m1): Likewise.
7487 (cortex-m0.small-multiply): Likewise.
7488 (cortex-m0plus.small-multiply): Likewise.
7489 (cortex-m1.small-multiply): Likewise.
7491 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
7492 Jakub Jelinek <jakub@redhat.com>
7494 PR tree-optimization/85692
7495 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
7496 source permute as well.
7498 2018-05-11 Martin Liska <mliska@suse.cz>
7501 * doc/extend.texi: Document LLVM style format for no_sanitize
7504 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
7506 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
7507 mode_supports_vsx_dform_quad to mode_supports_dq_form.
7508 (mode_supports_vsx_dform_quad): Likewise.
7509 (mode_supports_vmx_dform): Move these functions to be next to the
7510 other mode_supports functions.
7511 (mode_supports_dq_form): Likewise.
7512 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
7513 mode_supports_dq_form.
7514 (reg_offset_addressing_ok_p): Likewise.
7515 (offsettable_ok_by_alignment): Likewise.
7516 (rs6000_legitimate_offset_address_p): Likewise.
7517 (legitimate_lo_sum_address_p): Likewise.
7518 (rs6000_legitimize_address): Likewise.
7519 (rs6000_legitimize_reload_address): Likewise.
7520 (rs6000_secondary_reload_inner): Likewise.
7521 (rs6000_preferred_reload_class): Likewise.
7522 (rs6000_output_move_128bit): Likewise.
7524 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
7526 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
7527 Generate SImode target register for null target.
7528 <case IX86_BUILTIN_XGETBV>: Ditto.
7529 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
7530 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
7532 2018-05-10 Carl Love <cel@us.ibm.com>
7534 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
7535 dcbtt and dcbtstt if operands[2] is 0.
7537 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
7540 * config/i386/sse.md (usadv64qi): New expander.
7542 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
7544 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
7545 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
7546 -maltivec=be support.
7547 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
7548 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
7549 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
7550 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
7551 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
7552 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
7553 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
7554 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
7555 altivec_vsumsws): Adjust.
7556 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
7557 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
7558 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
7560 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
7561 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
7562 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
7563 (altivec_lve<VI_char>x): Delete expand.
7564 (*altivec_lve<VI_char>x_internal): Rename to...
7565 (altivec_lve<VI_char>x): ... this.
7566 (altivec_lvxl_<mode>): Delete expand.
7567 (*altivec_lvxl_<mode>_internal): Rename to ...
7568 (altivec_lvxl_<mode>): ... this.
7569 (altivec_stvxl_<mode>): Delete expand.
7570 (*altivec_stvxl_<mode>_internal): Rename to ...
7571 (altivec_stvxl_<mode>): ... this.
7572 (altivec_stve<VI_char>x): Delete expand.
7573 (*altivec_stve<VI_char>x_internal): Rename to ...
7574 (altivec_stve<VI_char>x): ... this.
7575 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
7576 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
7577 reduc_plus_scal_<mode>): Adjust.
7578 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
7580 (rs6000_cpu_cpp_builtins): Adjust.
7581 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
7582 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
7583 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
7584 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
7585 -maltivec=be support.
7586 (rs6000_split_vec_extract_var): Adjust.
7587 (rs6000_split_v4si_init): Adjust.
7588 (swap_selector_for_mode): Delete.
7589 (altivec_expand_lvx_be, altivec_expand_stvx_be,
7590 altivec_expand_stvex_be): Delete.
7591 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
7592 -maltivec=be support.
7593 (rs6000_gimple_fold_builtin): Ditto.
7594 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
7596 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
7597 (TARGET_DIRECT_MOVE_64BIT): Adjust.
7598 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
7599 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
7600 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
7601 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
7602 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
7603 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
7604 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
7605 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
7606 anonymous split): Adjust.
7607 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
7608 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
7610 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
7612 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
7613 when --with-gxx-include-dir is also specified.
7614 * configure: Regenerate.
7616 2018-05-09 Jim Wilson <jimw@sifive.com>
7619 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
7620 * config/riscv/t-withmultilib: New.
7621 * config/riscv/withmultilib.h: New.
7622 * doc/install.texi: Document RISC-V --with-multilib-list support.
7624 2018-05-09 Richard Biener <rguenther@suse.de>
7626 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
7628 (vect_bb_vectorization_profitable_p): Adjust. Compute
7629 actual scalar cost using the cost vector and the add_stmt_cost
7632 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
7634 PR rtl-optimization/85645
7635 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
7636 in the REG_CFA_REGISTER note for LR, don't leave it empty.
7638 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
7640 PR rtl-optimization/85645
7641 * shrink-wrap.c (spread_components): Return a boolean saying if
7642 anything was changed.
7643 (try_shrink_wrapping_separate): Iterate spread_components until
7644 nothing changes anymore.
7646 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
7648 PR rtl-optimization/85645
7649 * regrename.c (build_def_use): Also kill the chains that include the
7650 destination of a REG_CFA_REGISTER note.
7652 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
7654 PR rtl-optimization/85645
7655 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
7656 insn that has a REG_CFA_REGISTER note.
7658 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
7660 * cfgexpand.c (expand_clobber): New function.
7661 (expand_gimple_stmt_1): Use it.
7662 * tree-vect-stmts.c (vect_clobber_variable): New function,
7664 (vectorizable_simd_clone_call): ...here.
7665 (vectorizable_store): Emit a clobber either side of an
7666 IFN_STORE_LANES sequence.
7667 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
7669 2018-05-09 Tom de Vries <tom@codesourcery.com>
7672 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
7673 (define_insn "trap_if_false"): Add exit after trap.
7675 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
7677 PR rtl-optimization/85638
7678 * bb-reorder.c: Include common/common-target.h.
7679 (create_forwarder_block): New function extracted from...
7680 (fix_up_crossing_landing_pad): ...here. Rename into...
7681 (dw2_fix_up_crossing_landing_pad): ...this.
7682 (sjlj_fix_up_crossing_landing_pad): New function.
7683 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
7684 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
7685 from both partitions and exit the loop after one iteration.
7687 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
7690 * doc/extend.texi (PowerPC Built-in Functions): Rename this
7692 (Basic PowerPC Built-in Functions): The new name of the
7693 subsection previously known as "PowerPC Built-in Functions".
7694 (Basic PowerPC Built-in Functions Available on all Configurations):
7696 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
7698 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
7700 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
7702 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
7705 2018-05-08 Jim Wilson <jimw@sifive.com>
7707 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
7708 (LD_EMUL_SUFFIX): New.
7709 (LINK_SPEC): Use it.
7711 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
7713 * doc/extend.texi (PowerPC Built-in Functions): Rename this
7715 (Basic PowerPC Built-in Functions): The new name of the
7716 subsection previously known as "PowerPC Built-in Functions".
7717 (Basic PowerPC Built-in Functions Available on all Configurations):
7719 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
7721 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
7723 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
7725 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
7728 2018-05-08 Jakub Jelinek <jakub@redhat.com>
7731 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
7732 after cmpelim optimization.
7734 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
7736 * config.gcc: Support "goldmont".
7737 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
7738 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
7740 * config/i386/i386.c (m_GOLDMONT): Define.
7741 (processor_target_table): Add "goldmont".
7742 (PTA_GOLDMONT): Define.
7743 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
7744 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
7745 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
7746 (fold_builtin_cpu): Add "goldmont".
7747 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
7748 (ix86_option_override_internal): Add "goldmont".
7749 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
7750 (processor_type): Add PROCESSOR_GOLDMONT.
7751 * config/i386/i386.md: Add CPU "glm".
7752 * config/i386/glm.md: New file.
7753 * config/i386/x86-tune.def: Add m_GOLDMONT.
7754 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
7756 2018-05-08 Jakub Jelinek <jakub@redhat.com>
7759 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
7761 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
7762 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
7763 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
7766 * config/i386/i386.c (ix86_fold_builtin): Handle
7767 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
7770 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
7771 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
7773 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
7776 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
7777 (check_arch): Likewise.
7778 (check_fpu): Return the result rather than printing it.
7779 (end arch): Fix operator precedence.
7780 (end cpu): Likewise.
7781 (END): Print the result from check_fpu.
7783 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
7784 Alan Hayward <alan.hayward@arm.com>
7785 David Sherwood <david.sherwood@arm.com>
7787 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
7788 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
7789 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
7790 (*fcmuo<mode>_and): New patterns.
7792 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
7794 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
7795 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
7796 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
7797 (cmp_op, sve_imm_con): New code attributes.
7798 (SVE_COND_INT_CMP, imm_con): Delete.
7799 (cmp_op): Remove above unspecs from int attribute.
7800 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
7802 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
7803 comparison-specific unspecs.
7804 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
7805 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
7806 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
7807 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
7808 (*vec_fcm<cmp_op><mode>): Rename to...
7809 (*fcm<cmp_op><mode>): ...this and adjust likewise.
7810 (*vec_fcmuo<mode>): Rename to...
7811 (*fcmuo<mode>): ...this and adjust likewise.
7812 (*pred_fcm<cmp_op><mode>): New pattern.
7813 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
7814 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
7816 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
7818 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
7819 (aarch64_emit_sve_predicated_cond): New function.
7820 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
7821 (aarch64_emit_unspec_cond_or): Replace with...
7822 (aarch64_emit_sve_or_conds): ...this new function. Use
7823 aarch64_emit_sve_ptrue_op for the individual comparisons and
7824 aarch64_emit_binop to OR them together.
7825 (aarch64_emit_inverted_unspec_cond): Replace with...
7826 (aarch64_emit_sve_inverted_cond): ...this new function. Use
7827 aarch64_emit_sve_ptrue_op for the comparison and
7828 aarch64_emit_unop to invert the result.
7829 (aarch64_expand_sve_vec_cmp_float): Update after the above
7830 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
7832 2018-05-07 Nathan Sidwell <nathan@acm.org>
7834 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
7835 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
7836 (Backwards Compatibility): Likewise.
7838 2018-05-07 Luis Machado <luis.machado@linaro.org>
7842 2018-05-07 Luis Machado <luis.machado@linaro.org>
7844 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
7845 <prefetch_dynamic_strides>: New const bool field.
7846 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
7847 prefetch_dynamic_strides.
7848 (exynosm1_prefetch_tune): Likewise.
7849 (thunderxt88_prefetch_tune): Likewise.
7850 (thunderx_prefetch_tune): Likewise.
7851 (thunderx2t99_prefetch_tune): Likewise.
7852 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
7854 (aarch64_override_options_internal): Update to set
7855 PARAM_PREFETCH_DYNAMIC_STRIDES.
7856 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
7857 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
7858 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
7859 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
7860 prefetch-dynamic-strides setting.
7862 2018-05-07 Luis Machado <luis.machado@linaro.org>
7864 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
7865 <minimum_stride>: New const int field.
7866 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
7867 minimum_stride field.
7868 (exynosm1_prefetch_tune): Likewise.
7869 (thunderxt88_prefetch_tune): Likewise.
7870 (thunderx_prefetch_tune): Likewise.
7871 (thunderx2t99_prefetch_tune): Likewise.
7872 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
7873 (aarch64_override_options_internal): Update to set
7874 PARAM_PREFETCH_MINIMUM_STRIDE.
7875 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
7876 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
7877 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
7878 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
7879 stride is constant and is below the minimum stride threshold.
7881 2018-05-07 Luis Machado <luis.machado@linaro.org>
7883 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
7886 2018-05-07 Luis Machado <luis.machado@linaro.org>
7888 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
7889 <prefetch_dynamic_strides>: New const bool field.
7890 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
7891 prefetch_dynamic_strides.
7892 (exynosm1_prefetch_tune): Likewise.
7893 (thunderxt88_prefetch_tune): Likewise.
7894 (thunderx_prefetch_tune): Likewise.
7895 (thunderx2t99_prefetch_tune): Likewise.
7896 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
7898 (aarch64_override_options_internal): Update to set
7899 PARAM_PREFETCH_DYNAMIC_STRIDES.
7900 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
7901 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
7902 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
7903 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
7904 prefetch-dynamic-strides setting.
7906 2018-05-07 Luis Machado <luis.machado@linaro.org>
7908 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
7909 <minimum_stride>: New const int field.
7910 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
7911 minimum_stride field.
7912 (exynosm1_prefetch_tune): Likewise.
7913 (thunderxt88_prefetch_tune): Likewise.
7914 (thunderx_prefetch_tune): Likewise.
7915 (thunderx2t99_prefetch_tune): Likewise.
7916 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
7917 (aarch64_override_options_internal): Update to set
7918 PARAM_PREFETCH_MINIMUM_STRIDE.
7919 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
7920 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
7921 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
7922 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
7923 stride is constant and is below the minimum stride threshold.
7925 2018-05-06 Jakub Jelinek <jakub@redhat.com>
7928 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
7929 the type is addressable. Don't force op into register if it has
7932 2018-05-05 Roland McGrath <mcgrathr@google.com>
7935 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
7936 any section for which we don't know a specific type it should have,
7937 regardless of name. Previously this was done only for the exact
7938 names ".init_array", ".fini_array", and ".preinit_array".
7939 (default_elf_asm_named_section): Add comment about
7940 relationship with default_section_type_flags and SECTION_NOTYPE.
7941 (get_section): Don't consider it a type conflict if one side has
7942 SECTION_NOTYPE and the other doesn't, as long as neither has the
7943 SECTION_BSS et al used in the default_section_type_flags logic.
7945 2018-05-05 Tom de Vries <tom@codesourcery.com>
7948 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
7949 (workaround_barsyncs): New function.
7950 (nvptx_reorg): Use workaround_barsyncs.
7951 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
7952 (define_expand "nvptx_membar_cta"): New define_expand.
7953 (define_insn "*nvptx_membar_cta"): New insn.
7955 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
7957 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
7958 To improve optimization opportunities.
7959 * builtin-types.def: The new needed builtin types for the above.
7961 2018-05-04 Richard Biener <rguenther@suse.de>
7963 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
7964 * gimple-ssa-store-merging.c
7965 (imm_store_chain_info::output_merged_store): Remove redundant create,
7966 release split_store vector contents on failure.
7967 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
7968 scalar stmt vector on cache hit.
7970 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
7972 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
7974 * config.gcc (powerpc-xilinx-eabi*): Remove.
7975 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
7977 (fusion_addis_mem_combo_load): Ditto.
7978 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
7980 (rs6000_cpu_cpp_builtins): Ditto.
7981 * config/rs6000/rs6000-linux.c
7982 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
7983 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
7984 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
7986 (rs6000_setup_reg_addr_masks): Ditto.
7987 (rs6000_init_hard_regno_mode_ok): Ditto.
7988 (rs6000_option_override_internal): Ditto.
7989 (legitimate_lo_sum_address_p): Ditto.
7990 (rs6000_legitimize_address): Ditto.
7991 (rs6000_legitimize_reload_address): Ditto.
7992 (rs6000_legitimate_address_p): Ditto.
7993 (abi_v4_pass_in_fpr): Ditto.
7994 (setup_incoming_varargs): Ditto.
7995 (rs6000_gimplify_va_arg): Ditto.
7996 (rs6000_split_multireg_move): Ditto.
7997 (rs6000_savres_strategy): Ditto.
7998 (rs6000_emit_prologue_components): Ditto.
7999 (rs6000_emit_epilogue_components): Ditto.
8000 (rs6000_emit_prologue): Ditto.
8001 (rs6000_emit_epilogue): Ditto.
8002 (rs6000_elf_file_end): Ditto.
8003 (rs6000_function_value): Ditto.
8004 (rs6000_libcall_value): Ditto.
8005 * config/rs6000/rs6000.h: Ditto.
8006 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
8007 (TARGET_MINMAX): ... this. New.
8008 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
8009 * config/rs6000/rs6000.md: Remove Xilinx FP support.
8010 (*movsi_internal1_single): Delete.
8011 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
8012 mfpu=, mxilinx-fpu): Delete.
8013 * config/rs6000/singlefp.h: Delete.
8014 * config/rs6000/sysv4.h: Remove Xilinx FP support.
8015 * config/rs6000/t-rs6000: Ditto.
8016 * config/rs6000/t-xilinx: Delete.
8017 * gcc/config/rs6000/titan.md: Adjust for fp_type removal.
8018 * gcc/config/rs6000/vsx.md: Remove Xilinx FP support.
8019 (VStype_simple): Delete.
8020 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
8021 * config/rs6000/xfpu.h: Delete.
8022 * config/rs6000/xfpu.md: Delete.
8023 * config/rs6000/xilinx.h: Delete.
8024 * config/rs6000/xilinx.opt: Delete.
8025 * gcc/doc/invoke.texi (RS/6000 and PowerPC Options): Remove
8026 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
8028 2018-05-04 Tom de Vries <tom@codesourcery.com>
8031 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
8034 2018-05-04 Richard Biener <rguenther@suse.de>
8037 * tree-complex.c (update_complex_assignment): We are always in SSA form.
8038 (expand_complex_div_wide): Likewise.
8039 (expand_complex_operations_1): Likewise.
8040 (expand_complex_libcall): Preserve EH info of the original stmt.
8041 (tree_lower_complex): Handle removed blocks.
8042 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
8043 on complex multiplication and division libcall builtins.
8045 2018-05-04 Richard Biener <rguenther@suse.de>
8048 * fold-const.c (negate_expr_p): Restrict negation of operand
8049 zero of a division to when we know that can happen without
8051 (fold_negate_expr_1): Likewise.
8053 2018-05-04 Jakub Jelinek <jakub@redhat.com>
8056 * real.h (real_nextafter): Declare.
8057 * real.c (real_nextafter): New function.
8058 * fold-const-call.c (fold_const_nextafter): New function.
8059 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
8060 CASE_CFN_NEXTTOWARD.
8061 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
8062 even when arg1_mode is different from arg0_mode.
8064 2018-05-03 Nathan Sidwell <nathan@acm.org>
8066 * doc/extend.texi (Deprecated Features): Remove
8068 (Backwards Compatibility): Likewise.
8069 * doc/invoke.texi (C++ Language Options): Likewise.
8070 (C++ Dialect Options): Likewise.
8072 2018-05-03 Jakub Jelinek <jakub@redhat.com>
8075 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
8076 _mm512_mask_mullox_epi64): New intrinsics.
8078 2018-05-03 Tom de Vries <tom@codesourcery.com>
8081 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
8082 dump files): Add offload-tree.
8084 2018-05-03 Richard Biener <rguenther@suse.de>
8086 PR tree-optimization/85615
8087 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
8088 to loops not nested in BBs loop father to avoid creating multi-entry
8091 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8093 PR tree-optimization/70291
8094 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
8095 arguments. Change return type to tree. Emit libcall as a new
8096 statement rather than replacing existing one when inplace_p is true.
8097 (expand_complex_multiplication_components): New function.
8098 (expand_complex_multiplication): Expand floating-point complex
8099 multiplication using the above.
8100 (expand_complex_division): Rename inner_type parameter to type.
8101 Update expand_complex_libcall call-site.
8102 (expand_complex_operations_1): Update expand_complex_multiplication
8103 and expand_complex_division call-sites.
8105 2018-05-02 Jakub Jelinek <jakub@redhat.com>
8108 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
8109 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
8110 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
8111 the highest significant bit of the shift count mask is clear. In
8112 check whether and[sq]i3 is needed verify that all significant bits
8113 of the shift count other than the highest are set.
8115 2018-05-02 Tom de Vries <tom@codesourcery.com>
8118 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
8119 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
8120 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
8121 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
8122 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
8123 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
8124 __builtin_goacc_parlevel_size.
8126 2018-05-02 Richard Biener <rguenther@suse.de>
8128 PR tree-optimization/85597
8129 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
8130 do not use split vect_get_vec_defs call but call vect_get_slp_defs
8133 2018-05-02 Tom de Vries <tom@codesourcery.com>
8136 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
8137 dump files): Add ltrans-tree.
8139 2018-05-02 Tom de Vries <tom@codesourcery.com>
8142 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
8143 dump files): Add wpa-ipa.
8145 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
8147 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
8148 powerpc*-*-linux*paired* target.
8149 * config/rs6000/750cl.h: Delete.
8150 * config/rs6000/paired.h: Delete.
8151 * config/rs6000/paired.md: Delete.
8152 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
8154 * config/rs6000/rs6000-builtin.def: Remove paired float support.
8155 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
8156 comment. Remove paired float support.
8157 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
8158 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
8160 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
8161 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
8163 * config/rs6000/rs6000.c: Remove paired float support.
8164 (paired_expand_vector_init, paired_expand_vector_move,
8165 paired_emit_vector_compare, paired_emit_vector_cond_expr,
8166 (paired_expand_lv_builtin, paired_expand_stv_builtin,
8167 paired_expand_builtin, paired_expand_predicate_builtin,
8168 paired_init_builtins): Delete.
8169 * config/rs6000/rs6000.h: Remove paired float support.
8170 * config/rs6000/rs6000.md: Remove paired float support.
8171 (move_from_CR_ov_bit): Delete.
8172 * config/rs6000/rs6000.opt (mpaired): Delete.
8173 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
8174 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
8176 2018-05-02 Richard Biener <rguenther@suse.de>
8179 * gimplify.c (gimplify_save_expr): When in SSA form allow
8180 SAVE_EXPRs to compute to SSA vars.
8182 2018-05-02 Jakub Jelinek <jakub@redhat.com>
8185 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
8186 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
8187 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
8188 clobber operands[2], instead use a new pseudo. Formatting fixes.
8190 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
8192 PR tree-optimization/85586
8193 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
8194 exit early for statements in the same group if the accesses are
8197 2018-05-02 Tom de Vries <tom@codesourcery.com>
8200 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
8203 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
8205 PR tree-optimization/85143
8206 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
8208 2018-05-01 Tom de Vries <tom@codesourcery.com>
8211 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
8212 not found" error message.
8214 2018-05-01 Tom de Vries <tom@codesourcery.com>
8217 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
8218 * vec.c (test_ordered_remove_if): New function.
8219 (vec_c_tests): Call test_ordered_remove_if.
8220 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
8221 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
8222 * tree-vect-patterns.c (vect_pattern_recog_1): Use
8223 VEC_ORDERED_REMOVE_IF.
8225 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8227 PR tree-optimization/82665
8228 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
8229 pointer subtraction where arguments come from a memchr call.
8231 2018-05-01 Jakub Jelinek <jakub@redhat.com>
8233 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
8234 --push-state --as-needed and --pop-state instead of --as-needed and
8235 --no-as-needed if ld supports it.
8236 * configure: Regenerated.
8239 * doc/install.texi2html: Replace _002d with - and _002a with * in
8240 generated html files using sed.
8242 2018-04-30 David Malcolm <dmalcolm@redhat.com>
8245 * gcc-rich-location.c (blank_line_before_p): New function.
8246 (use_new_line): New function.
8247 (gcc_rich_location::add_fixit_insert_formatted): New function.
8248 * gcc-rich-location.h
8249 (gcc_rich_location::add_fixit_insert_formatted): New function.
8251 2018-04-30 David Malcolm <dmalcolm@redhat.com>
8253 * selftest.c (assert_streq): Rename "expected" and "actual" to
8254 "val1" and "val2". Extend NULL-handling to cover both inputs
8255 symmetrically, while still requiring both to be non-NULL for a pass.
8256 * selftest.h (assert_streq): Rename "expected" and "actual" to
8258 (ASSERT_EQ): Likewise.
8259 (ASSERT_EQ_AT): Likewise.
8260 (ASSERT_KNOWN_EQ): Likewise.
8261 (ASSERT_KNOWN_EQ_AT): Likewise.
8262 (ASSERT_NE): Likewise.
8263 (ASSERT_MAYBE_NE): Likewise.
8264 (ASSERT_MAYBE_NE_AT): Likewise.
8265 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
8266 the assertion to pass.
8267 (ASSERT_STREQ_AT): Likewise.
8269 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
8271 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
8272 interaction with -pie.
8274 2018-04-30 David Malcolm <dmalcolm@redhat.com>
8276 * selftest.h: Fix alphabetization of per-source-file selftest
8279 2018-04-30 Jason Merrill <jason@redhat.com>
8281 PR c++/61982 - dead stores to destroyed objects.
8282 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
8285 2018-04-30 Jason Merrill <jason@redhat.com>
8287 * tree.c (build_clobber): New.
8288 * tree.h: Declare it.
8289 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
8291 2018-04-30 David Malcolm <dmalcolm@redhat.com>
8293 * diagnostic-show-locus.c (layout::layout): Update for
8294 location_get_source_line returning a char_span.
8295 (struct char_span): Move to input.h.
8296 (struct correction): Update for fields in char_span becoming
8298 (struct source_line): Update for location_get_source_line
8299 returning a char_span.
8300 (layout::print_line): Likewise.
8301 * edit-context.c (edited_file::print_content): Likewise.
8302 (edited_file::print_diff_hunk): Likewise.
8303 (edited_file::print_run_of_changed_lines): Likewise.
8304 (edited_file::get_num_lines): Likewise.
8305 (edited_line::edited_line): Likewise.
8306 * final.c (asm_show_source): Likewise.
8307 * input.c (location_get_source_line): Convert return type
8308 from const char * to char_span, losing the final "line_len"
8310 (dump_location_info): Update for the above.
8311 (get_substring_ranges_for_loc): Likewise. Use a char_span
8312 when handling the literal within the line.
8313 (test_reading_source_line): Update for location_get_source_line
8314 returning a char_span.
8315 * input.h (class char_span): Move here from
8316 diagnostic-show-locus.c, converting from a struct to a class.
8317 Make data members private.
8318 (char_span::operator bool): New.
8319 (char_span::length): New.
8320 (char_span::get_buffer): New.
8321 (char_span::operator[]): New.
8322 (char_span::subspan): Make const.
8323 (char_span::xstrdup): New.
8324 (location_get_source_line): Convert return type from const char *
8325 to char_span, losing the final "line_size" param.
8327 2018-04-30 Jan Hubicka <jh@suse.cz>
8329 * lto-wrapper.c (ltrans_priorities): New static var.
8330 (cmp_priority): New.
8331 (run_gcc): Read priorities and if doing parallel build order
8332 the Makefile by them.
8334 2018-04-30 David Malcolm <dmalcolm@redhat.com>
8336 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
8338 2018-04-30 Richard Biener <rguenther@suse.de>
8340 * tree-cfg.c (verify_address): Remove base argument, add
8341 flag whether to check TREE_ADDRESSABLE and do that.
8342 (verify_expr): Remove.
8343 (verify_types_in_gimple_reference): Add pieces from verify_expr.
8344 (verify_gimple_assign_single): Likewise.
8345 (verify_gimple_switch): Likewise.
8346 (verify_expr_location_1): Dereference tp once. Add (disabled)
8347 piece from verify_expr.
8348 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
8350 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
8352 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
8354 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
8356 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
8357 (small_data_pattern): Likewise.
8358 (arc_rewrite_small_data): Likewise.
8359 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
8360 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
8361 (get_symbol_alignment): New function.
8362 (legitimate_small_data_address_p): Likewise.
8363 (legitimate_scaled_address): Update, call
8364 legitimate_small_data_address_p.
8365 (output_sdata): New static variable.
8366 (arc_print_operand): Update how we handle small data operands.
8367 (arc_print_operand_address): Likewise.
8368 (arc_legitimate_address_p): Update, use
8369 legitimate_small_data_address_p.
8370 (arc_rewrite_small_data_p): Remove.
8371 (arc_rewrite_small_data_1): Likewise.
8372 (arc_rewrite_small_data): Likewise.
8373 (small_data_pattern): Likewise.
8374 (compact_sda_memory_operand): Update to use
8375 legitimate_small_data_address_p and get_symbol_alignment.
8376 (prepare_move_operands): Don't rewite sdata pattern.
8377 (prepare_extend_operands): Remove.
8378 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
8380 (zero_extendqisi2): Likewise.
8381 (zero_extendhisi2): Likewise.
8382 (extendqihi2): Likewise.
8383 (extendqisi2): Likewise.
8384 (extendhisi2): Likewise.
8388 * config/arc/constraints.md (Usd): Change it to memory constraint.
8390 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
8392 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
8393 as source of std instructions.
8394 * config/arc/arc.md (movsi_insn): Update pattern predicate to
8395 allow 6-bit constants as source for store instructions.
8396 (movdi_insn): Update instruction pattern to allow 6-bit constants
8397 as source for store instructions.
8399 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
8401 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
8403 2018-04-30 Nathan Sidwell <nathan@acm.org>
8404 Sandra Loosemore <sandra@codesourcery.com>
8406 * dumpfile.c (dump_open): Allow '-' for stdout.
8407 * doc/invoke.texi (Developer Options): Document dump filename
8408 determination early. Document stdin/stdout selection.
8410 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
8412 Microblaze Target: PIC data text relative
8414 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
8415 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
8417 * gcc/config/microblaze/microblaze.h (microblaze_constant_address_p):
8418 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
8419 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
8420 New addressing mode for data-text relative position indepenedent code.
8421 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
8422 'ADDRESS_SYMBOLIC_TXT_REL'.
8423 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
8424 (microblaze_legitimate_pic_operand): Exclude function calls from
8425 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
8426 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
8428 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
8429 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
8430 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
8431 for 'address + offset'.
8432 (microblaze_expand_prologue): Add new function prologue call for
8434 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
8435 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
8436 table in case of TARGET_PIC_DATA_TEXT_REL.
8437 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
8438 * gcc/config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
8439 Add new macros 'UNSPEC_TEXT',
8440 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
8441 + exclude function calls from 'UNSPEC_PLT' in case of data text
8443 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
8444 new target hook for generating address diff vector tables in case of
8446 * doc/tm.texi : Regenerate.
8447 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
8448 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
8449 of addr diff vector generation.
8450 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
8451 target hook definition.
8452 * targhooks.h, gcc/targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
8453 Add default function for generate_pic_addr_diff_vec -> flag_pic.
8454 * doc/invoke.texi (Add new pic option): Add new microblaze pic
8455 option for data text relative.
8457 2018-04-30 Richard Biener <rguenther@suse.de>
8459 * tree-chrec.h (evolution_function_is_constant_p): Remove
8461 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
8463 2018-04-30 Richard Biener <rguenther@suse.de>
8466 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
8468 2018-04-30 Richard Biener <rguenther@suse.de>
8470 PR tree-optimization/28364
8471 PR tree-optimization/85275
8472 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
8473 copying first exit test.
8475 2018-04-28 Mark Wielaard <mark@klomp.org>
8477 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
8479 (dwarf_AT): Handle DW_AT_addr_base.
8480 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
8482 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
8485 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
8486 (*ashl<dwi>3_doubleword_mask_1): Ditto.
8487 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
8488 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
8490 2018-04-28 Richard Biener <rguenther@suse.de>
8492 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
8493 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
8494 to reflect use. Only add interesting stmts.
8496 2018-04-27 Martin Jambor <mjambor@suse.cz>
8499 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
8500 the jump function allows for passing through aggregate values.
8502 2018-04-27 David Malcolm <dmalcolm@redhat.com>
8504 * input.h (in_system_header_at): Convert from macro to inline
8506 (from_macro_expansion_at): Likewise.
8507 (from_macro_definition_at): Likewise.
8509 2018-04-27 Jeff Law <law@redhat.com>
8511 * config.gcc: Mark tile* targets as deprecated/obsolete.
8513 2018-04-27 Richard Biener <rguenther@suse.de>
8515 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
8518 2018-04-27 Richard Biener <rguenther@suse.de>
8520 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
8522 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
8524 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
8525 with Yd constraint. Set "preferred_for_speed" attribute from
8526 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
8528 (*movdi_internal): Ditto.
8529 (movti_interunit splitters): Remove
8530 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
8531 (movdi_interunit splitters): Ditto.
8532 * config/i386/constraints.md (Ye): Remove.
8533 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
8535 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8538 * config/aarch64/constraints.md (Usg): Limit to 31.
8541 2018-04-27 Jakub Jelinek <jakub@redhat.com>
8543 PR tree-optimization/85529
8544 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
8545 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
8546 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
8547 zero extension or masking of the MSB bit.
8548 (optimize_range_tests): Add FIRST_BB argument, pass it through
8549 to optimize_range_tests_var_bound.
8550 (maybe_optimize_range_tests, reassociate_bb): Adjust
8551 optimize_range_tests callers.
8553 2018-04-26 Richard Biener <rguenther@suse.de>
8554 Jakub Jelinek <jakub@redhat.com>
8556 * cgraph.h (symbol_table): Just declare debug method here.
8557 * symtab.c (symbol_table::debug): Define.
8559 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
8561 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
8563 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
8565 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
8566 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
8567 (*movdi_internal): Substitute Yi and Yj constraint with x
8568 and Ym and Yn constraint with y constraint. Update "isa"
8569 attribute and set "preferred_for_speed" attribute from
8570 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
8571 (*movsi_internal): Ditto.
8572 (*movdf_internal): Ditto.
8573 (*movsf_internal): Ditto.
8574 (*zero_extendsidi2): Ditto.
8575 * config/i386/sse.md (vec_set<mode>_0): Ditto.
8576 (sse2_loadld): Ditto.
8577 (*vec_extract<ssevecmodelower>_0): Ditto.
8578 (*vec_extractv4si_0_zext_sse4): Ditto.
8579 (vec_concatv2di): Ditto.
8580 (*vec_dup<mode>): Ditto.
8581 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
8582 * config/i386/constraints.md (Yi): Remove.
8587 2018-04-26 Nathan Sidwell <nathan@acm.org>
8589 * dumpfile.c (dump_open): New.
8590 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
8591 (dump_finish): Detect stdio/stderr by value not name.
8593 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
8595 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
8597 2018-04-26 Tom de Vries <tom@codesourcery.com>
8600 * config/nvptx/nvptx.c (verify_neutering_jumps)
8601 (verify_neutering_labels): New function
8602 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
8604 2018-04-26 Tom de Vries <tom@codesourcery.com>
8607 * config/nvptx/nvptx.c (needs_neutering_p): New function.
8608 (nvptx_single): Use needs_neutering_p to skip over insns that do not
8611 2018-04-26 Richard Biener <rguenther@suse.de>
8612 Tom de Vries <tom@codesourcery.com>
8615 * lto-streamer-out.c (output_function): Fixup loops if required to match
8616 discovery done in the reader.
8618 2018-04-26 Richard Biener <rguenther@suse.de>
8620 PR tree-optimization/85116
8621 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
8622 have a loop exit from the single latch predecessor. Remove
8623 case of header with just condition.
8624 (ch_base::copy_headers): Exclude infinite loops from any
8626 (pass_ch::execute): Record exits.
8628 2018-04-26 Richard Biener <rguenther@suse.de>
8630 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
8631 prologue cost vector and pass it to vect_get_load_cost.
8632 (vect_get_peeling_costs_all_drs): Likewise.
8633 (vect_peeling_hash_get_lowest_cost): Likewise.
8634 (vect_enhance_data_refs_alignment): Likewise.
8636 2018-04-26 Richard Biener <rguenther@suse.de>
8639 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
8640 checking of integer<->pointer conversions.
8641 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
8642 sign-/zero-extending pointer types.
8643 (expand_omp_for_static_chunk): Likewise.
8645 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
8646 Jean Lee <xiaoyur347@gmail.com>
8648 * config/mips/mips.c (mips_asan_shadow_offset): New function.
8649 (TARGET_ASAN_SHADOW_OFFSET): Define.
8650 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
8651 true for -fsanitize=address.
8653 2018-04-25 Mark Wielaard <mark@klomp.org>
8655 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
8658 2018-04-25 Jakub Jelinek <jakub@redhat.com>
8660 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
8661 than "alu", remove explicit "memory" and "imm_disp" attributes.
8662 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
8665 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
8666 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
8667 gen_lowpart_no_emit.
8669 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
8672 * config/i386/i386.c (ix86_expand_builtin): Change memory
8673 operand to XI, extend p0 to Pmode.
8674 * config/i386/i386.md: Change unspec volatile and operand
8675 1 mode to XI, change operand 0 mode to P.
8677 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
8679 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
8680 GET_MODE_MASK before any checking.
8681 (nds32_can_use_bset_p): Likewise.
8682 (nds32_can_use_btgl_p): Likewise.
8684 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
8686 * config/nds32/nds32-doubleword.md: New define_split pattern for
8687 illegal register number.
8689 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
8691 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
8693 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
8695 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
8697 2018-04-25 Richard Biener <rguenther@suse.de>
8699 * lto-streamer.h (LTO_major_version): Bump to 8.
8701 2018-04-25 Jakub Jelinek <jakub@redhat.com>
8703 * BASE-VER: Set to 9.0.0.
8705 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
8707 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
8708 in __abskf2 and __powikf2.
8710 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8713 * config/aarch64/constraints.md (Usg, Usj): New constraints.
8714 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
8715 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
8716 Use the above on operand 2. Reindent.
8717 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
8719 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
8722 * common/config/i386/i386-common.c (ix86_handle_option): Don't
8724 * config/i386/i386.opt (mcet): Removed.
8725 * doc/install.texi: Remove -mcet documentation.
8726 * doc/invoke.texi: Likewise.
8728 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
8731 * doc/install.texi: Remove -mcet from bootstrap-cet.
8733 2018-04-24 Jakub Jelinek <jakub@redhat.com>
8736 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
8737 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
8741 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
8742 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
8743 containing a CONST_VECTOR.
8745 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
8747 * doc/install.texi: Update newlib dependency for nvptx.
8749 2018-04-24 Jakub Jelinek <jakub@redhat.com>
8752 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
8753 instead of INTVAL when shifting x left.
8755 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
8757 PR tree-optimization/85478
8758 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
8759 vect_grouped_store_supported for single element vectors.
8761 2018-04-24 Richard Biener <rguenther@suse.de>
8764 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
8765 load cost increase to the case of non-constant step.
8767 2018-04-24 Jakub Jelinek <jakub@redhat.com>
8770 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
8771 destination if any_malformed_asm.
8773 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
8776 * expr.c (store_field): In the bitfield case, if the value comes from
8777 a function call and is returned in registers by means of a PARALLEL,
8778 do not change the mode of the temporary unless BLKmode and VOIDmode.
8780 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
8782 PR rtl-optimization/85423
8783 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
8784 dependencies to debug insns when the previous insn is non-debug.
8786 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
8788 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
8789 enums into a single definition.
8790 (fls): Fix predicates and printing.
8793 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
8795 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
8796 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
8797 and short u6 immediate.
8798 (check_if_valid_sleep_operand): Remove.
8799 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
8801 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
8803 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
8804 flag_always_save_lp condition.
8805 * config/nds32/nds32.opt (malways-save-lp): New option.
8807 2018-04-22 Shiva Chen <shiva0217@gmail.com>
8809 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
8810 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
8811 * config/nds32/nds32.h
8812 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
8813 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
8815 2018-04-22 Shiva Chen <shiva0217@gmail.com>
8817 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
8818 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
8820 2018-04-22 Shiva Chen <shiva0217@gmail.com>
8821 Chung-Ju Wu <jasonwucj@gmail.com>
8823 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
8825 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
8826 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
8828 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
8830 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
8832 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
8834 * config/nds32/nds32-protos.h (nds32_data_alignment,
8835 nds32_local_alignment): Declare.
8836 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
8837 nds32_local_alignment): New functions.
8838 (TARGET_CONSTANT_ALIGNMENT): Define.
8839 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
8841 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
8843 * config/nds32/nds32.c
8844 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
8845 (TARGET_MODES_TIEABLE_P): Likewise.
8847 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
8849 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
8852 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
8853 Chung-Ju Wu <jasonwucj@gmail.com>
8855 * config/nds32/constants.md (unspec_volatile_element): Add enum values
8856 for unaligned access.
8857 * config/nds32/nds32-intrinsic.c: Implementation of expanding
8859 * config/nds32/nds32-intrinsic.md: Likewise.
8860 * config/nds32/nds32_intrinsic.h: Likewise.
8861 * config/nds32/nds32.h (nds32_builtins): Likewise.
8862 * config/nds32/nds32.opt (munaligned-access): New option.
8863 * config/nds32/nds32.c (nds32_asm_file_start): Display
8864 flag_unaligned_access status.
8866 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
8868 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
8869 -mno-relax is present.
8870 * config/riscv/linux.h (LINK_SPEC): Ditto.
8872 2018-04-20 Martin Sebor <msebor@redhat.com>
8875 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
8877 (gimple_fold_builtin_stxcpy_chk): Same.
8878 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
8880 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
8883 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
8884 __powikf2 when long double is IEEE 128-bit.
8886 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
8888 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
8889 step to make sure stack always aligned.
8891 2018-04-20 Carl Love <cel@us.ibm.com>
8894 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
8895 size check for arg0.
8897 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
8898 Tom de Vries <tom@codesourcery.com>
8901 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
8902 Emit insns for calls too.
8903 (nvptx_find_par): Always look for worker-level predecessor insn.
8904 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
8906 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
8907 (nvptx_process_pars): Propagate frames for calls.
8909 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
8912 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
8914 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
8915 (ix86_handle_option): Don't handle OPT_mibt.
8916 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
8918 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
8920 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
8921 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
8922 (ix86_target_macros): Define __CET__ with flag_cf_protection
8923 for -fcf-protection.
8924 * config/i386/i386.c (isa2_opts): Remove -mibt.
8925 * config/i386/i386.h (TARGET_IBT): Removed.
8926 (TARGET_IBT_P): Likewise.
8927 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
8928 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
8929 * config/i386/i386.opt (mcet): Update help message.
8932 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
8933 -mcet as an alias for -mshstk.
8935 2018-04-20 Richard Biener <rguenther@suse.de>
8938 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
8939 complexity by forcing a single use of the multiply operand.
8941 2018-04-20 Martin Jambor <mjambor@suse.cz>
8944 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
8945 recursion dependency to only apply to non-clones.
8947 2018-04-20 Martin Jambor <mjambor@suse.cz>
8950 * ipa-cp.c (create_specialized_node): Check that clones of
8951 self-recursive edges exist during IPA-CP.
8953 2018-04-19 Toon Moene <toon@moene.org>
8955 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
8958 2018-04-19 Jakub Jelinek <jakub@redhat.com>
8960 PR tree-optimization/85467
8961 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
8962 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
8963 VECTOR_CST element to type.
8965 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
8968 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
8969 * config/i386/i386.md (builtin_setjmp_setup): Removed.
8970 (builtin_longjmp): Likewise.
8971 (save_stack_nonlocal): New pattern.
8972 (restore_stack_nonlocal): Likewise.
8974 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
8977 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
8978 Replace ASM_OUTPUT_LABEL with fprintf.
8980 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
8983 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
8984 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
8985 * config/i386/i386-c.c (ix86_target_macros_internal): Also
8986 define __IBT__ and __SHSTK__ for -fcf-protection.
8987 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
8989 (ix86_trampoline_init): Likewise.
8990 (x86_output_mi_thunk): Likewise.
8991 (ix86_notrack_prefixed_insn_p): Likewise.
8992 (ix86_option_override_internal): Don't disallow -fcf-protection.
8993 * config/i386/i386.md (rdssp<mode>): Also enable for
8995 (incssp<mode>): Likewise.
8996 (nop_endbr): Likewise.
8997 * config/i386/i386.opt (mcet): Change help message to built-in
9001 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
9002 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
9003 enable CET built-in functions.
9005 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
9007 * common/config/i386/i386-common.c
9008 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
9009 OPTION_MASK_ISA_MOVDIRI_UNSET,
9010 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
9011 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
9012 * config.gcc (movdirintrin.h): New header.
9013 * config/i386/cpuid.h (bit_MOVDIRI,
9014 bit_MOVDIR64B): New bits.
9015 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
9017 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
9018 (VOID, PVOID, PCVOID)): New function types.
9019 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
9020 __builtin_ia32_directstoreu_u64,
9021 __builtin_ia32_movdir64b): New builtins.
9022 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
9023 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
9025 (ix86_valid_target_attribute_inner_p): Ditto.
9026 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
9027 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
9028 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
9029 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
9030 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
9031 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
9032 (movdiri<mode>, movdir64b_<mode>): New.
9033 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
9034 * config/i386/immintrin.h: Include movdirintrin.h.
9035 * config/i386/movdirintrin.h: New file.
9036 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
9038 2018-04-19 Richard Biener <rguenther@suse.de>
9041 * cfg.c (clear_bb_flags): When loop state says we have
9042 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
9044 2018-04-19 Richard Biener <rguenther@suse.de>
9046 PR tree-optimization/84737
9047 * tree-vect-data-refs.c (vect_copy_ref_info): New function
9048 copying restrict info.
9049 (vect_setup_realignment): Use it.
9050 * tree-vectorizer.h (vect_copy_ref_info): Declare.
9051 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
9052 the first DR to all generated stores.
9053 (vectorizable_load): Likewise for loads.
9055 2018-04-19 Jakub Jelinek <jakub@redhat.com>
9057 PR tree-optimization/85446
9058 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
9059 the integral and pointer types to have the same precision.
9061 * doc/install.texi: Document --disable-cet being the default and
9064 2018-04-18 Martin Liska <mliska@suse.cz>
9066 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
9069 2018-04-18 Martin Liska <mliska@suse.cz>
9072 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
9075 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
9076 arguments if they are comparable.
9078 2018-04-18 Martin Liska <mliska@suse.cz>
9081 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
9084 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
9087 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
9090 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
9091 ENDBR after calling __morestack.
9093 2018-04-18 David Malcolm <dmalcolm@redhat.com>
9096 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
9097 by using gcc_base_ver to generate a gcc_driver_version, and use
9098 it when generating GCC_DRIVER_NAME.
9099 * configure: Regenerate.
9101 2018-04-18 Jakub Jelinek <jakub@redhat.com>
9104 * config.gcc: Obsolete powerpc*-*-*spe*.
9106 2018-04-17 Jakub Jelinek <jakub@redhat.com>
9109 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
9110 (stabstr_D): Change type of unum from unsigned int to
9111 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
9114 2018-04-17 Jim Wilson <jimw@sifive.com>
9117 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
9118 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
9119 Set arg_pointer_offset after using pretend_args_size.
9121 2018-04-17 Jakub Jelinek <jakub@redhat.com>
9123 PR rtl-optimization/85431
9124 * dse.c (record_store): Ignore zero width stores.
9127 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
9128 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
9129 __builtin_stack_restore rather than after it.
9130 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
9131 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
9132 argument instead of virtual_dynamic_stack_rtx.
9134 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
9136 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
9138 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9139 Add note to error message to explain internal mapping of overloaded
9140 built-in function name to non-overloaded built-in function name.
9141 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
9144 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
9147 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
9148 where the inputs overlap with the output.
9150 2018-04-17 Jakub Jelinek <jakub@redhat.com>
9152 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
9153 (=v, v) alternative and explicit "memory" attribute.
9154 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
9155 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
9157 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
9158 "sselog1" type instead of "sselog".
9159 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
9160 "sselog". Remove explicit "memory" attribute.
9161 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
9162 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
9164 (vec_extract_hi_v32hi): Merge all alternatives into one, use
9165 "sselog1" type instead of "sselog". Remove explicit "memory"
9167 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
9168 use "sselog1" type instead of "sselog". Remove explicit "memory"
9170 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
9171 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
9173 (vec_extract_hi_v64qi): Merge all alternatives into one, use
9174 "sselog1" type instead of "sselog". Remove explicit "memory"
9176 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
9177 use "sselog1" type instead of "sselog". Remove explicit "memory"
9181 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
9184 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
9187 2018-04-17 Martin Jambor <mjambor@suse.cz>
9190 * ipa-cp.c (create_specialized_node): Call
9191 expand_all_artificial_thunks if necessary.
9193 2018-04-17 Martin Liska <mliska@suse.cz>
9196 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
9197 in message, remote space in between '_G' and '('.
9199 2018-04-17 Jakub Jelinek <jakub@redhat.com>
9202 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
9203 avx512f_vmcmp<mode>3<round_saeonly_name>,
9204 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
9205 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
9206 avx512f_rndscale<mode><round_saeonly_name>,
9207 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
9208 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
9209 Use %<iptr>2 instead of %2 for -masm=intel.
9210 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
9211 avx512f_vcvttss2usi<round_saeonly_name>,
9212 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
9214 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
9215 avx512f_vcvttsd2usi<round_saeonly_name>,
9216 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
9217 Use %q1 instead of %1 for -masm=intel.
9218 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
9219 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
9220 of %3 for -masm=intel.
9221 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
9223 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
9225 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
9227 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
9228 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
9230 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
9232 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
9233 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
9234 %g1 and one with %0 and %1.
9235 (avx512er_vmrcp28<mode><round_saeonly_name>,
9236 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
9238 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
9239 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
9240 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
9241 of %0 and %{%4%} for -masm=intel.
9242 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
9243 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
9244 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
9245 order of %0 and %{%5%}%{z%} for -masm=intel.
9247 2018-04-17 Jan Hubicka <jh@suse.cz>
9250 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
9252 2018-04-17 Martin Liska <mliska@suse.cz>
9255 * multiple_target.c (create_dispatcher_calls): Set apostrophes
9256 for target_clone error message. Make default implementation
9257 clone to be a local declaration.
9258 (separate_attrs): Add new argument and check for an empty
9260 (expand_target_clones): Handle it.
9261 (ipa_target_clone): Make redirection just for target_clones
9264 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
9265 Tom de Vries <tom@codesourcery.com>
9268 * omp-expand.c (expand_oacc_for): Add dummy false branch for
9269 tiled basic blocks without omp continue statements.
9271 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
9274 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
9275 vec_extract expression as having side effects to make sure it gets
9278 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
9281 * config/i386/i386.c (get_builtin_code_for_version): Check
9284 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
9287 * gcc/config.gcc: Support "skylake".
9288 * gcc/config/i386/i386-c.c (ix86_target_macros_internal): Handle
9290 * gcc/config/i386/i386.c (m_SKYLAKE): Define.
9291 (processor_target_table): Add "skylake".
9292 (ix86_option_override_internal): Add "skylake".
9293 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
9294 PROCESSOR_CANNONLAKE.
9295 (get_builtin_code_for_version): Fix priority for
9296 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
9297 PROCESSOR_SKYLAKE-AVX512.
9298 * gcc/config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
9299 (processor_type): Add PROCESSOR_SKYLAKE.
9301 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
9302 Jason Merrill <jason@redhat.com>
9305 * convert.c (convert_to_integer_1): Use direct recursion for
9306 enumeral types and types with a precision less than the number
9307 of bits in their mode.
9309 2018-04-16 Julia Koval <julia.koval@intel.com>
9312 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
9313 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
9315 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
9318 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
9319 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
9320 and -mno-direct-move.
9322 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
9325 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
9326 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
9327 Ensure negative shifts result in {0}.
9329 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
9331 PR rtl-optimization/79916
9332 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
9333 regs (if any) to define how to gnerate SD moves when LRA is in
9336 2018-04-13 Jakub Jelinek <jakub@redhat.com>
9338 PR rtl-optimization/85393
9339 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
9340 * except.c (expand_dw2_landing_pad_for_region): Make static.
9341 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
9342 a label and unconditional jump to old_bb, rather than
9343 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
9346 PR rtl-optimization/85376
9347 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
9348 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
9349 instead of a specific value.
9351 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
9352 Bin Cheng <bin.cheng@arm.com>
9354 PR tree-optimization/82965
9355 PR tree-optimization/83991
9356 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
9357 by_profile_only parameter.
9358 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
9359 information if the loop was predicted to iterate too many times.
9360 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
9362 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
9365 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
9368 2018-04-13 Martin Liska <mliska@suse.cz>
9369 Jakub Jelinek <jakub@redhat.com>
9372 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
9373 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
9374 * builtins.c (expand_builtin_memory_copy_args): Use
9375 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
9376 handle dest_addr == pc_rtx.
9378 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
9381 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
9382 asked to not generate direct moves.
9383 (fix_trunc<mode>si2_stfiwx): Similar.
9384 (fix_trunc<mode>si2_internal): Similar.
9386 2018-04-12 Jakub Jelinek <jakub@redhat.com>
9389 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
9390 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
9391 lookup if dest in some wider mode is known to be const0_rtx and
9392 if so, record permanent equivalence for it to be ZERO_EXTEND of
9393 the narrower mode destination.
9395 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
9397 * lto-streamer-out.c (output_function): Revert 259346.
9398 * omp-expand.c (expand_oacc_for): Likewise.
9400 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
9402 PR rtl-optimization/85354
9403 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
9404 * sel-sched.c (sel_global_init): ... here.
9406 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
9409 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
9410 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
9411 mode for PE-COFF targets.
9412 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
9413 (i386_pe_asm_lto_end): Likewise.
9414 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
9415 (TARGET_ASM_LTO_END): Likewise.
9416 * config/i386/winnt.c (saved_debug_info_level): New static variable.
9417 (i386_pe_asm_lto_start): New function.
9418 (i386_pe_asm_lto_end): Likewise.
9420 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
9421 Richard Biener <rguenther@suse.de>
9424 * lto-streamer-out.c (output_function): Fix CFG loop state before
9426 * omp-expand.c (expand_oacc_for): Handle calls to internal
9427 functions like regular functions.
9429 2018-04-12 Richard Biener <rguenther@suse.de>
9432 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
9433 for the early LTO debug to properly generate references to it
9434 during DIE emission. Do not re-use that for the skeleton for
9436 (dwarf2out_early_finish): Likewise.
9438 2018-04-12 Jakub Jelinek <jakub@redhat.com>
9441 * config/i386/sse.md
9442 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
9443 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
9444 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
9445 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
9446 and output is a reg, avoid creating invalid lowpart subreg, but
9447 instead split into a 512-bit move. Don't split if not AVX512VL,
9448 input is xmm16+ reg and output is a mem.
9449 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
9450 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
9451 xmm16+ reg and output is a mem.
9453 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9455 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
9456 also for flag_dwarf2_cfi_asm.
9458 2018-04-12 Jakub Jelinek <jakub@redhat.com>
9460 PR rtl-optimization/85342
9461 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
9462 a bool scalar var inside of the loop instead. Don't try to update
9463 recog_data.operand after failed apply_change_group.
9465 2018-04-12 Tom de Vries <tom@codesourcery.com>
9468 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
9469 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
9470 array with flexible array member as array without given dimension.
9471 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
9472 argument for undefined param to true.
9474 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
9477 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
9478 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
9479 from PowerPC section.
9480 * config/rs6000/sysv4.opt (mcall-): Improve help text.
9481 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
9482 help text that is too long.
9483 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
9484 help text that is too long.
9485 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
9486 help text that is too long.
9488 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
9490 * config/alpha/alpha.md (stack_probe_internal): Rename
9491 from "probe_stack". Update all callers.
9493 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
9495 PR rtl-optimization/84566
9496 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
9497 sched_macro_fuse_insns.
9499 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
9502 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
9503 (compute_block_dependences): ... from here.
9505 2018-04-11 Jakub Jelinek <jakub@redhat.com>
9507 PR tree-optimization/85331
9508 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
9509 from int to HOST_WIDE_INT.
9511 2018-04-11 Martin Jambor <mjambor@suse.cz>
9514 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
9515 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
9516 not the same as the source val.
9517 (cgraph_edge_brings_value_p): New parameter.
9518 (gather_edges_for_value): Pass destination value to
9519 cgraph_edge_brings_value_p.
9520 (perhaps_add_new_callers): Likewise.
9521 (get_info_about_necessary_edges): Likewise and exclude values brought
9522 only by self-recursive edges.
9523 (create_specialized_node): Redirect only clones of self-calling edges.
9524 (+self_recursive_pass_through_p): New function.
9525 (find_more_scalar_values_for_callers_subset): Use it.
9526 (find_aggregate_values_for_callers_subset): Likewise.
9527 (known_aggs_to_agg_replacement_list): Removed.
9528 (decide_whether_version_node): Re-calculate known constants for all
9529 remaining context clones.
9531 2018-04-11 Richard Biener <rguenther@suse.de>
9534 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
9535 from early DWARF output.
9536 (dwarf2out_early_finish): Output line info unconditionally into
9537 early DWARF and add reference to it.
9539 2018-04-11 Jakub Jelinek <jakub@redhat.com>
9542 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
9543 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
9544 other than V2DFmode using iptr mode attribute.
9545 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
9547 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
9549 PR rtl-optimization/84659
9550 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
9552 2018-04-11 Jakub Jelinek <jakub@redhat.com>
9555 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
9557 (output_loc_list): Pass address of a dummy size variable even in the
9558 locview handling loop.
9559 (index_location_lists): Add comment on why skip_loc_list_entry can't
9562 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
9565 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
9568 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
9571 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
9572 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
9573 and -mstring-compare-inline-limit.
9575 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
9578 * gcc/config/rs6000/rs6000.md (allocate_stack): Put the residual size
9579 for stack clash protection in a register whenever we need it to be in
9582 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
9584 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
9585 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
9587 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
9590 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
9592 (mlong-double-): Ditto.
9593 * config/rs6000/sysv4.opt (msdata=): Ditto.
9594 (mtls-size=): Ditto.
9596 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
9598 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
9599 erroneous entries for
9600 "vector int vec_ldl (int, long int *)", and
9601 "vector unsigned int vec_ldl (int, unsigned long int *)".
9602 Add comments and entries for
9603 "vector bool char vec_ldl (int, bool char *)",
9604 "vector bool short vec_ldl (int, bool short *)",
9605 "vector bool int vec_ldl (int, bool int *)",
9606 "vector bool long long vec_ldl (int, bool long long *)",
9607 "vector pixel vec_ldl (int, pixel *)",
9608 "vector long long vec_ldl (int, long long *)",
9609 "vector unsigned long long vec_ldl (int, unsigned long long *)".
9610 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
9611 type tree bool_long_long_type_node and correct definition of
9612 bool_V2DI_type_node to make reference to this new type tree.
9613 (rs6000_mangle_type): Replace erroneous reference to
9614 bool_long_type_node with bool_long_long_type_node.
9615 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
9616 comments to emphasize sign distinctions for char and int types and
9617 replace RS6000_BTI_bool_long constant with
9618 RS6000_BTI_bool_long_long constant. Also add comment to restrict
9619 use of RS6000_BTI_pixel.
9620 (bool_long_type_node): Remove this macro definition.
9621 (bool_long_long_type_node): New macro definition
9623 2018-04-10 Jakub Jelinek <jakub@redhat.com>
9625 PR rtl-optimization/85300
9626 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
9627 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
9628 simplify_unary_operation fails.
9630 2018-04-10 Martin Liska <mliska@suse.cz>
9632 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
9633 cgraph_edge and ipa_ref.
9635 2018-04-10 Jakub Jelinek <jakub@redhat.com>
9639 * config/i386/sse.md
9640 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
9641 computation of the VEC_MERGE selector from mask.
9642 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
9643 Fix decoding of the VEC_MERGE selector into mask.
9645 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
9647 PR tree-optimization/85286
9648 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
9650 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
9652 * final.c (final_1): Set insn_last_address as well as
9653 insn_current_address.
9655 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9658 * explow.c (emit_stack_probe): Call validize_mem on memory location
9659 before passing it to gen_probe_stack. Create address operand and
9660 legitimize it for the probe_stack_address case.
9662 2018-04-09 Jan Hubicka <jh@suse.cz>
9665 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
9666 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
9667 * tree.c (free_lang_data_in_type): Fix handling of binfos;
9669 (free_lang_data): Rebuild type inheritance graph.
9671 2018-04-09 Martin Sebor <msebor@redhat.com>
9673 * invoke.texi (-finline-small-functions): Mention other optimization
9675 (-findirect-inlining, -fpartial-inlining): Same.
9676 (-finline-functions-called-once): Same.
9677 (-freorder-blocks-and-partition): Same.
9679 2018-04-09 Jan Hubicka <jh@suse.cz>
9682 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
9683 jumps; choose last target that matches the criteria (i.e.
9684 no partition changes for non-crossing jumps).
9685 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
9686 support for redirecting crossing jumps to non-crossing.
9688 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
9690 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
9691 also for naked functions.
9693 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
9695 * config/arc/arc.md (add_shift): New pattern.
9696 (add_shift2): Likewise.
9697 (sub_shift): Likewise.
9698 (sub_shift_cmp0_noout): Likewise.
9699 (compare_si_ashiftsi): Likewise.
9700 (xbfu_cmp0_noout): New combine pattern.
9701 (xbfu_cmp0"): Likewise.
9702 (movsi_set_cc_insn): Place the predicable variant first.
9703 (commutative_binary_cmp0_noout): Remove clobber.
9704 (commutative_binary_cmp0): New pattern.
9705 (noncommutative_binary_cmp0): Likewise.
9706 (noncommutative_binary_cmp0_noout): Likewise.
9707 (noncommutative_binary_comparison_result_used): Removed.
9708 (rsub_cmp0): New pattern.
9709 (rsub_cmp0_noout): Likewise.
9710 (extzvsi): Changed, keep only meaningful variants.
9711 (SQH, SEZ): New iterators.
9712 (SQH_postfix): New mode attribute.
9713 (SEZ_prefix): New code attribute.
9714 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
9715 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
9716 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
9718 (noncommutative_operator): Check the availability of barrel
9721 2018-04-09 Richard Biener <rguenther@suse.de>
9723 PR tree-optimization/85284
9724 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
9725 Only use the niter constraining form of simple_iv when the exit
9728 2018-04-09 Tom de Vries <tom@codesourcery.com>
9731 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
9732 (define_expand "*memory_barrier"): New define_expand.
9733 (define_insn "memory_barrier"): New insn.
9735 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
9737 PR rtl-optimization/80463
9738 PR rtl-optimization/83972
9739 PR rtl-optimization/83480
9741 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
9742 correct producer for the insn.
9743 (tidy_control_flow): Fixup seqnos in case of debug insns.
9745 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
9747 PR rtl-optimization/83913
9749 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
9750 different sched-times when merging exprs.
9752 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
9754 PR rtl-optimization/83962
9756 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
9757 tidy_fallthru_edge and tidy_control_flow.
9759 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
9761 PR rtl-optimization/83530
9763 * sel-sched.c (force_next_insn): New global variable.
9764 (remove_insn_for_debug): When force_next_insn is true, also leave only
9765 next insn in the ready list.
9766 (sel_sched_region): When the region wasn't scheduled, make another pass
9767 over it with force_next_insn set to 1.
9769 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
9771 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
9773 * config/nds32/constants.md (unspec_volatile_element): Add enum values
9774 for interrupt control.
9775 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
9776 functions for interrupt control.
9777 * config/nds32/nds32-intrinsic.md: Likewise.
9778 * config/nds32/nds32_intrinsic.h: Likewise.
9779 * config/nds32/nds32.h (nds32_builtins): Likewise.
9781 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
9783 * config/nds32/nds32.c (nds32_init_machine_status,
9784 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
9785 strict_aligned_p field.
9786 (nds32_expand_to_rtl_hook): New function.
9787 (TARGET_EXPAND_TO_RTL_HOOK): Define.
9788 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
9790 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
9791 Chung-Ju Wu <jasonwucj@gmail.com>
9793 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
9794 * config/nds32/nds32-n7.md: New file.
9795 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
9796 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
9798 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
9799 * config/nds32/nds32.md (pipeline_model): Add n7.
9800 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
9801 * config/nds32/pipelines.md: Include n7 settings.
9803 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
9804 Chung-Ju Wu <jasonwucj@gmail.com>
9806 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
9807 * config/nds32/nds32-e8.md: New file.
9808 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
9809 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
9811 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
9812 * config/nds32/nds32.md (pipeline_model): Add e8.
9813 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
9814 * config/nds32/pipelines.md: Include e8 settings.
9816 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
9817 Chung-Ju Wu <jasonwucj@gmail.com>
9819 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
9820 * config/nds32/nds32-n8.md: New file.
9821 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
9822 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
9824 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
9825 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
9826 * config/nds32/nds32.md (pipeline_model): Add n8.
9827 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
9828 * config/nds32/pipelines.md: Include n8 settings.
9830 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
9831 Chung-Ju Wu <jasonwucj@gmail.com>
9833 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
9834 * config/nds32/nds32-n9-2r1w.md: New file.
9835 * config/nds32/nds32-n9-3r2w.md: New file.
9836 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
9837 nds32_register_ports): New or modify for cpu n9.
9838 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
9840 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
9841 * config/nds32/nds32-utils.c: New file.
9842 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
9843 TARGET_MUL_SLOW): Define.
9844 * config/nds32/nds32.md (pipeline_model): New attribute.
9845 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
9846 New options that support cpu n9.
9847 * config/nds32/pipelines.md: Include n9 settings.
9848 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
9850 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
9852 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
9853 information if necessary.
9854 (output_cond_branch_compare_zero): Likewise.
9855 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
9856 (nds32_target_alignment): Refine for alignment.
9857 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
9858 (FUNCTION_BOUNDARY): Modify.
9859 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
9861 * config/nds32/nds32.opt (malways-align, malign-functions): New.
9863 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
9865 * config/nds32/constants.md (unspec_volatile_element): Add values for
9866 TLB operation and data prefetch.
9867 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
9868 functions for TLB operation and data prefetch.
9869 * config/nds32/nds32-intrinsic.md: Likewise.
9870 * config/nds32/nds32_intrinsic.h: Likewise.
9871 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
9872 (nds32_print_operand): Likewise.
9873 * config/nds32/nds32.h (nds32_builtins): Likewise.
9875 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
9876 Andrew Pinski <pinsika@gcc.gnu.org>
9879 * match.pd: Use constant_boolean_node of correct type instead of
9880 boolean_true_node or boolean_false_node for simplifying
9881 pointer comparisons to zero.
9883 2018-04-07 Jakub Jelinek <jakub@redhat.com>
9885 PR tree-optimization/80021
9886 * tree.c (verify_type_variant): Make error call in verify_variant_match
9887 translatable and remove final full stop.
9889 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
9891 * config/nds32/constants.md (unspec_volatile_element): Add
9892 UNSPEC_VOLATILE_EH_RETURN.
9893 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
9894 nds32_output_stack_pop): Support dwarf exception handling process.
9895 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
9896 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
9897 exception handling process.
9898 (nds32_compute_stack_frame): Likewise.
9899 (nds32_return_addr_rtx): Likewise.
9900 (nds32_initial_elimination_offset): Likewise.
9901 (nds32_expand_prologue): Likewise.
9902 (nds32_expand_epilogue): Likewise.
9903 (nds32_dynamic_chain_address): New function.
9904 * config/nds32/nds32.h (machine_function): Add fields for dwarf
9906 (DYNAMIC_CHAIN_ADDRESS): Define.
9907 (EH_RETURN_DATA_REGNO): Define.
9908 (EH_RETURN_STACKADJ_RTX): Define.
9909 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
9910 patterns for dwarf exception handling.
9912 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
9914 * config/nds32/nds32.h: Clean up obsolete macros.
9916 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
9918 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
9919 Add enum values for particular instructions.
9920 * config/nds32/nds32-intrinsic.c: Implementation of expanding
9921 particular intrinsic functions.
9922 * config/nds32/nds32-intrinsic.md: Likewise.
9923 * config/nds32/nds32_intrinsic.h: Likewise.
9924 * config/nds32/nds32.h (nds32_builtins): Likewise.
9925 * config/nds32/nds32.md (type): Add pbsad and pbsada.
9926 (btst, ave): New patterns for particular instructions.
9928 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
9930 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
9931 Add enum values for atomic load/store and memory sync.
9932 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
9934 * config/nds32/nds32-intrinsic.md: Likewise.
9935 * config/nds32/nds32_intrinsic.h: Likewise.
9936 * config/nds32/nds32.h (nds32_builtins): Likewise.
9938 2018-04-07 Jakub Jelinek <jakub@redhat.com>
9940 PR tree-optimization/85257
9941 * fold-const.c (native_encode_vector): If not all elts could fit
9942 and off is -1, return 0 rather than offset.
9943 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
9944 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
9945 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
9946 adjust buffer in native_interpret_expr call.
9948 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
9950 * config/nds32/constants.md (unspec_volatile_element): Add cache
9951 control enum values.
9952 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
9953 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
9954 * config/nds32/nds32.c (nds32_cctl_names): New.
9955 (nds32_print_operand): Handle cache control register names.
9956 * config/nds32/nds32.h (nds32_builtins): New enum values.
9957 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
9959 * config/nds32/nds32.md (type): Add mmu.
9960 * config/nds32/pipelines.md (simple_insn): Add mmu.
9962 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
9964 * config/nds32/nds32.md (type): Remove call.
9965 * config/nds32/pipelines.md (simple_insn): Likewise.
9967 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
9969 * config/nds32/constants.md (unspec_volatile_element): Add
9970 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
9971 UNSPEC_VOLATILE_FMFCFG.
9972 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
9973 description for fmfcfg and fmfcsr.
9974 (bdesc_1arg): Add fmtcsr.
9975 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
9976 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
9977 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
9978 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
9979 unspec_fmfcfg): New patterns.
9980 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
9981 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
9982 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
9983 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
9984 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
9985 __nds32__fmfcfg): Define.
9987 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
9989 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
9990 intrinsic register names.
9991 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
9992 intrinsic register enum values and macros.
9994 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
9996 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
9997 for load/store addressing form.
9998 (nds32_print_operand_address): Likewise.
10000 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
10003 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
10004 based on LABEL_REF. Remove useless assertion.
10005 (pic_address_needs_scratch): Fix formatting.
10006 (sparc_legitimize_pic_address): Minor tweaks.
10007 (sparc_delegitimize_address): Adjust assertion accordingly.
10008 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
10009 into symbolic_operand.
10010 (movsi_high_pic_label_ref): Likewise.
10011 (movsi_lo_sum_pic_label_ref): Likewise.
10012 (movdi_pic_label_ref): Likewise.
10013 (movdi_high_pic_label_ref): Likewise.
10014 (movdi_lo_sum_pic_label_ref): Likewise.
10016 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
10018 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
10019 custom LIB_SPEC setup.
10021 2018-04-06 Ruslan Bukin <br@bsdpad.com>
10022 Kito Cheng <kito.cheng@gmail.com>
10024 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
10025 * config/riscv/freebsd.h: New.
10027 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
10029 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
10030 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
10033 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
10034 Kito Cheng <kito.cheng@gmail.com>
10036 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
10037 nds32_output_call, nds32_symbol_binds_local_p): New functions.
10038 * config/nds32/nds32-protos.h (nds32_output_call,
10039 nds32_output_return): Declare.
10040 * config/nds32/nds32.md: Refine all the call and return patterns.
10042 2018-04-06 Jakub Jelinek <jakub@redhat.com>
10045 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
10046 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
10048 PR rtl-optimization/84872
10049 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
10050 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
10051 EDGE_CROSSING edge.
10053 2018-04-06 Tamar Christina <tamar.christina@arm.com>
10055 * expr.c (copy_blkmode_to_reg): Revert 254862.
10056 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
10058 2018-04-06 Richard Biener <rguenther@suse.de>
10060 PR middle-end/85244
10061 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
10062 after seeing a component reference with an adjacent field. Treat
10063 refs to arrays at struct end of external decls similar to
10064 refs to unconstrained commons.
10066 2018-04-06 Jakub Jelinek <jakub@redhat.com>
10069 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
10070 look through SAVE_EXPRs with non-side-effects argument. Adjust
10072 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
10075 2018-04-06 Richard Biener <rguenther@suse.de>
10077 PR middle-end/85180
10078 * alias.c (find_base_term): New wrapper around find_base_term
10079 unwinding CSELIB_VAL_PTR changes.
10080 (find_base_term): Do not restore CSELIB_VAL_PTR during the
10083 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10085 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
10087 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
10088 constant definitions.
10089 ("nop"): lr 0,0 -> nopr r0
10090 ("nop_lr0", "nop_lr1"): New insn definitions.
10092 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
10094 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
10095 NDS32_V3PUSH_AVAILABLE_P macro.
10097 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
10098 Chung-Ju Wu <jasonwucj@gmail.com>
10100 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
10101 (nds32*-*-*): Add float and fpu_config into supported_defaults.
10102 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
10103 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
10104 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
10105 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
10106 * config/nds32/constraints.md: New constraints and checking for hard
10107 float configuration.
10108 * config/nds32/iterators.md: New mode iterator and attribute for hard
10109 float configuration.
10110 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
10112 * config/nds32/nds32-fpu.md: New file.
10113 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
10114 deal with hard float code generation.
10115 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
10117 (abi_type, float_reg_number): New enum type.
10118 * config/nds32/nds32-predicates.c: New predicates for hard float.
10119 * config/nds32/nds32-protos.h: Declare functions for hard float.
10120 * config/nds32/nds32.c: Implementation for hard float configuration.
10121 * config/nds32/nds32.h: Definitions for hard float configuration.
10122 * config/nds32/nds32.md: Include hard float machine description and
10123 modify patterns for hard float configuration.
10124 * config/nds32/nds32.opt: New options for hard float configuration.
10125 * config/nds32/predicates.md: New predicates for hard float
10128 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
10130 * common/config/nds32/nds32-common.c
10131 (nds32_option_optimization_table): Enable -mreleax-hint by default.
10133 2018-04-05 Jakub Jelinek <jakub@redhat.com>
10135 PR middle-end/85195
10136 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
10137 CONSTRUCTOR_ELT (ctor, ...)->value.
10139 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
10142 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
10144 2018-04-05 Tom de Vries <tom@codesourcery.com>
10147 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
10150 2018-04-05 Shiva Chen <shiva0217@gmail.com>
10151 Kito Cheng <kito.cheng@gmail.com>
10153 * config/nds32/constraints.md (U33): Fine-tune checking condition.
10154 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
10155 * config/nds32/nds32.h (nds32_16bit_address_type): Add
10156 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
10158 2018-04-05 Shiva Chen <shiva0217@gmail.com>
10159 Kito Cheng <kito.cheng@gmail.com>
10161 * config/nds32/constraints.md (Ufe): New memory constraint.
10162 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
10163 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
10164 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
10166 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
10167 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
10169 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
10171 * config/nds32/nds32.md: Use optimize_size in the condition for
10172 alu-shift instructions.
10174 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
10176 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
10178 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
10180 * config/nds32/nds32.md (negsi2): Refine pattern.
10182 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
10183 Chung-Ju Wu <jasonwucj@gmail.com>
10185 * config/nds32/iterators.md (shift_rotate): New code iterator.
10186 (shift): New code attribute.
10187 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
10188 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
10189 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
10190 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
10191 bit-wise operations.
10192 (andsi3, *andsi3): Ditto.
10193 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
10194 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
10195 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
10196 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
10197 nds32_ior_operand, nds32_xor_operand): New predicates.
10199 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
10201 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
10202 (addsi3, subsi3): ... this.
10204 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
10206 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
10208 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
10210 * config/nds32/nds32.md: Adjust indention.
10212 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
10214 * config/nds32/nds32.md (feature): New attribute.
10216 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
10218 * config/nds32/nds32.md (subtype): New attribute.
10220 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
10223 * config/arm/arm-builtins.c (arm_expand_builtin): Change
10224 expansion to perform a bitwise AND of the argument followed by a
10225 boolean negation of the result.
10227 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
10229 PR rtl-optimization/84878
10230 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
10231 the basic block. Assert the use reference is not artificial and that
10232 it has an associated insn.
10234 2018-04-04 Michael Matz <matz@suse.de>
10236 * builtins.c (compute_objsize): Pass correct operand
10237 to array_at_struct_end_p.
10239 2018-04-04 Richard Biener <rguenther@suse.de>
10242 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
10243 from contexts for DINFO_LEVEL_TERSE and below.
10245 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
10247 * config/nds32/nds32-doubleword.md (move_<mode>): Require
10248 resiter_operand condition.
10249 * config/nds32/nds32.md (*move<mode>): Ditto.
10251 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
10252 Monk Chiang <sh.chiang04@gmail.com>
10254 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
10256 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
10258 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
10260 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
10261 Kito Cheng <kito.cheng@gmail.com>
10263 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
10264 nds32_cond_code_str, output_cond_branch,
10265 output_cond_branch_compare_zero, nds32_expand_cbranch,
10266 nds32_expand_cstore, nds32_expand_movcc,
10267 nds32_output_cbranchsi4_equality_zero,
10268 nds32_output_cbranchsi4_equality_reg,
10269 nds32_output_cbranchsi4_equality_reg_or_const_int,
10270 nds32_output_cbranchsi4_greater_less_zero: New functions.
10271 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
10272 nds32_expand_cstore, nds32_expand_movcc,
10273 nds32_output_cbranchsi4_equality_zero,
10274 nds32_output_cbranchsi4_equality_reg,
10275 nds32_output_cbranchsi4_equality_reg_or_const_int,
10276 nds32_output_cbranchsi4_greater_less_zero): Declare.
10277 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
10278 nds32_rimm11s_operand): New predicates.
10279 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
10280 * config/nds32/nds32.md: Rewrite all the branch and conditional move
10283 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
10285 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
10286 * config/nds32/nds32.md: Ditto.
10287 * config/nds32/pipelines.md: Ditto.
10289 2018-04-04 Richard Biener <rguenther@suse.de>
10291 PR tree-optimization/85168
10292 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
10293 propagating abnormals.
10295 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
10297 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
10299 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
10300 Kito Cheng <kito.cheng@gmail.com>
10302 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
10303 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
10304 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
10305 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
10306 * config/nds32/nds32.md (sibcall_internal): New.
10307 (sibcall_register): Remove.
10308 (sibcall_immediate): Remove.
10309 (sibcall_value_internal): New.
10310 (sibcall_value_register): Remove.
10311 (sibcall_value_immediate): Remove.
10312 * config/nds32/predicates.md (nds32_general_register_operand): New.
10313 (nds32_call_address_operand): New.
10315 2018-04-03 Jakub Jelinek <jakub@redhat.com>
10317 PR rtl-optimization/85167
10318 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
10319 bb_defs if *split_p, instead preinitialize it to NULL.
10321 PR tree-optimization/85156
10322 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
10323 evaluating the argument multiple times.
10325 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
10327 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
10329 (_mm_cvtpd_ps): Likewise.
10330 (_mm_cvttpd_epi32): Likewise.
10331 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
10332 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
10333 vector, pixel, and bool following altivec.h include.
10335 2018-04-03 Martin Sebor <msebor@redhat.com>
10337 * doc/extend.texi (Common Function Attributes): Clarify.
10338 (const attribute): Likewise.
10339 (pure attribute): Likewise.
10341 2018-04-03 Jakub Jelinek <jakub@redhat.com>
10344 * config/i386/i386.c (ix86_expand_vector_set): Use
10345 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
10347 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
10349 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
10350 instructions when changing rounding bits to preserve precision bits
10351 in the x87 control word.
10353 2018-04-03 Martin Liska <mliska@suse.cz>
10355 PR tree-optimization/82491
10356 * rtl.h (strip_offset_and_add): Replace += suboffset with
10357 poly_uint64 () + suboffset.
10359 2018-03-29 Martin Liska <mliska@suse.cz>
10360 Martin Jambor <mjambor@suse.cz>
10363 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
10364 param_type is not an integral or pointer type.
10366 2018-04-03 Richard Biener <rguenther@suse.de>
10368 * sese.h (recompute_all_dominators): Remove.
10370 2018-04-02 Martin Sebor <msebor@redhat.com>
10372 * doc/invoke.texi (-Wrestrict): Fix typos.
10374 2018-04-02 Jim Wilson <jimw@sifive.com>
10376 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
10377 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
10378 (<optab>di3, <optab>si3_extend): Likewise.
10379 (<optab>si3_mask, <optab>si3_mask_1): New.
10380 (<optab>di3_mask, <optab>di3_mask_1): New.
10381 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
10382 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
10383 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
10385 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
10387 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
10390 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
10392 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
10393 (nds32_canonicalize_comparison): New function.
10395 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
10396 Kito Cheng <kito.cheng@gmail.com>
10397 Kuan-Lin Chen <kuanlinchentw@gmail.com>
10399 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
10400 * config/nds32/constants.md (unspec_volatile_element): Add
10401 UNSPEC_VOLATILE_RELAX_GROUP.
10402 * config/nds32/nds32-relax-opt.c: New file.
10403 * config/nds32/nds32-predicates.c
10404 (nds32_symbol_load_store_p): New function.
10405 * config/nds32/nds32-protos.h
10406 (nds32_symbol_load_store_p): Declare function.
10407 (make_pass_nds32_relax_opt): Declare new rtl pass function.
10408 * config/nds32/nds32.c
10409 (nds32_register_pass): New function to register pass.
10410 (nds32_register_passes): New function to register passes.
10411 * config/nds32/nds32.md (relax_group): New pattern.
10412 * config/nds32/nds32.opt (mrelax-hint): New option.
10413 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
10415 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
10417 * config/nds32/t-nds32: Modify files dependency.
10419 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
10421 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
10422 (PROFILE_HOOK): Define its implementation.
10424 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
10426 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
10427 type and 32-bit size.
10429 2018-04-01 Jakub Jelinek <jakub@redhat.com>
10431 PR middle-end/85090
10432 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
10433 (V_128_256): New mode iterator.
10434 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
10435 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
10436 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
10438 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
10439 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
10441 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
10444 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
10445 NaN inputs correctly.
10447 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
10450 * config/rs6000/vsx.md (??r): New mode attribute.
10451 (*vsx_mov<mode>_64bit): Use it.
10452 (*vsx_mov<mode>_32bit): Likewise.
10454 2018-03-30 Martin Sebor <msebor@redhat.com>
10456 PR tree-optimization/84818
10457 * builtins.c (check_access): Use warning_n.
10459 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10462 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
10464 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
10467 2018-03-30 Julia Koval <julia.koval@intel.com>
10470 * x86-tune.def (movx, partial_reg_dependency): Enable for
10473 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
10475 PR inline-asm/84985
10476 * lra-constraints.c (process_alt_operands): Move setting
10477 this_alternative_matches below.
10479 2018-03-29 Martin Liska <mliska@suse.cz>
10482 * doc/invoke.texi: Document how LTO works with debug info.
10483 Describe auto-load support of binutils. Mention 'x86-64'
10484 as valid option value of -march option.
10486 2018-03-29 Jakub Jelinek <jakub@redhat.com>
10488 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
10491 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
10492 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
10493 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
10496 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
10499 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
10500 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
10501 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
10502 for RS6000_BTM_POWERPC64.
10503 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
10504 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
10505 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
10508 (DIVDEU): Likewise.
10510 2018-03-28 Carl Love <cel@us.ibm.com>
10513 2017-09-27 Carl Love <cel@us.ibm.com>
10515 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
10516 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
10517 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
10520 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10522 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
10523 instead of __vector bool.
10524 (_mm_max_pu8): Likewise.
10525 (_mm_min_pi16): Likewise.
10527 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
10530 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
10531 (DIVWEUO): Likewise.
10532 (DIVDEO): Likewise.
10533 (DIVDEUO): Likewise.
10534 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
10535 DIVWEUO and DIVDEUO.
10536 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
10537 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
10538 (div_extend): Likewise.
10539 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
10541 (__builtin_divweuo): Likewise.
10542 (__builtin_divdeo): Likewise.
10543 (__builtin_divdeuo): Likewise.
10545 2018-03-28 Jakub Jelinek <jakub@redhat.com>
10548 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
10549 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
10551 PR tree-optimization/82004
10552 * gimple-match-head.c (optimize_pow_to_exp): New function.
10553 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
10554 Don't fold to exp if optimize_pow_to_exp is false.
10556 2018-03-28 Martin Liska <mliska@suse.cz>
10559 * calls.c (initialize_argument_information): Fix trailing space.
10560 * common.opt: Fix typo and provide better explanation for
10561 -fsanitize-coverage option.
10562 * config/i386/i386.opt: Fix typo.
10564 2018-03-28 Jakub Jelinek <jakub@redhat.com>
10565 Martin Liska <mliska@suse.cz>
10568 * gimplify.c (asan_poison_variable): Don't do the check for
10569 gimplify_omp_ctxp here.
10570 (gimplify_decl_expr): Do it here.
10571 (gimplify_target_expr): Likewise.
10573 2018-03-28 Martin Liska <mliska@suse.cz>
10576 * config/i386/i386.c (ix86_function_arg_advance): Do not call
10577 chkp_type_bounds_count if MPX is not enabled.
10579 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
10581 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
10583 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
10586 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
10587 function to create the function decl for complex long double
10588 multiply and divide for -mabi=ieeelongdouble.
10589 (init_float128_ieee): Call it.
10591 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
10594 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
10595 -fcf-protection=branch -mibt.
10596 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
10598 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10601 * config/arm/arm.c (arm_valid_symbolic_address): Handle
10602 arm_word_relocations.
10604 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
10607 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
10608 extern array declarations.
10610 2018-03-27 Richard Biener <rguenther@suse.de>
10612 PR middle-end/84067
10613 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
10614 explicit single_use checks.
10616 2018-03-27 Richard Biener <rguenther@suse.de>
10618 PR tree-optimization/85082
10619 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
10622 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10624 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
10625 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
10626 Turn on fasynchronous-unwind-tables and funwind-tables.
10628 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
10631 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
10632 (*bmi_blsr_<mode>_ccz): Ditto.
10634 2018-03-26 Tom de Vries <tom@codesourcery.com>
10636 PR tree-optimization/85063
10637 * omp-general.c (offloading_function_p): New function. Factor out
10639 * omp-offload.c (pass_omp_target_link::gate): ... here.
10640 * omp-general.h (offloading_function_p): Declare.
10641 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
10642 with attribute omp declare target for offloading functions.
10644 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
10646 PR tree-optimization/84005
10647 * tree-data-ref.h (get_base_for_alignment): Declare.
10648 * tree-data-ref.c (get_base_for_alignment_1): New function.
10649 (get_base_for_alignment): Likewise.
10650 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
10651 get_base_for_alignment to find a suitable base object, instead
10652 of always using drb->base_address.
10654 2018-03-23 Jakub Jelinek <jakub@redhat.com>
10656 PR inline-asm/85022
10657 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
10658 known size by default.
10660 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
10662 PR inline-asm/85030
10663 * lra-constraints.c (process_alt_operands): Don't match BLKmode
10664 and non BLKmode operands.
10666 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10669 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
10670 Clean up attributes.
10672 2018-03-23 Richard Biener <rguenther@suse.de>
10675 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
10676 we are going to emit early debug for LTO.
10678 2018-03-23 Jakub Jelinek <jakub@redhat.com>
10680 PR inline-asm/85034
10681 * function.c (match_asm_constraints_1): Don't optimize if input
10682 doesn't satisfy general_operand predicate for output's mode.
10684 PR inline-asm/85022
10685 * alias.c (write_dependence_p): Don't require for x_canonicalized
10686 non-VOIDmode if x has VOIDmode.
10689 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
10690 just don't try to optimize it rather than assert it never happens.
10692 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
10694 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
10695 macro expansions for definition of ST_INTERNAL_<mode> and
10696 LD_INTERNAL_<mode> builtins.
10697 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
10699 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
10701 (altivec_expand_st_builtin): Likewise.
10702 (altivec_expand_builtin): Remove calls to deleted functions.
10703 (rs6000_address_for_altivec): Delete this function.
10704 * config/rs6000/vector.md: Remove expands for
10705 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
10707 2018-03-22 Sudakshina Das <sudi.das@arm.com>
10710 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
10711 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
10712 re-computing once computed.
10713 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
10714 (arm_init_machine_status): Initialize
10715 machine->static_chain_stack_bytes.
10717 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
10720 * doc/extend.texi: Add four new prototypes for vec_ld.
10721 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
10722 definitions for more logical presentation.
10723 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
10724 entries for V1TI variants of __builtin_altivec_ld builtin.
10725 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
10726 handling of V1TI variant of LVX icode pattern.
10727 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
10728 (rs6000_gimple_fold_builtin): Likewise.
10729 (altivec_init_builtins): Add code to define
10730 __builtin_altivec_lvx_v1ti function.
10732 2018-03-22 Jakub Jelinek <jakub@redhat.com>
10734 PR inline-asm/84941
10735 * function.c (match_asm_constraints_1): Don't do the optimization
10736 if input isn't a REG, SUBREG, MEM or constant.
10738 2018-03-22 Tom de Vries <tom@codesourcery.com>
10740 PR tree-optimization/84956
10741 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
10742 bb_has_abnormal_pred.
10744 2018-03-22 Jakub Jelinek <jakub@redhat.com>
10747 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
10748 DECL_INITIAL (decl) to decl at the end.
10749 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
10750 adjust the comment.
10752 2018-03-21 Joseph Myers <joseph@codesourcery.com>
10754 * doc/extend.texi (__builtin_tgmath): Document when complex
10755 integer types are treated as _Complex _Float64.
10757 2018-03-21 Tom de Vries <tom@codesourcery.com>
10759 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
10761 2018-03-21 Jakub Jelinek <jakub@redhat.com>
10763 PR tree-optimization/84960
10764 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
10765 if it is ENTRY block, move them into single succ of ENTRY in that case.
10767 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
10769 PR tree-optimization/84811
10770 * poly-int.h (poly_span_traits): Remove the T3 parameter and
10771 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
10772 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
10773 (known_subrange_p): Update accordingly. Cast each value involved
10774 in the size comparison, rather than casting the result of the
10777 2018-03-21 Jakub Jelinek <jakub@redhat.com>
10779 PR tree-optimization/84982
10780 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
10781 by flipping the least significant bit rather than all bits from
10782 bitpos to bitpos + bitsize - 1.
10784 2018-03-21 Nathan Sidwell <nathan@acm.org>
10786 * doc/extend.texi (Deprecated Features): Remove mention of
10787 long-deleted deprecations.
10789 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10792 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
10793 * configure: Regenerate.
10795 2018-03-21 Tom de Vries <tom@codesourcery.com>
10797 PR tree-optimization/83126
10798 * tree-parloops.c (num_phis): New function.
10799 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
10801 2018-03-21 Nathan Sidwell <nathan@acm.org>
10803 * doc/extend.texi (Deprecated Features): Update deprecated flags,
10804 mention anon-struct/union members and trailing attributes.
10806 2018-03-21 Bin Cheng <bin.cheng@arm.com>
10808 PR tree-optimization/84969
10809 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
10810 builtin memset partitions if they set different rhs values.
10812 2018-03-21 Jakub Jelinek <jakub@redhat.com>
10814 PR rtl-optimization/84989
10815 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
10816 VEC_DUPLICATE with scalar result mode.
10818 2018-03-21 Martin Liska <mliska@suse.cz>
10821 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
10822 not intended return statement.
10824 2018-03-21 Martin Liska <mliska@suse.cz>
10827 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
10828 (chkp_find_bound_slots_1): Limit number of iterations.
10830 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
10833 * Minor grammar fixes for x86 options.
10835 2018-03-20 Jakub Jelinek <jakub@redhat.com>
10838 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
10839 holding REG_CFA_RESTORE notes, instead turn them into a USE.
10841 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
10844 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
10845 (altivec_lvx_<mode>_1op): Likewise.
10846 (altivec_stvx_<mode>_2op): Likewise.
10847 (altivec_stvx_<mode>_1op): Likewise.
10848 (altivec_lvx_<VM2:mode>): New define_expand.
10849 (altivec_stvx_<VM2:mode>): Likewise.
10850 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
10851 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
10852 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
10853 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
10854 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
10855 (rs6000_gen_lvx): Likewise.
10856 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
10857 (altivec_expand_stv_builtin): Likewise.
10858 (altivec_expand_builtin): Likewise.
10859 * config/rs6000/vector.md: Likewise.
10861 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10864 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
10867 2018-03-20 Richard Biener <rguenther@suse.de>
10870 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
10871 sign-conversions as zero, fall back to standard scalar_stmt
10874 2018-03-20 Martin Liska <mliska@suse.cz>
10877 * predict.c (rebuild_frequencies): Handle case when we have
10878 PROFILE_ABSENT, but flag_guess_branch_prob is false.
10880 2018-03-20 Jakub Jelinek <jakub@redhat.com>
10883 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
10884 flag_section_anchors.
10885 * varasm.c (use_blocks_for_decl_p): Remove hack for
10886 dw2_force_const_mem.
10889 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
10891 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
10892 be created, use lowpart_subreg of operands[0] rather than operands[0]
10894 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
10895 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
10896 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
10897 and n constraint instead of aarch64_shift_imm_di and Usd.
10898 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
10899 (*aarch64_<optab>_reg_minus<mode>3): ... this.
10901 2018-03-20 Sudakshina Das <sudi.das@arm.com>
10904 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
10905 to favor GPR over NEON registers.
10906 (<shift>di3_neon): Likewise.
10908 2018-03-20 Tom de Vries <tom@codesourcery.com>
10911 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
10912 (nvptx_process_pars): Emit bar.sync asap and alap.
10914 2018-03-20 Tom de Vries <tom@codesourcery.com>
10917 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
10918 seen_label if seen_label is already set.
10920 2018-03-20 Jakub Jelinek <jakub@redhat.com>
10923 * config/i386/i386.c (fold_builtin_cpu): For features above 31
10924 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
10925 Use 1U instead of 1. Formatting fixes.
10928 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
10929 instead of TREE_TYPE (s1) for the return value.
10931 2018-03-19 Jakub Jelinek <jakub@redhat.com>
10933 PR tree-optimization/84946
10934 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
10935 bitsize + bitsize in poly_uint64 rather than poly_int64.
10938 * dwarf2asm.c: Include fold-const.c.
10939 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
10940 of decl rather than decl itself.
10942 PR rtl-optimization/84643
10943 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
10945 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
10948 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
10949 calling assemble_variable.
10951 2018-03-19 Sudakshina Das <sudi.das@arm.com>
10954 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
10955 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
10957 2018-03-19 Jim Wilson <jimw@sifive.com>
10960 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
10961 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
10962 (riscv_first_stack_step): Likewise.
10963 (riscv_option_override): Use STACK_BOUNDARY instead of
10964 MIN_STACK_BOUNDARY.
10965 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
10966 MIN_STACK_BOUNDARY.
10967 (BIGGEST_ALIGNMENT): Set to 128.
10968 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
10969 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
10972 2018-03-19 Richard Biener <rguenther@suse.de>
10974 PR tree-optimization/84933
10975 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
10976 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
10978 2018-03-19 Richard Biener <rguenther@suse.de>
10980 PR tree-optimization/84859
10981 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
10982 (cond_if_else_store_replacement): Perform sinking operation on
10983 single-store BBs regardless of MAX_STORES_TO_SINK setting.
10984 Generalize what a BB with a single eligible store is.
10986 2018-03-19 Richard Biener <rguenther@suse.de>
10988 PR tree-optimization/84929
10989 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
10990 chrec_is_positive against non-chrec arg.
10992 2018-03-19 Tamar Christina <tamar.christina@arm.com>
10995 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
10997 2018-03-18 Martin Liska <mliska@suse.cz>
10999 PR rtl-optimization/84635
11000 * regrename.c (build_def_use): Use matches_mode only when
11003 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
11005 PR tree-optimization/84913
11006 * tree-vect-loop.c (vectorizable_reduction): Don't try to
11007 vectorize chains of COND_EXPRs.
11009 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
11011 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
11013 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
11015 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
11017 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
11019 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
11021 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
11022 Kito Cheng <kito.cheng@gmail.com>
11024 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
11025 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
11026 (nds32_adjust_reg_alloc_order): New function.
11027 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
11029 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
11031 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
11032 nds32_print_operand, nds32_print_operand_address): Use
11033 HOST_WIDE_INT_PRINT_DEC instead.
11035 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
11037 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
11039 2018-03-17 Jakub Jelinek <jakub@redhat.com>
11042 * config/i386/i386.c (initial_ix86_tune_features,
11043 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
11044 unsigned long long.
11045 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
11046 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
11047 rather than 1u << ix86_tune. Formatting fix.
11048 (ix86_option_override_internal): Change ix86_arch_mask from
11049 unsigned int to unsigned HOST_WIDE_INT, initialize to
11050 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
11051 (ix86_function_specific_restore): Likewise.
11053 2018-03-16 Jakub Jelinek <jakub@redhat.com>
11056 * postreload.c (reload_combine_recognize_pattern): Perform
11057 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
11058 truncate_int_for_mode the result for the destination's mode.
11061 * hsa-gen.c (mem_type_for_type): Fix comment typo.
11062 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
11064 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
11067 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
11070 * lra-assigns.c (lra_split_hard_reg_for): Don't use
11071 regno_allocno_class_array and sorted_pseudos.
11072 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
11073 insns where regno is used.
11075 2018-03-16 Martin Liska <mliska@suse.cz>
11078 * multiple_target.c (create_dispatcher_calls): Redirect
11079 reference in the symbol table.
11081 2018-03-16 Martin Liska <mliska@suse.cz>
11084 * multiple_target.c (create_dispatcher_calls): Redirect also
11087 2018-03-16 Jakub Jelinek <jakub@redhat.com>
11091 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
11092 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
11093 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
11095 2018-03-16 Julia Koval <julia.koval@intel.com>
11097 * doc/invoke.texi (Skylake Server): Add CLWB.
11098 Cannonlake): Remove CLWB.
11100 2018-03-16 Jakub Jelinek <jakub@redhat.com>
11102 PR tree-optimization/84841
11103 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
11105 (FLOAT_ONE_CONST_TYPE): Define.
11106 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
11107 (sort_by_operand_rank): Put entries with higher constant_type last
11108 rather than first to match comments.
11110 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
11112 * config/nios2/nios2.md (movsi_internal): Fix thinko in
11115 2018-03-15 Jakub Jelinek <jakub@redhat.com>
11118 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
11119 check and use address of target always.
11121 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
11124 * config/i386/i386.c (indirect_thunk_needed): Update comments.
11125 (indirect_thunk_bnd_needed): Likewise.
11126 (indirect_thunks_used): Likewise.
11127 (indirect_thunks_bnd_used): Likewise.
11128 (indirect_return_needed): New.
11129 (indirect_return_bnd_needed): Likewise.
11130 (output_indirect_thunk_function): Add a bool argument for
11132 (output_indirect_thunk_function): Don't generate alias for
11133 function return thunk.
11134 (ix86_code_end): Call output_indirect_thunk_function to generate
11135 function return thunks.
11136 (ix86_output_function_return): Set indirect_return_bnd_needed
11137 and indirect_return_needed instead of indirect_thunk_bnd_needed
11138 and indirect_thunk_needed.
11140 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
11142 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
11143 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
11144 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
11146 2018-03-15 David Malcolm <dmalcolm@redhat.com>
11147 Paul Hua <paul.hua.gm@gmail.com>
11150 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
11152 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
11154 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
11155 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
11156 resp. SFmode cases.
11158 2018-03-15 Tamar Christina <tamar.christina@arm.com>
11161 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
11162 instead of GET_MODE_SIZE when comparing Units.
11164 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
11167 * varasm.c (hash_section): Return an unchangeble hash value
11168 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
11169 Return !aarch64_can_use_per_function_literal_pools_p ().
11171 2018-03-15 Jakub Jelinek <jakub@redhat.com>
11174 * optabs.c (emit_conditional_move): Pass address of cmode's copy
11175 rather than address of cmode as last argument to prepare_cmp_insn.
11177 2018-03-15 Julia Koval <julia.koval@intel.com>
11179 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
11180 F_AVX512VNNI, F_AVX512BITALG): New.
11182 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
11185 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
11186 insn for floating-point loads and stores.
11188 2018-03-14 Carl Love <cel@us.ibm.com>
11190 * config/rs6000/rs6000-c.c: Add macro definitions for
11191 ALTIVEC_BUILTIN_VEC_PERMXOR.
11192 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
11193 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
11194 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
11195 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
11197 * config/doc/extend.texi: Add prototypes for vec_permxor.
11199 2018-03-14 David Malcolm <dmalcolm@redhat.com>
11202 * diagnostic-show-locus.c (class layout_point): Convert m_line
11203 from int to linenum_type.
11204 (line_span::comparator): Use linenum "compare" function when
11205 comparing line numbers.
11206 (test_line_span): New function.
11207 (layout_range::contains_point): Convert param "row" from int to
11209 (layout_range::intersects_line_p): Likewise.
11210 (layout::will_show_line_p): Likewise.
11211 (layout::print_source_line): Likewise.
11212 (layout::should_print_annotation_line_p): Likewise.
11213 (layout::print_annotation_line): Likewise.
11214 (layout::print_leading_fixits): Likewise.
11215 (layout::annotation_line_showed_range_p): Likewise.
11216 (struct line_corrections): Likewise for field m_row.
11217 (line_corrections::line_corrections): Likewise for param "row".
11218 (layout::print_trailing_fixits): Likewise.
11219 (layout::get_state_at_point): Likewise.
11220 (layout::get_x_bound_for_row): Likewise.
11221 (layout::print_line): Likewise.
11222 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
11223 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
11224 * input.c (selftest::test_linenum_comparisons): New function.
11225 (selftest::input_c_tests): Call it.
11226 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
11227 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
11228 * selftest.h (ASSERT_GT): New macro.
11229 (ASSERT_GT_AT): New macro.
11230 (ASSERT_LT): New macro.
11231 (ASSERT_LT_AT): New macro.
11233 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
11235 PR rtl-optimization/84780
11236 * combine.c (distribute_links): Don't make a link based on pc_rtx.
11238 2018-03-14 Martin Liska <mliska@suse.cz>
11240 * tree.c (record_node_allocation_statistics): Use
11241 get_stats_node_kind.
11242 (get_stats_node_kind): New function extracted from
11243 record_node_allocation_statistics.
11244 (free_node): Use get_stats_node_kind.
11246 2018-03-14 Richard Biener <rguenther@suse.de>
11248 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
11249 that the value-set of ANTIC_IN doesn't grow.
11252 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
11254 (BB_VISITED_WITH_VISITED_SUCCS): New define.
11255 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
11257 2018-03-14 Julia Koval <julia.koval@intel.com>
11259 * config.gcc (icelake-client, icelake-server): New.
11261 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
11262 (initial_ix86_arch_features): Ditto.
11263 (PTA_SKYLAKE): Add SGX.
11264 (PTA_ICELAKE): Remove.
11265 (PTA_ICELAKE_CLIENT): New.
11266 (PTA_ICELAKE_SERVER): New.
11267 (ix86_option_override_internal): Split up icelake on icelake client and
11269 (get_builtin_code_for_version): Ditto.
11270 (fold_builtin_cpu): Ditto.
11271 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
11272 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
11273 * config/i386/i386.h (processor_type): Ditto.
11274 * doc/invoke.texi: Ditto.
11276 2018-03-14 Jakub Jelinek <jakub@redhat.com>
11279 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
11280 INTEGER_CST offset, add it together with bitpos / 8 and
11281 sign extend based on POINTER_SIZE.
11285 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
11288 * config/i386/constraints.md (Yc): New register constraint.
11289 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
11290 Use Yc constraint for alternative 2 of operand 0. Remove
11291 preferred_for_speed attribute.
11293 2018-03-14 Richard Biener <rguenther@suse.de>
11295 PR tree-optimization/84830
11296 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
11297 with the old one to avoid oscillations.
11299 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
11302 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
11304 (assign_by_spills): Return a flag of reload assignment failure.
11305 Do not process the reload assignment failures. Do not spill other
11306 reload pseudos if they has the same reg class. Update n if
11308 (lra_assign): Add a return arg. Set up from the result of
11309 assign_by_spills call.
11310 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
11311 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
11312 usage_insns if it is not NULL.
11313 (spill_hard_reg_in_range): New function.
11314 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
11315 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
11316 function prototypes.
11317 (lra_assign): Change prototype.
11318 * lra.c (lra): Add code to deal with fails by splitting hard reg
11321 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
11323 * config/riscv/riscv.opt (mrelax): New option.
11324 * config/riscv/riscv.c (riscv_file_start): Emit ".option
11325 "norelax" when riscv_mrelax is disabled.
11326 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
11328 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11331 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
11332 reassociation for int modes.
11334 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
11336 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
11337 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
11339 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
11340 * config/aarch64/aarch64-sve.md
11341 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
11342 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
11343 (*extend<mode><Vwide>2): Rename to...
11344 (aarch64_sve_extend<mode><Vwide>2): ...this.
11345 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
11346 renaming the old pattern to...
11347 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
11349 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
11350 define_expand, renaming the old pattern to...
11351 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
11352 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
11353 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
11354 account when deciding which SVE instruction the optab should use.
11355 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
11357 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
11359 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
11360 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
11361 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
11362 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
11363 (tlsdesc_small_<mode>): Turn a define_expand and use
11364 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
11365 (tlsdesc_small_advsimd_<mode>): ...this.
11366 (tlsdesc_small_sve_<mode>): New pattern.
11368 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
11370 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
11371 (UNSPEC_UMUL_HIGHPART): New constants.
11372 (MUL_HIGHPART): New int iteraor.
11373 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
11374 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
11376 (*<su>mul<mode>3_highpart): New define_insn.
11378 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
11381 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
11384 2018-03-13 Martin Liska <mliska@suse.cz>
11387 * (sem_item_optimizer::sem_item_optimizer): Initialize new
11389 (sem_item_optimizer::~sem_item_optimizer): Release it.
11390 (sem_item_optimizer::merge_classes): Register variable aliases.
11391 (sem_item_optimizer::fixup_pt_set): New function.
11392 (sem_item_optimizer::fixup_points_to_sets): Likewise.
11393 * ipa-icf.h: Declare new variables and functions.
11395 2018-03-13 Jakub Jelinek <jakub@redhat.com>
11397 PR middle-end/84834
11398 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
11399 integer_pow2p@2 and test integer_pow2p in condition.
11400 (A < 0 ? C : 0): Similarly for @1.
11402 PR middle-end/84831
11403 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
11404 characters starting at p contain '\0' character, don't look beyond
11408 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
11409 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
11412 * reg-stack.c (change_stack): Change update_end var from int to
11413 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
11414 also call set_block_for_insn on the newly added insns and rescan.
11417 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
11418 on the last operand.
11421 * tree.c (stabilize_reference_1): Return save_expr (e) for
11422 STATEMENT_LIST even if it doesn't have side-effects.
11424 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
11426 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
11428 2018-03-12 Renlin Li <renlin.li@arm.com>
11430 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
11431 aarch64_output_scalar_simd_mov_immediate.
11433 2018-03-12 Martin Sebor <msebor@redhat.com>
11435 PR tree-optimization/83456
11436 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
11437 for perfectly overlapping calls to memcpy.
11438 (gimple_fold_builtin_memory_chk): Same.
11439 (gimple_fold_builtin_strcpy): Handle no-warning.
11440 (gimple_fold_builtin_stxcpy_chk): Same.
11441 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
11443 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
11445 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
11446 parameter. Use it for SFmode.
11447 (rs6000_function_arg_advance_1): Adjust.
11448 (rs6000_function_arg): Adjust.
11449 (rs6000_gimplify_va_arg): Pass false for that new parameter.
11451 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
11453 PR rtl-optimization/84169
11454 PR rtl-optimization/84780
11455 * combine.c (can_combine_p): Check for a 2-insn combination whether
11456 the destination register is used between the two insns, too.
11458 2018-03-12 Richard Biener <rguenther@suse.de>
11460 PR tree-optimization/84803
11461 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
11462 for refs DR analysis didn't process.
11464 2018-03-12 Richard Biener <rguenther@suse.de>
11466 PR tree-optimization/84777
11467 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
11468 force-vectorize loops ignore whether we are optimizing for size.
11470 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
11472 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
11473 (TARGET_MD_ASM_ADJUST): Define.
11475 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
11476 Kito Cheng <kito.cheng@gmail.com>
11477 Chung-Ju Wu <jasonwucj@gmail.com>
11479 * config/nds32/nds32.c (nds32_compute_stack_frame,
11480 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
11481 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
11482 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
11483 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
11484 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
11485 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
11486 * config/nds32/nds32.md (prologue, epilogue): Use macro
11487 NDS32_V3PUSH_AVAILABLE_P to do checking.
11489 2018-03-11 Jakub Jelinek <jakub@redhat.com>
11492 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
11493 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
11494 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
11495 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
11496 addition of most attributes on !orig_type_die or the attribute not
11497 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
11499 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
11500 Chung-Ju Wu <jasonwucj@gmail.com>
11502 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
11503 __NDS32_VH__ macro.
11504 * config/nds32/nds32.opt (mvh): New option.
11506 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
11507 Chung-Ju Wu <jasonwucj@gmail.com>
11509 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
11511 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
11512 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
11515 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
11516 Chung-Ju Wu <jasonwucj@gmail.com>
11518 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
11520 * config/nds32/nds32-multiple.md (strlensi): New pattern.
11521 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
11523 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
11524 Kito Cheng <kito.cheng@gmail.com>
11525 Chung-Ju Wu <jasonwucj@gmail.com>
11527 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
11528 UNSPEC_FFMISM and UNSPEC_FLMISM.
11529 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
11530 for ffb, ffmism and flmism.
11531 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
11532 (unspec_ffmism): Ditto.
11533 (unspec_flmism): Ditto.
11534 (nds32_expand_builtin_impl): Check if string extension is available.
11535 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
11536 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
11538 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
11541 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
11544 * lra-assigns.c (assign_by_spills): Return a flag of reload
11545 assignment failure. Do not process the reload assignment
11546 failures. Do not spill other reload pseudos if they has the same
11548 (lra_assign): Add a return arg. Set up from the result of
11549 assign_by_spills call.
11550 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
11551 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
11552 usage_insns if it is not NULL.
11553 (spill_hard_reg_in_range): New function.
11554 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
11555 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
11556 function prototypes.
11557 (lra_assign): Change prototype.
11558 * lra.c (lra): Add code to deal with fails by splitting hard reg
11561 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
11564 * config/i386/i386.opt: Replace Enforcment with Enforcement.
11566 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
11569 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
11570 (dw_val_node): Add val_symbolic_view.
11571 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
11572 (symview_upper_bound): New.
11573 (new_line_info_table): Initialize symviews_since_reset.
11574 (dwarf2out_source_line): Count symviews_since_reset and set
11575 symview_upper_bound.
11576 (dw_val_equal_p): Handle symview.
11577 (add_AT_symview): New.
11578 (print_dw_val): Handle symview.
11579 (attr_checksum, attr_checksum_ordered): Likewise.
11580 (same_dw_val_p, size_of_die): Likewise.
11581 (value_format, output_die): Likewise.
11582 (add_high_low_attributes): Use add_AT_symview for entry_view.
11583 (dwarf2out_finish): Reset symview_upper_bound, clear
11586 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
11589 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
11590 Add strict argument and use it.
11591 (rs6000_split_multireg_move): Update for new strict argument.
11592 (mem_operand_gpr): Disallow all non-offsettable addresses.
11593 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
11595 2018-03-09 Jakub Jelinek <jakub@redhat.com>
11598 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
11599 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
11600 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
11603 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
11604 decl, use remap_type if we want to use the type.
11606 2018-03-09 Martin Sebor <msebor@redhat.com>
11608 PR tree-optimization/84526
11609 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
11611 (builtin_access::generic_overlap): Be prepared to handle non-array
11614 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
11616 PR rtl-optimization/84682
11617 * lra-constraints.c (process_address_1): Check is_address flag
11618 for address constraints.
11619 (process_alt_operands): Likewise.
11620 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
11621 preprocess_constraints.
11622 * recog.h (preprocess_constraints): Add oploc parameter.
11624 * recog.c (preprocess_constraints): Test address_operand for
11625 CT_ADDRESS constraints.
11627 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
11630 * lra-assigns.c (assign_by_spills): Return a flag of reload
11631 assignment failure. Do not process the reload assignment
11632 failures. Do not spill other reload pseudos if they has the same
11634 (lra_assign): Add a return arg. Set up from the result of
11635 assign_by_spills call.
11636 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
11637 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
11638 usage_insns if it is not NULL.
11639 (spill_hard_reg_in_range): New function.
11640 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
11641 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
11642 function prototypes.
11643 (lra_assign): Change prototype.
11644 * lra.c (lra): Add code to deal with fails by splitting hard reg
11647 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11650 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
11651 Accept complain bool parameter. Only emit errors if it is true.
11652 (arm_parse_cpu_option_name): Likewise.
11653 (arm_target_thumb_only): Adjust callers of the above.
11654 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
11655 prototype to take a default true bool parameter.
11656 (arm_parse_arch_option_name): Likewise.
11658 2018-03-09 David Malcolm <dmalcolm@redhat.com>
11659 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
11663 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
11664 * configure: Regenerate.
11665 * configure.ac ("linker --version-script option"): New.
11666 ("linker soname option"): New.
11668 2018-03-09 Richard Biener <rguenther@suse.de>
11670 PR tree-optimization/84775
11671 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
11672 immediate uses of predicate stmts and mark them modified.
11675 PR tree-optimization/84178
11676 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
11678 (version_loop_for_if_conversion): Delay update_ssa call.
11679 (tree_if_conversion): Delay update_ssa until after predicate
11682 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
11685 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
11686 when the function accesses prior frames.
11688 2018-03-08 Jakub Jelinek <jakub@redhat.com>
11691 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
11692 gen_llsym, otherwise call maybe_gen_llsym.
11694 PR inline-asm/84742
11695 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
11696 has ',' character inside of it.
11698 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11701 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
11702 as clobbering CC_REGNUM.
11704 2018-03-08 Richard Biener <rguenther@suse.de>
11706 PR middle-end/84552
11707 * tree-scalar-evolution.c: Include tree-into-ssa.h.
11708 (follow_copies_to_constant): Do not follow SSA names registered
11711 2018-03-08 Richard Biener <rguenther@suse.de>
11713 PR tree-optimization/84178
11714 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
11716 (version_loop_for_if_conversion): Delay update_ssa call.
11717 (tree_if_conversion): Delay update_ssa until after predicate
11720 2018-03-08 David Malcolm <dmalcolm@redhat.com>
11722 PR tree-optimization/84178
11723 * tree-if-conv.c (release_bb_predicate): Remove the
11724 the assertion that the stmts have NULL use_ops.
11725 Discard the statements, asserting that they haven't
11726 yet been added to a BB.
11728 2018-03-08 Richard Biener <rguenther@suse.de>
11730 PR tree-optimization/84746
11731 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
11732 (phi_translate): Pass in destination ANTIC_OUT set.
11733 (phi_translate_1): Likewise. For a simplified result lookup
11734 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
11735 (phi_translate_set): Adjust.
11736 (do_pre_regular_insertion): Likewise.
11737 (do_pre_partial_partial_insertion): Likewise.
11739 2018-03-08 Martin Liska <mliska@suse.cz>
11741 PR gcov-profile/84735
11742 * doc/gcov.texi: Document usage of profile files.
11743 * gcov-io.h: Document changes in the format.
11745 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
11749 * dwarf2out.c (struct dw_line_info_table): Update comments for
11751 (FORCE_RESET_NEXT_VIEW): New.
11752 (FORCE_RESETTING_VIEW_P): New.
11753 (RESETTING_VIEW_P): Check for -1 too.
11754 (ZERO_VIEW_P): Likewise.
11755 (new_line_info_table): Force-reset next view.
11756 (dwarf2out_begin_function): Likewise.
11757 (dwarf2out_source_line): Simplify zero_view_p initialization.
11758 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
11759 view directly. Omit view when omitting .loc at line 0.
11761 2018-03-08 Jakub Jelinek <jakub@redhat.com>
11763 PR tree-optimization/84740
11764 * tree-switch-conversion.c (process_switch): Call build_constructors
11765 only if info.phi_count is non-zero.
11767 PR tree-optimization/84739
11768 * tree-tailcall.c (find_tail_calls): Check call arguments against
11769 DECL_ARGUMENTS (current_function_decl) rather than
11770 DECL_ARGUMENTS (func) when checking for tail recursion.
11772 2018-03-07 Jakub Jelinek <jakub@redhat.com>
11774 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
11775 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
11776 Volker Reichelt's entry and add entries for people that perform
11777 GCC fuzzy testing and report numerous bugs.
11779 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
11782 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
11783 readonly data in sdata, if that is disabled.
11784 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
11785 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
11786 -mreadonly-in-sdata option.
11788 2018-03-07 Martin Sebor <msebor@redhat.com>
11790 PR tree-optimization/84468
11791 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
11792 basic block when looking for nul assignment.
11794 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
11797 * except.h (output_function_exception_table): Adjust prototype.
11798 * except.c (output_function_exception_table): Remove FNNAME parameter
11799 and add SECTION parameter. Ouput one part of the table at a time.
11800 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
11801 the first part of the exception table and emit unwind directives.
11802 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
11803 (i386_pe_seh_cold_init): Likewise.
11804 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
11805 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
11806 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
11807 (ix86_output_call_insn): Emit a nop in one more case for SEH.
11808 * config/i386/winnt.c: Include except.h.
11809 (struct seh_frame_state): Add reg_offset, after_prologue and
11810 in_cold_section fields.
11811 (i386_pe_seh_end_prologue): Set seh->after_prologue.
11812 (i386_pe_seh_cold_init): New function.
11813 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
11814 to seh->in_cold_section.
11815 (seh_emit_push): Record the offset of the push.
11816 (seh_emit_save): Record the offet of the save.
11817 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
11818 Test seh->after_prologue to disregard the epilogue.
11819 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
11820 (i386_pe_end_cold_function): New function.
11822 2018-03-07 Jakub Jelinek <jakub@redhat.com>
11825 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
11826 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
11829 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
11831 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
11832 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
11834 PR middle-end/84723
11835 * multiple_target.c: Include tree-inline.h and intl.h.
11836 (expand_target_clones): Diagnose and fail if node->definition and
11837 !tree_versionable_function_p (node->decl).
11839 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
11841 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
11843 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
11844 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11845 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
11847 2018-03-06 Jakub Jelinek <jakub@redhat.com>
11850 * combine.c (try_combine): Use reg_or_subregno instead of handling
11851 just paradoxical SUBREGs and REGs.
11853 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
11855 * config/arc/arc.c (arc_finalize_pic): Remove function.
11856 (arc_must_save_register): We use single base PIC register, remove
11857 checks to save/restore the PIC register.
11858 (arc_expand_prologue): Likewise.
11859 * config/arc/arc-protos.h (arc_set_default_type_attributes):
11861 (arc_verify_short): Likewise.
11862 (arc_attr_type): Likewise.
11863 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
11864 (walk_stores): Likewise.
11865 (arc_address_cost): Make it static.
11866 (arc_verify_short): Likewise.
11867 (branch_dest): Likewise.
11868 (arc_attr_type): Likewise.
11869 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
11870 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
11871 (arc_final_prescan_insn): Remove inserting the nops due to
11872 hardware hazards. It is done in reorg step.
11873 (insn_length_variant_t): Remove.
11874 (insn_length_parameters_t): Likewise.
11875 (arc_insn_length_parameters): Likewise.
11876 (arc_get_insn_variants): Likewise.
11877 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
11879 2018-03-06 Jakub Jelinek <jakub@redhat.com>
11881 PR inline-asm/84683
11882 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
11885 PR tree-optimization/84687
11886 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
11888 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
11890 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11892 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
11893 Rename to ppc_speculation_barrier.
11894 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
11895 __builtin_ppc_speculation_barrier.
11897 2018-03-05 Jakub Jelinek <jakub@redhat.com>
11900 * combine.c (combine_simplify_rtx): Don't try to simplify if
11901 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
11904 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
11906 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
11907 to 32 bytes when compiling for POWER9.
11909 2018-03-05 Jakub Jelinek <jakub@redhat.com>
11912 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
11913 regparm >= 3 with no arg reg available also for calls with
11914 flag_force_indirect_call. Pass decl to ix86_function_regparm.
11917 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
11919 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
11921 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
11924 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
11926 2018-03-05 Richard Biener <rguenther@suse.de>
11928 PR tree-optimization/84486
11929 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
11930 When inserting a __builtin_assume_aligned call set the LHS
11931 SSA name alignment info accordingly.
11933 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
11935 PR tree-optimization/84114
11936 * config/aarch64/aarch64.c (aarch64_reassociation_width)
11937 Avoid reassociation of FLOAT_MODE addition.
11939 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
11941 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
11942 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
11943 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
11944 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
11945 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
11946 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
11947 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
11949 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
11950 __builtin_ia32_wbinvd): New builtins.
11951 (SPECIAL_ARGS2): New.
11952 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
11953 (SPECIAL_ARGS2): New.
11954 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
11955 (ix86_valid_target_attribute_inner_p): Ditto.
11956 (ix86_init_mmx_sse_builtins): Add special_args2.
11957 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
11958 TARGET_WBNOINVD_P): New.
11959 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
11960 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
11961 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
11962 * config/i386/immintrin.h (_wbinvd): New intrinsic.
11963 * config/i386/pconfigintrin.h: New file.
11964 * config/i386/wbnoinvdintrin.h: Ditto.
11965 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
11967 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
11969 2018-03-05 Richard Biener <rguenther@suse.de>
11971 PR tree-optimization/84670
11972 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
11974 (BB_VISITED_WITH_VISITED_SUCCS): New define.
11975 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
11976 (compute_antic_aux): Only assert the number of values in ANTIC_IN
11977 doesn't grow if all successors (recursively) were visited at least
11980 2018-03-05 Richard Biener <rguenther@suse.de>
11982 PR tree-optimization/84650
11983 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
11984 if executed in the loop pipeline.
11986 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
11988 * doc/configfiles.texi (Configuration Files): Move info about
11989 conditionalizing $target-protos.h to...
11990 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
11991 differs from $target-protos.h.
11993 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
11994 Chung-Ju Wu <jasonwucj@gmail.com>
11996 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
11997 * config/nds32/nds32-multiple.md (setmemsi): Define.
11998 * config/nds32/nds32-memory-manipulation.c
11999 (nds32_gen_dup_4_byte_to_word_value): New.
12000 (emit_setmem_word_loop): New.
12001 (emit_setmem_byte_loop): New.
12002 (nds32_expand_setmem_loop): New.
12003 (nds32_expand_setmem_loop_v3m): New.
12004 (nds32_expand_setmem_unroll): New.
12005 (nds32_expand_setmem): New.
12007 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
12008 Chung-Ju Wu <jasonwucj@gmail.com>
12010 * config/nds32/nds32-memory-manipulation.c
12011 (nds32_emit_load_store): New.
12012 (nds32_emit_post_inc_load_store): New.
12013 (nds32_emit_mem_move): New.
12014 (nds32_emit_mem_move_block): New.
12015 (nds32_expand_movmemsi_loop_unknown_size): New.
12016 (nds32_expand_movmemsi_loop_known_size): New.
12017 (nds32_expand_movmemsi_loop): New.
12018 (nds32_expand_movmemsi_unroll): New.
12019 (nds32_expand_movmemqi): Rename ...
12020 (nds32_expand_movmemsi): ... to this.
12021 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
12022 (movmemsi): ... to this.
12023 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
12024 (nds32_expand_movmemsi): ... to this.
12026 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
12027 Monk Chiang <sh.chiang04@gmail.com>
12028 Chung-Ju Wu <jasonwucj@gmail.com>
12030 * config/nds32/nds32-protos.h
12031 (nds32_expand_load_multiple): New arguments.
12032 (nds32_expand_store_multiple): Ditto.
12033 (nds32_valid_multiple_load_store): Rename ...
12034 (nds32_valid_multiple_load_store_p): ... to this.
12035 * config/nds32/nds32-memory-manipulation.c
12036 (nds32_expand_load_multiple): Refine implementation.
12037 (nds32_expand_store_multiple): Ditto.
12038 * config/nds32/nds32-multiple.md
12039 (load_multiple): Update nds32_expand_load_multiple interface.
12040 (store_multiple): Update nds32_expand_store_multiple interface.
12041 * config/nds32/nds32-predicates.c
12042 (nds32_valid_multiple_load_store): Rename ...
12043 (nds32_valid_multiple_load_store_p): ... to this and refine
12045 * config/nds32/predicates.md
12046 (nds32_load_multiple_and_update_address_operation): New predicate.
12047 (nds32_store_multiple_and_update_address_operation): New predicate.
12049 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
12050 Chung-Ju Wu <jasonwucj@gmail.com>
12052 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
12053 (combo): New attribute.
12054 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
12056 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
12058 * config/nds32/nds32.opt: Change -mcmodel= default value.
12060 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
12061 Monk Chiang <sh.chiang04@gmail.com>
12062 Chung-Ju Wu <jasonwucj@gmail.com>
12064 * config/nds32/constants.md (unspec_element): New enum.
12065 * config/nds32/constraints.md (Umw): New constraint.
12066 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
12067 * config/nds32/nds32-intrinsic.md: Likewise.
12068 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
12069 (nds32_valid_smw_lwm_base_p): New.
12070 (nds32_output_smw_single_word): New.
12071 (nds32_output_lmw_single_word): New.
12072 (nds32_expand_unaligned_load): New.
12073 (nds32_expand_unaligned_store): New.
12074 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
12075 (nds32_output_smw_single_word): Declare.
12076 (nds32_output_lmw_single_word): Declare.
12077 (nds32_expand_unaligned_load): Declare.
12078 (nds32_expand_unaligned_store): Declare.
12079 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
12080 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
12081 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
12082 NDS32_BUILTIN_UASTORE_DW.
12083 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
12086 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
12087 Kito Cheng <kito.cheng@gmail.com>
12088 Chung-Ju Wu <jasonwucj@gmail.com>
12090 * config/nds32/nds32-intrinsic.c
12091 (nds32_expand_builtin_null_ftype_reg): Delete.
12092 (nds32_expand_builtin_reg_ftype_imm): Ditto.
12093 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
12094 (nds32_read_argument): New.
12095 (nds32_legitimize_target): Ditto.
12096 (nds32_legitimize_argument): Ditto.
12097 (nds32_check_constant_argument): Ditto.
12098 (nds32_expand_unop_builtin): Ditto.
12099 (nds32_expand_unopimm_builtin): Ditto.
12100 (nds32_expand_binop_builtin): Ditto.
12101 (nds32_builtin_decl_impl): Ditto.
12102 (builtin_description): Ditto.
12103 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
12104 (nds32_init_builtins_impl): Ditto.
12105 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
12106 (nds32_builtin_decl): New.
12107 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
12108 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
12110 2018-03-02 Jeff Law <law@redhat.com>
12112 * reorg.c (stop_search_p): Handle DEBUG_INSN.
12113 (redundant_insn, fill_simple_delay_slots): Likewise.
12114 (fill_slots_from_thread): Likewise.
12115 * resource.c (mark_referenced_resources): Likewise.
12116 (mark_set_resources, find_dead_or_set_registers): Likewise.
12118 2018-03-02 Jakub Jelinek <jakub@redhat.com>
12120 * substring-locations.h (format_warning_va): Formatting fix for
12121 ATTRIBUTE_GCC_DIAG.
12122 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
12124 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
12125 * substring-locations.c: Include intl.h.
12126 (format_warning_va): Turned into small wrapper around
12127 format_warning_n_va, renamed to ...
12128 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
12129 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
12131 (format_warning_at_substring_n): New function.
12132 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
12133 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
12134 format_warning_at_substring with just a shorter name instead of
12135 const function pointer.
12136 (fmtwarn_n): New function.
12137 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
12138 appropriate, get rid of all the fmtstr temporaries, move conditionals
12139 with G_() wrapped string literals directly into fmtwarn arguments,
12140 cast dir.len to (int), formatting fixes.
12142 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
12144 * doc/invoke.texi: Remove "Cilk Plus" references.
12146 2018-03-02 Jakub Jelinek <jakub@redhat.com>
12147 Richard Biener <rguenther@suse.de>
12150 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
12151 for error or warning attributes if CALL_FROM_THUNK_P is set.
12154 2018-03-02 Jakub Jelinek <jakub@redhat.com>
12157 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
12158 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
12161 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
12162 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
12164 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
12165 instead of -1U in last predictors element's probability member.
12167 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
12170 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
12171 arguments if they are comparable.
12173 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
12175 PR tree-optimization/84634
12176 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
12177 masks and masked_loop_p with a single loop_masks, making sure it's
12178 null for bb vectorization.
12180 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
12182 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
12183 (vect_analyze_data_ref_access): Use loop->safe_len rather than
12184 loop->force_vectorize to check whether there is no alias.
12186 2018-03-02 Jakub Jelinek <jakub@redhat.com>
12189 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
12191 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
12193 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
12194 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
12195 instead of a loop around prev_real_insn.
12196 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
12199 PR inline-asm/84625
12200 * config/i386/i386.c (ix86_print_operand): Use conditional
12201 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
12204 2018-03-02 Richard Biener <rguenther@suse.de>
12206 PR tree-optimization/84427
12207 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
12208 (bitmap_set_subtract_values): Rewrite to handle multiple
12211 (prune_clobbered_mems): Likewise.
12212 (phi_translate): Take edge instead of pred/phiblock.
12213 (phi_translate_1): Likewise.
12214 (phi_translate_set): Likewise. Insert all translated
12215 exprs for a value into the set, keeping possibly multiple
12216 expressions per value.
12217 (compute_antic_aux): Adjust for phi_translate changes.
12218 When intersecting union the expressions and prune those
12219 not in the final value set, keeping possibly multiple
12220 expressions per value. Do not use value-insertion
12221 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
12222 all expressions. Add verification that the value-sets
12223 only shrink during iteration.
12224 (compute_partial_antic_aux): Adjust for the phi_translate changes.
12225 (do_pre_regular_insertion): Likewise.
12226 (do_pre_partial_partial_insertion): Likewise.
12228 2018-03-02 Richard Biener <rguenther@suse.de>
12231 * config/darwin.c (saved_debug_info_level): New static global.
12232 (darwin_asm_lto_start): Disable debug info generation for LTO out.
12233 (darwin_asm_lto_end): Restore debug info generation settings.
12235 2018-03-01 Martin Liska <mliska@suse.cz>
12238 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12239 volatile arguments.
12241 2018-03-01 Richard Biener <rguenther@suse.de>
12244 * dwarf2out.c (gen_variable_die): Properly handle late VLA
12245 type annotation with LTO when debug was disabled at compile-time.
12247 2018-03-01 Matthew Fortune <mfortune@gmail.com>
12249 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
12251 (mips_final_postscan_insn): Likewise.
12253 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
12255 PR rtl-optimization/84528
12256 * alias.c (init_alias_target): Add commentary.
12257 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
12258 a unique base value if the frame pointer is not eliminated
12259 to the stack pointer.
12261 2018-03-01 Tom de Vries <tom@codesourcery.com>
12263 PR rtl-optimization/83327
12264 * lra-int.h (hard_regs_spilled_into): Declare.
12265 * lra.c (hard_regs_spilled_into): Define.
12266 (init_reg_info): Init hard_regs_spilled_into.
12267 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
12268 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
12269 (process_bb_lives): Handle hard_regs_spilled_into.
12270 (lra_create_live_ranges_1): Before doing liveness propagation, clear
12271 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
12273 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
12275 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
12276 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
12277 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
12278 * config/rs6000/aix72.h: New file.
12280 2018-02-28 Jakub Jelinek <jakub@redhat.com>
12282 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
12283 instead of warning_at with conditional singular and plural messages
12287 * stor-layout.c (update_alignment_for_field): For
12288 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
12289 && !DECL_PACKED (field), do the alignment update, just use
12290 only desired_align instead of MAX (type_align, desired_align)
12292 (place_field): Don't do known_align < desired_align handling
12293 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
12294 is non-NULL, instead do it after rli->prev_field handling and
12295 only if not within a bitfield word. For DECL_PACKED (field)
12296 use type_align of BITS_PER_UNIT.
12298 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
12300 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
12301 superfluous parentheses and trailing spaces.
12303 2018-02-28 Richard Biener <rguenther@suse.de>
12305 PR tree-optimization/84584
12306 * graphite-scop-detection.c (scop_detection::add_scop): Discard
12307 SCoPs with fake exit edge.
12309 2018-02-28 Martin Liska <mliska@suse.cz>
12312 * timevar.c (timer::print): Fix format to properly print 100%
12315 2018-02-28 Richard Biener <rguenther@suse.de>
12317 PR middle-end/84607
12318 * genmatch.c (capture_info::walk_match): Do not mark
12319 captured expressions without operands as expr_p given
12320 they act more like predicates and should be subject to
12321 "lost tail" side-effect preserving.
12323 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
12325 PR rtl-optimization/81611
12326 * auto-inc-dec.c (attempt_change): Move dead note from
12327 mem_insn if it's the next use of regno
12328 (find_address): Take address use of reg holding
12329 non-incremented value. Add parm to limit search to the named
12331 (merge_in_block): Attempt to use a mem insn that is the next
12332 use of the original regno.
12334 2018-02-27 Martin Sebor <msebor@redhat.com>
12337 * gcc/doc/invoke.texi (-Wmissing-attributes): New option.
12338 * gcc/print-tree.c (print_node): Handle DECL_UNINLINABLE.
12340 2018-02-27 Martin Sebor <msebor@redhat.com>
12342 PR translation/84207
12343 * diagnostic-core.h (warning_n, error_n, inform_n): Change
12344 n argument to unsigned HOST_WIDE_INT.
12345 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
12346 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
12347 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
12348 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
12350 2018-02-27 Richard Biener <rguenther@suse.de>
12352 PR tree-optimization/84512
12353 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
12354 Do not use the estimate returned from record_stmt_cost for
12355 the scalar iteration cost but sum properly using add_stmt_cost.
12357 2018-02-27 Richard Biener <rguenther@suse.de>
12359 PR tree-optimization/84466
12360 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
12361 Adjust last change to less strictly validate use operands.
12363 2018-02-27 Martin Liska <mliska@suse.cz>
12365 PR gcov-profile/84548
12366 * gcov.c (process_file): Allow partial overlap and consider it
12367 also as group functions.
12368 (output_lines): Properly calculate range of lines for a group.
12370 2018-02-27 Martin Liska <mliska@suse.cz>
12372 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
12373 'ggc' suffixes. Change first column width.
12374 (timer::print): Fix formatting of the column.
12376 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
12378 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
12379 preserve inline entry blocks for the sake of debug inline
12380 entry point markers alone.
12381 (remove_unused_locals): Suggest in comments a better place to
12382 force the preservation of inline entry blocks that are
12383 otherwise unused, but do not preserve them.
12385 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
12387 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
12389 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
12392 * config/i386/constraints.md (Bs): Replace
12393 ix86_indirect_branch_register with
12394 TARGET_INDIRECT_BRANCH_REGISTER.
12396 * config/i386/i386.md (indirect_jump): Likewise.
12397 (tablejump): Likewise.
12398 (*sibcall_memory): Likewise.
12399 (*sibcall_value_memory): Likewise.
12400 Peepholes of indirect call and jump via memory: Likewise.
12401 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
12402 (*sibcall_value_GOT_32): Likewise.
12403 * config/i386/predicates.md (indirect_branch_operand): Likewise.
12404 (GOT_memory_operand): Likewise.
12405 (call_insn_operand): Likewise.
12406 (sibcall_insn_operand): Likewise.
12407 (GOT32_symbol_operand): Likewise.
12408 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
12410 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
12412 PR rtl-optimization/83496
12413 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
12414 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
12415 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
12416 redundant insn, if any.
12417 (relax_delay_slots): Likewise.
12418 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
12420 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
12422 PR tree-optimization/83965
12423 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
12424 that grouped statements are part of a reduction chain. Return
12425 true if the statement is not marked as a reduction itself but
12426 is part of a group.
12427 (vect_recog_dot_prod_pattern): Don't check whether the statement
12428 is part of a group here.
12429 (vect_recog_sad_pattern): Likewise.
12430 (vect_recog_widen_sum_pattern): Likewise.
12432 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
12435 * final.c (rest_of_clean_state): Also look for calls inside sequences.
12437 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
12440 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
12442 (ix86_output_indirect_function_return): New prototype.
12443 (ix86_split_simple_return_pop_internal): Likewise.
12444 * config/i386/i386.c (indirect_return_via_cx): New.
12445 (indirect_return_via_cx_bnd): Likewise.
12446 (indirect_thunk_name): Handle return va CX_REG.
12447 (output_indirect_thunk_function): Create alias for
12448 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
12449 (ix86_output_indirect_jmp): Remove the bool argument.
12450 (ix86_output_indirect_function_return): New function.
12451 (ix86_split_simple_return_pop_internal): Likewise.
12452 * config/i386/i386.md (*indirect_jump): Don't pass false
12453 to ix86_output_indirect_jmp.
12454 (*tablejump_1): Likewise.
12455 (simple_return_pop_internal): Change it to define_insn_and_split.
12456 Call ix86_split_simple_return_pop_internal to split it for
12457 -mfunction-return=.
12458 (simple_return_indirect_internal): Call
12459 ix86_output_indirect_function_return instead of
12460 ix86_output_indirect_jmp.
12462 2018-02-26 Jakub Jelinek <jakub@redhat.com>
12465 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
12466 memset and value initialization afterwards.
12468 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
12470 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
12472 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12475 * common/config/aarch64/aarch64-common.c
12476 (aarch_option_optimization_table[]): Switch
12477 off fomit-frame-pointer
12479 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
12480 Chung-Ju Wu <jasonwucj@gmail.com>
12482 * config/nds32/nds32-multiple.md (load_multiple): Disallow
12484 (store_multiple): Ditto.
12486 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
12488 * config.gcc: Add --with-cpu support for nds32 target.
12489 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
12490 * config/nds32/nds32.opt: Add -mcpu= option.
12492 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
12494 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
12495 isel=yes): Warn for these deprecated options.
12497 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
12499 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
12500 ISA_2_5_MASKS_EMBEDDED.
12502 2018-02-23 Jakub Jelinek <jakub@redhat.com>
12504 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
12505 p->max as pointers rather than using iterative_hash_expr.
12507 2018-02-23 Carl Love <cel@us.ibm.com>
12509 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
12510 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
12512 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
12513 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
12514 P8V_BUILTIN_VEC_VUNSIGNED2.
12516 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
12519 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
12520 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
12522 * lra-constraints.c (curr_insn_transform): Set up
12523 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
12525 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
12526 * lra-eliminations.c (spill_pseudos): Ditto.
12527 (process_insn_for_elimination): Ditto.
12528 * lra-lives.c (reg_early_clobber_p): Use the new macros.
12529 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
12530 LRA_NON_CLOBBERED_ALT.
12532 2018-02-22 Martin Sebor <msebor@redhat.com>
12534 PR tree-optimization/84480
12535 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
12536 to maybe_diag_stxncpy_trunc. Call it.
12537 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
12538 from gimple_fold_builtin_strcpy. Print inlining stack.
12539 (handle_builtin_stxncpy): Print inlining stack.
12540 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
12542 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
12545 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
12546 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
12547 and -fcheck-pointer-bounds are used together.
12548 (indirect_thunk_prefix): New enum.
12549 (indirect_thunk_need_prefix): New function.
12550 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
12551 "_nt" instead of "_bnd" for NOTRACK prefix.
12552 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
12553 (output_indirect_thunk_function): Likewise.
12555 (ix86_code_end): Update output_indirect_thunk_function calls.
12556 (ix86_output_indirect_branch_via_reg): Replace
12557 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
12558 (ix86_output_indirect_branch_via_push): Likewise.
12559 (ix86_output_function_return): Likewise.
12560 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
12561 incompatible with -fcf-protection=branch and
12562 -fcheck-pointer-bounds.
12564 2018-02-22 Steve Ellcey <sellcey@cavium.com>
12567 * config/aarch64/aarch64.c (aarch64_print_address_internal):
12568 Change gcc_assert call to output_operand_lossage.
12570 2018-02-22 Steve Ellcey <sellcey@cavium.com>
12572 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
12574 2018-02-22 DJ Delorie <dj@redhat.com>
12575 Sebastian Perta <sebastian.perta@renesas.com>
12576 Oleg Endo <olegendo@gcc.gnu.org>
12578 * config/rx/rx.c (rx_rtx_costs): New function.
12579 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
12581 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
12583 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
12585 2018-02-22 Martin Liska <mliska@suse.cz>
12588 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
12589 Add "native" as a possible value.
12591 2018-02-22 Martin Liska <mliska@suse.cz>
12594 * config/i386/i386.c (ix86_option_override_internal):
12595 Add "native" as a possible value for -march and -mtune.
12597 2018-02-22 Jakub Jelinek <jakub@redhat.com>
12600 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
12601 to all type variants.
12603 PR tree-optimization/84503
12604 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
12605 width as info->bitpos + info->bitsize - start.
12606 (merged_store_group::merge_overlapping): Simplify width computation.
12607 (check_no_overlap): New function.
12608 (imm_store_chain_info::try_coalesce_bswap): Compute expected
12609 start + width and last_order of the group, fail if check_no_overlap
12611 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
12612 to group if check_no_overlap fails.
12614 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
12616 * config/rs6000/altivec.md: Delete contraint arguments to
12617 define_expand, define_split, and define_peephole2, and in
12618 define_insn_and_split if always unused.
12619 * config/rs6000/darwin.md: Ditto.
12620 * config/rs6000/dfp.md: Ditto.
12621 * config/rs6000/rs6000.md: Ditto.
12622 * config/rs6000/sync.md: Ditto.
12623 * config/rs6000/vector.md: Ditto.
12624 * config/rs6000/vsx.md: Ditto.
12626 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
12628 * config/rs6000/altivec.md: Write output control strings as braced
12629 blocks instead of double-quoted strings.
12630 * config/rs6000/darwin.md: Ditto.
12631 * config/rs6000/rs6000.md: Ditto.
12632 * config/rs6000/vector.md: Ditto.
12633 * config/rs6000/vsx.md: Ditto.
12635 2018-02-21 Jason Merrill <jason@redhat.com>
12637 PR c++/84314 - ICE with templates and fastcall attribute.
12638 * attribs.c (build_type_attribute_qual_variant): Remove assert.
12640 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
12642 * ipa-cp.c (determine_versionability): Fix comment typos.
12644 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
12647 * ipa-cp.c (determine_versionability): Do not version functions caling
12650 2018-02-21 Martin Liska <mliska@suse.cz>
12653 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
12654 Add "native" as a possible value.
12655 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
12656 the macro when native cpu detection is available.
12658 2018-02-21 Martin Liska <mliska@suse.cz>
12661 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
12662 Add "native" as a possible value.
12663 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
12664 when native cpu detection is available.
12666 2018-02-21 Jakub Jelinek <jakub@redhat.com>
12667 Martin Sebor <msebor@redhat.com>
12669 PR tree-optimization/84478
12670 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
12672 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
12673 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
12674 support which is conservatively correct, for 2 only stay conservative
12675 for maxlen. Formatting and comment capitalization fixes. Add STRICT
12676 argument to the 2 argument get_range_strlen, adjust 6 arg
12677 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
12679 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
12680 (gimple_fold_builtin_strlen): Pass true as last argument to
12683 2018-02-20 Martin Sebor <msebor@redhat.com>
12685 PR middle-end/84095
12686 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
12687 (builtin_memref::set_base_and_offset): Same. Handle inner references.
12688 (builtin_memref::builtin_memref): Factor out parts into
12689 set_base_and_offset and call it.
12691 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
12693 PR middle-end/84406
12694 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
12695 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
12696 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
12697 search at the associated MODE_INT.
12699 2018-02-20 Jeff Law <law@redhat.com>
12701 PR middle-end/82123
12702 PR tree-optimization/81592
12703 PR middle-end/79257
12704 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
12705 for range data rather than using global data.
12706 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
12707 range data rather than using global data.
12708 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
12709 pass it to children as needed.
12710 (struct directive::fmtresult): Similarly.
12711 (struct directive::set_width): Similarly.
12712 (struct directive::set_precision): Similarly.
12713 (format_integer, format_directive, parse_directive): Similarly.
12714 (format_none): Accept unnamed vr_values parameter.
12715 (format_percent, format_floating, format_character): Similarly.
12716 (format_string, format_plain): Similarly.
12717 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
12718 the EVRP range analyzer for range data rather than using global data.
12719 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
12720 gimple-ssa-evrp-analyze.h
12721 (class sprintf_dom_walker): Add after_dom_children member function.
12722 Add evrp_range_analyzer member.
12723 (sprintf_dom_walker::before_dom_children): Call into the EVRP
12724 range analyzer as needed.
12725 (sprintf_dom_walker::after_dom_children): New member function.
12726 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
12728 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
12729 (evrp_range_analyzer::pop_to_marker): Likewise.
12731 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
12733 PR tree-optimization/84419
12734 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
12735 with the required type if its current type is compatible but
12738 2018-02-20 Jakub Jelinek <jakub@redhat.com>
12740 PR middle-end/82004
12741 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
12742 after vectorization.
12744 2018-02-20 Martin Liska <mliska@suse.cz>
12747 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
12748 possible values if we don't have a hint.
12750 2018-02-20 Martin Liska <mliska@suse.cz>
12754 * final.c (shorten_branches): Build align_tab array with one
12756 * opts.c (finish_options): Add alignment option limit check.
12757 (MAX_CODE_ALIGN): Likewise.
12758 (MAX_CODE_ALIGN_VALUE): Likewise.
12759 * doc/invoke.texi: Document maximum allowed option value for
12760 all -falign-* options.
12762 2018-02-19 Jakub Jelinek <jakub@redhat.com>
12765 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
12766 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
12767 * var-tracking.c (emit_note_insn_var_location): Remove all references
12768 to NOTE_INSN_CALL_ARG_LOCATION.
12769 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
12770 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
12771 Use copy_rtx_if_shared.
12772 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
12773 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
12774 (dwarf2out_var_location): Remove handling of
12775 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
12777 * final.c (final_scan_insn): Remove all references to
12778 NOTE_INSN_CALL_ARG_LOCATION.
12779 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
12780 before dumping final insns.
12781 * except.c (emit_note_eh_region_end): Remove all references to
12782 NOTE_INSN_CALL_ARG_LOCATION.
12783 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
12784 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
12785 * config/arc/arc.c (hwloop_optimize): Likewise.
12786 * config/arm/arm.c (create_fix_barrier): Likewise.
12787 * config/s390/s390.c (s390_chunkify_start): Likewise.
12788 * config/sh/sh.c (find_barrier): Likewise.
12789 * config/i386/i386.c (rest_of_insert_endbranch,
12790 ix86_seh_fixup_eh_fallthru): Likewise.
12791 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
12792 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
12793 * config/frv/frv.c (frv_function_prologue): Likewise.
12794 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
12796 (note_outside_basic_block_p): Remove all references to
12797 NOTE_INSN_CALL_ARG_LOCATION.
12798 * gengtype.c (adjust_field_rtx_def): Likewise.
12799 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
12801 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
12802 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
12805 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
12808 PR tree-optimization/84452
12809 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
12810 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
12813 2018-02-19 Martin Liska <mliska@suse.cz>
12816 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
12818 2018-02-19 Martin Liska <mliska@suse.cz>
12819 Richard Sandiford <richard.sandiford@linaro.org>
12821 PR tree-optimization/82491
12822 * gimple-fold.c (get_base_constructor): Make earlier bail out
12825 2018-02-19 Carl Love <cel@us.ibm.com>
12827 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
12828 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
12830 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
12831 P8V_BUILTIN_VEC_NEG.
12833 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
12835 * config/rl78/rl78.md (movdf): New define expand.
12837 2018-02-19 Martin Liska <mliska@suse.cz>
12840 * doc/invoke.texi: Fix typo.
12841 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
12843 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
12845 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
12846 handle rs6000_single_float and rs6000_double_float specially for
12849 2018-02-16 Jeff Law <law@redhat.com>
12851 * config/rx/rx.c (add_pop_cfi_notes): New function.;
12852 (pop_regs): Use it.
12854 2018-02-16 Jakub Jelinek <jakub@redhat.com>
12857 * ipa-inline.c (inline_small_functions): Fix a typo.
12859 2018-02-16 Nathan Sidwell <nathan@acm.org>
12861 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
12863 2018-02-16 Carl Love <cel@us.ibm.com>
12865 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
12866 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
12867 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
12868 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
12869 expansion to P8V_BUILTIN_VEC_FLOAT2.
12871 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
12873 PR rtl-optimization/70023
12874 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
12875 src_regno into account.
12877 2018-02-16 Carl Love <cel@us.ibm.com>
12879 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
12880 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
12881 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
12882 * config/rs6000/rs6000.c: Remove case statements for
12883 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
12884 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
12885 and P9V_BUILTIN_VEC_VINSERT4B.
12886 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
12887 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
12888 * config/rs6000/vsx.md:
12889 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
12892 2018-02-16 Carl Love <cel@us.ibm.com>
12894 * config/rs6000/altivec.h: Add builtin names vec_extract4b
12896 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
12898 * config/rs6000/rs6000-c.c: Add the definitions for
12899 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
12900 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
12901 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
12902 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
12903 definition for insert4b and define insn *insert3b_internal.
12904 * doc/extend.texi: Add documentation for vec_extract4b.
12906 2018-02-16 Nathan Sidwell <nathan@acm.org>
12908 * doc/extend.texi (Backwards Compatibility): Mention friend
12909 injection. Note for-scope is deprecated.
12910 * doc/invoke.texi (-ffriend-injection): Deprecate.
12912 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
12914 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
12915 that moved to I2, also allow destinations that are a paradoxical
12916 subreg (instead of a normal reg).
12918 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
12921 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
12924 2018-02-16 Richard Biener <rguenther@suse.de>
12926 PR tree-optimization/84037
12927 PR tree-optimization/84016
12929 * config/i386/i386.c (ix86_builtin_vectorization_cost):
12930 Adjust vec_construct for the fact we need additional higher latency
12931 128bit inserts for AVX256 and AVX512 vector builds.
12932 (ix86_add_stmt_cost): Scale vector construction cost for
12935 2018-02-16 Richard Biener <rguenther@suse.de>
12937 PR tree-optimization/84417
12938 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
12939 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
12940 (non_rewritable_lvalue_p): Likewise, use poly-ints.
12942 2018-02-16 Martin Liska <mliska@suse.cz>
12945 * internal-fn.def (ASAN_CHECK): Set proper flags.
12946 (ASAN_MARK): Likewise.
12948 2018-02-16 Julia Koval <julia.koval@intel.com>
12950 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
12951 from PTA_CANNONLAKE.
12953 2018-02-16 Jakub Jelinek <jakub@redhat.com>
12956 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
12957 Use ++iter rather than iter++ for std::list iterators.
12958 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
12959 defer deleting them until all nodes in the forest are processed. Do
12960 free even leaf nodes. Change to_process into auto_vec.
12963 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
12964 * vec.h (vec_default_construct): Use memset instead of placement new
12965 if BROKEN_VALUE_INITIALIZATION is defined.
12966 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
12967 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
12970 PR rtl-optimization/83723
12971 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
12972 * lra.c (lra_substitute_pseudo): Likewise. If true, use
12973 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
12975 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
12977 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
12979 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
12981 PR rtl-optimization/81443
12982 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
12983 from inner REGs to paradoxical SUBREGs.
12985 2018-02-16 Richard Biener <rguenther@suse.de>
12987 PR tree-optimization/84399
12988 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
12989 For operands we can analyze at their definition make sure we can
12990 analyze them at each use as well.
12992 2018-02-16 Richard Biener <rguenther@suse.de>
12994 PR tree-optimization/84190
12995 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
12996 volatile accesses if the decl isn't volatile.
12998 2018-02-15 Jason Merrill <jason@redhat.com>
13000 PR c++/84314 - ICE with templates and fastcall attribute.
13001 * attribs.c (build_type_attribute_qual_variant): Don't clobber
13002 TYPE_CANONICAL on an existing type.
13004 2018-02-15 Jakub Jelinek <jakub@redhat.com>
13006 PR tree-optimization/84383
13007 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
13008 dstoff nor call operand_equal_p if dstbase is NULL.
13010 PR tree-optimization/84334
13011 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
13012 also a CONSTANT_CLASS_P, punt.
13014 2018-02-14 Jim Wilson <jimw@sifive.com>
13016 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
13017 first SMALL_OPERAND check. New local min_second_step. Move assert
13018 to where locals are set. Add TARGET_RVC support.
13019 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
13021 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
13023 * doc/invoke.texi: Correct -Wformat-overflow code sample.
13025 2018-02-14 Martin Sebor <msebor@redhat.com>
13027 PR tree-optimization/83698
13028 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
13029 arrays constrain the offset range to their bounds.
13030 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
13031 (builtin_access::overlap): Avoid setting the size of overlap if it's
13033 (maybe_diag_overlap): Also consider arrays when deciding what values
13034 of offsets to include in diagnostics.
13036 2018-02-14 Martin Sebor <msebor@redhat.com>
13039 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
13040 that correspond to the kind of a declaration.
13042 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
13045 * config/pa/pa.md: Load address of PIC label using the linkage table
13046 if the label is nonlocal.
13048 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
13050 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
13051 warning message if user requests -maltivec=be.
13052 * doc/invoke.texi: Document deprecation of -maltivec=be.
13054 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
13057 * config/rs6000/rs6000-c.c: Update definitions for
13058 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
13059 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
13061 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
13064 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
13065 add _get_ssp intrinsics. Remove argument from
13066 __builtin_ia32_rdssp[d|q].
13067 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
13068 * config/i386/i386-builtin.def: Remove argument from
13069 __builtin_ia32_rdssp[d|q].
13070 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
13071 ix86_expand_special_args_builtin for _rdssp[d|q].
13072 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
13073 Clear register before usage.
13074 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
13075 Add documentation for new _get_ssp and _inc_ssp intrinsics.
13077 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
13079 PR tree-optimization/84357
13080 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
13081 operand 1 of an ARRAY_REF too.
13083 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
13086 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
13087 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
13089 (set_of_reg): New struct.
13090 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
13091 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
13092 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
13094 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
13095 Split into bitclr, bitset, bitinvert patterns if appropriate.
13096 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
13097 use rx_fuse_in_memory_bitop.
13098 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
13099 to named insn, correct maximum insn length.
13101 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
13104 * machmode.def: Define a complex mode for PARTIAL_INT.
13105 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
13107 * doc/rtl.texi: Document CSPImode.
13108 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
13110 (msp430_hard_regno_nregs_with_padding): Likewise.
13112 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
13115 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
13117 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
13119 PR rtl-optimization/84169
13120 * combine.c (try_combine): New variable split_i2i3. Set it to true if
13121 we generated a parallel as new i3 and we split that to new i2 and i3
13122 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
13123 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
13124 those to i2, not i1. Partially rewrite this scan code.
13126 2018-02-13 Jakub Jelinek <jakub@redhat.com>
13129 * stor-layout.c (place_field): For variable length fields, adjust
13130 offset_align afterwards not just based on the field's alignment,
13131 but also on the size.
13133 PR middle-end/84309
13134 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
13135 of exps and logs in the use_exp2 case.
13137 2018-02-13 Jeff Law <law@redhat.com>
13139 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
13140 entry for "vector".
13142 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
13145 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
13149 * common.opt (gas-loc-support, gas-locview-support): New.
13150 (ginline-points, ginternal-reset-location-views): New.
13151 * doc/invoke.texi: Document them. Use @itemx where intended.
13152 (gvariable-location-views): Adjust.
13153 * target.def (reset_location_view): New.
13154 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
13155 (TARGET_RESET_LOCATION_VIEW): New.
13156 * doc/tm.texi: Rebuilt.
13157 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
13158 (dwarf2out_default_as_locview_support): New.
13159 (output_asm_line_debug_info): Use option variables.
13160 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
13161 (output_loc_list): Likewise.
13162 (add_high_low_attributes): Check option variables.
13163 Don't output entry view attribute in strict mode.
13164 (gen_inlined_subroutine_die): Check option variables.
13165 (dwarf2out_inline_entry): Likewise.
13166 (init_sections_and_labels): Likewise.
13167 (dwarf2out_early_finish): Likewise.
13168 (maybe_reset_location_view): New, from...
13169 (dwarf2out_var_location): ... here. Call it.
13170 * debug.h (dwarf2out_default_as_loc_support): Declare.
13171 (dwarf2out_default_as_locview_support): Declare.
13172 * hooks.c (hook_int_rtx_insn_0): New.
13173 * hooks.h (hook_int_rtx_insn_0): Declare.
13174 * toplev.c (process_options): Take -gas-loc-support and
13175 -gas-locview-support from dwarf2out. Enable
13176 -gvariable-location-views by default only with locview
13177 assembler support. Enable -ginternal-reset-location-views by
13178 default only if the target defines the corresponding hook.
13179 Enable -ginline-points by default if location views are
13180 enabled; force it disabled if statement frontiers are
13182 * tree-inline.c (expand_call_inline): Check option variables.
13183 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
13185 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
13187 PR tree-optimization/84321
13188 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
13189 handling. Also check whether the anti-range contains any values
13190 that satisfy the mask; switch to a VR_RANGE if not.
13192 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
13195 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
13197 2018-02-13 Martin Jambor <mjambor@suse.cz>
13200 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
13201 of call statements, also set location of a load to a temporary.
13203 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
13205 * config/rl78/rl78.c (add_vector_labels): New function.
13206 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
13207 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
13208 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
13209 which checks that no arguments are passed.
13210 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
13211 * doc/extend.texi: Documentation for the new attribute.
13213 2018-02-13 Andreas Schwab <schwab@suse.de>
13215 * config/riscv/linux.h (CPP_SPEC): Define.
13217 2018-02-13 Jakub Jelinek <jakub@redhat.com>
13220 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
13221 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
13222 OPTION_MASK_ISA_AES as first argument to def_builtin_const
13223 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
13224 instead of OPTION_MASK_ISA_PCLMUL as first argument to
13225 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
13226 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
13227 temporarily for AES and PCLMUL builtins.
13229 PR tree-optimization/84339
13230 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
13231 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
13234 PR middle-end/84309
13235 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
13236 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
13237 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
13239 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
13241 * omp-simd-clone.h: New file.
13242 * omp-simd-clone.c: Include omp-simd-clone.h.
13243 (expand_simd_clones): No longer static.
13244 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
13245 cgraph.h and omp-simd-clone.h.
13246 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
13247 (vect_recog_widen_shift_pattern): Formatting fix.
13248 (vect_pattern_recog_1): Don't check optab for calls.
13251 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
13252 operands[2] into a REG before using gen_lowpart on it.
13254 2018-02-12 Jeff Law <law@redhat.com>
13257 * config/sh/sh.c (find_barrier): Consider a sibling call
13260 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
13261 successfully back substituting a reg.
13263 2018-02-12 Richard Biener <rguenther@suse.de>
13265 PR tree-optimization/84037
13266 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
13267 parameter, move visited init to caller.
13268 (vect_slp_analyze_operations): Separate cost from validity
13269 check, initialize visited once for all instances.
13270 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
13272 * tree-vect-stmts.c (vect_model_simple_cost): Make early
13274 (vect_model_promotion_demotion_cost): Likewise.
13275 (vectorizable_bswap): Guard cost modeling with !slp_node
13276 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
13278 (vectorizable_call): Likewise.
13279 (vectorizable_conversion): Likewise.
13280 (vectorizable_assignment): Likewise.
13281 (vectorizable_shift): Likewise.
13282 (vectorizable_operation): Likewise.
13283 (vectorizable_store): Likewise.
13284 (vectorizable_load): Likewise.
13285 (vectorizable_condition): Likewise.
13286 (vectorizable_comparison): Likewise.
13288 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
13291 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
13292 (ASAN_MARK): Fix fnspec to account for return value, change pointer
13293 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
13295 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
13297 PR middle-end/83665
13298 * params.def (inline-min-speedup): Increase from 8 to 15.
13299 (max-inline-insns-auto): Decrease from 40 to 30.
13300 * ipa-split.c (consider_split): Add some buffer for function to
13301 be considered inlining candidate.
13302 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
13305 2018-02-12 Richard Biener <rguenther@suse.de>
13307 PR tree-optimization/84037
13308 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
13309 matched stmts if we cannot swap the non-matched ones.
13311 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
13313 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
13314 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
13315 _mm_maskz_scalef_round_ss): New intrinsics.
13316 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
13317 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
13318 __builtin_ia32_scalefss_round): Remove.
13319 (__builtin_ia32_scalefsd_mask_round,
13320 __builtin_ia32_scalefss_mask_round): New intrinsics.
13321 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
13322 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
13323 ((match_operand:VF_128 2 "<round_nimm_predicate>"
13324 "<round_constraint>")): Changed to ...
13325 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
13326 "<round_scalar_constraint>")): ... this.
13327 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
13328 %0, %1, %2<round_op3>}"): Changed to ...
13329 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
13330 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
13331 %2<round_scalar_mask_op3>}"): ... this.
13332 * config/i386/subst.md (round_scalar_nimm_predicate): New.
13334 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
13336 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
13337 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
13338 (_mm_maskz_sqrt_round_ss): New intrinsics.
13339 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
13340 (__builtin_ia32_sqrtsd_mask_round)
13341 (__builtin_ia32_sqrtss_mask_round): New builtins.
13342 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
13343 (__builtin_ia32_sqrtss_round): Remove.
13344 (__builtin_ia32_sqrtsd_mask_round)
13345 (__builtin_ia32_sqrtss_mask_round): New builtins.
13346 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
13347 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
13348 ((match_operand:VF_128 1 "vector_operand"
13349 "xBm,<round_constraint>")): Changed to ...
13350 ((match_operand:VF_128 1 "vector_operand"
13351 "xBm,<round_scalar_constraint>")): ... this.
13352 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
13353 %0, %2, %<iptr>1<round_op3>}): Changed to ...
13354 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
13355 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
13356 %<iptr>1<round_scalar_mask_op3>}): ... this.
13357 ((set_attr "prefix" "<round_prefix>")): Changed to ...
13358 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
13360 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
13363 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
13364 Cast vec_cmpeq result to correct type.
13365 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
13366 Cast vec_cmpgt result to correct type.
13368 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
13370 * final.c (final_scan_insn_1): Renamed from...
13371 (final_scan_insn): ... this. New wrapper, to recover
13372 seen from the outermost call in recursive ones.
13373 * config/sparc/sparc.c (output_return): Drop seen from call.
13374 (output_sibcall): Likewise.
13375 * config/visium/visium.c (output_branch): Likewise.
13377 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
13379 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
13382 2018-02-10 Alan Modra <amodra@gmail.com>
13385 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
13386 Specify LR as an input.
13388 2018-02-10 Jakub Jelinek <jakub@redhat.com>
13391 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
13392 remove_member_access_dummy_vars): New functions.
13393 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
13394 lower_omp_1, execute_lower_omp): Use them.
13396 PR rtl-optimization/84308
13397 * shrink-wrap.c (spread_components): Release todo vector.
13399 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
13401 PR rtl-optimization/57193
13402 * ira-color.c (struct allocno_color_data): Add member
13403 conflict_allocno_hard_prefs.
13404 (update_conflict_allocno_hard_prefs): New.
13405 (bucket_allocno_compare_func): Add a preference based on
13406 conflict_allocno_hard_prefs.
13407 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
13408 (color_allocnos): Remove a dead code. Initiate
13409 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
13411 2018-02-09 Jakub Jelinek <jakub@redhat.com>
13414 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
13415 constraint from =wa to wa. Avoid a subreg on the output operand,
13416 instead use a pseudo and subreg it in a move.
13417 (p9_xxbrd_<mode>): Changed to ...
13418 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
13419 (p9_xxbrd_v2df): New expander.
13420 (p9_xxbrw_<mode>): Changed to ...
13421 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
13422 (p9_xxbrw_v4sf): New expander.
13424 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
13426 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
13428 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
13431 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
13432 multiply in 32-bit mode.
13433 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
13434 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
13437 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
13439 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
13440 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
13441 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
13442 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
13444 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
13447 * dwarf2out.c (is_trivial_indirect_ref): New function.
13448 (dwarf2out_late_global_decl): Do not generate a location
13449 attribute for variables that have a non-trivial DECL_VALUE_EXPR
13450 and that are not defined in the current unit.
13452 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
13454 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
13455 instead of a libcall for UNORDERED.
13457 2018-02-09 Tamar Christina <tamar.christina@arm.com>
13460 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
13461 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
13463 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13466 * config/s390/s390.c (s390_set_current_function): Invoke
13467 s390_indirect_branch_settings also if fndecl didn't change.
13469 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
13471 * config/rs6000/rs6000.md (blockage): Set length to zero.
13473 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
13475 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
13477 2018-02-09 Jakub Jelinek <jakub@redhat.com>
13480 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
13481 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
13485 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
13486 PARALLEL incoming that failed vt_get_decl_and_offset check.
13488 PR middle-end/84237
13489 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
13490 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
13492 (get_variable_section): For decls in named .bss* sections pass true as
13493 second argument to bss_initializer_p.
13495 2018-02-09 Marek Polacek <polacek@redhat.com>
13496 Jakub Jelinek <jakub@redhat.com>
13499 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
13500 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
13501 Sync some changes from cxx_fold_indirect_ref.
13503 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
13505 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
13507 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
13508 (BLOCK_INLINE_ENTRY_LABEL): New.
13509 (dwarf2out_var_location): Disregard inline entry markers.
13510 (inline_entry_data): New struct.
13511 (inline_entry_data_hasher): New hashtable type.
13512 (inline_entry_data_hasher::hash): New.
13513 (inline_entry_data_hasher::equal): New.
13514 (inline_entry_data_table): New variable.
13515 (add_high_low_attributes): Add DW_AT_entry_pc and
13516 DW_AT_GNU_entry_view attributes if a pending entry is found
13517 in inline_entry_data_table. Add old entry_pc attribute only
13518 if debug nonbinding markers are disabled.
13519 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
13520 markers are enabled.
13521 (block_within_block_p, dwarf2out_inline_entry): New.
13522 (dwarf2out_finish): Check that no entries remained in
13523 inline_entry_data_table.
13524 * final.c (reemit_insn_block_notes): Handle inline entry notes.
13525 (final_scan_insn, notice_source_line): Likewise.
13526 (rest_of_clean_state): Skip inline entry markers.
13527 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
13529 * gimple.c (gimple_build_debug_inline_entry): New.
13530 * gimple.h (enum gimple_debug_subcode): Add
13531 GIMPLE_DEBUG_INLINE_ENTRY.
13532 (gimple_build_debug_inline_entry): Declare.
13533 (gimple_debug_inline_entry_p): New.
13534 (gimple_debug_nonbind_marker_p): Adjust.
13535 * insn-notes.def (INLINE_ENTRY): New.
13536 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
13537 inline entry marker notes.
13538 (print_insn): Likewise.
13539 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
13540 (INSN_DEBUG_MARKER_KIND): Likewise.
13541 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
13542 * tree-inline.c (expand_call_inline): Build and insert
13543 debug_inline_entry stmt.
13544 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
13545 inline entry blocks early, if nonbind markers are enabled.
13546 (dump_scope_block): Dump fragment info.
13547 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
13548 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
13549 (gimple_build_debug_inline_entry): New.
13550 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
13551 Enable/disable inline entry points too.
13552 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
13553 (DEBUG_INSN): Describe inline entry markers.
13555 * common.opt (gvariable-location-views): New.
13556 (gvariable-location-views=incompat5): New.
13557 * config.in: Rebuilt.
13558 * configure: Rebuilt.
13559 * configure.ac: Test assembler for view support.
13560 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
13561 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
13562 * dwarf2out.c (var_loc_view): New typedef.
13563 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
13564 (dwarf2out_locviews_in_attribute): New.
13565 (dwarf2out_locviews_in_loclist): New.
13566 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
13567 (enum dw_line_info_opcode): Add LI_adv_address.
13568 (struct dw_line_info_table): Add view.
13569 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
13570 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
13571 (zero_view_p): New variable.
13572 (ZERO_VIEW_P): New macro.
13573 (output_asm_line_debug_info): New.
13574 (struct var_loc_node): Add view.
13575 (add_AT_view_list, AT_loc_list): New.
13576 (add_var_loc_to_decl): Add view param. Test it against last.
13577 (new_loc_list): Add view params. Record them.
13578 (AT_loc_list_ptr): Handle loc and view lists.
13579 (view_list_to_loc_list_val_node): New.
13580 (print_dw_val): Handle dw_val_class_view_list.
13581 (size_of_die): Likewise.
13582 (value_format): Likewise.
13583 (loc_list_has_views): New.
13584 (gen_llsym): Set vl_symbol too.
13585 (maybe_gen_llsym, skip_loc_list_entry): New.
13586 (dwarf2out_maybe_output_loclist_view_pair): New.
13587 (output_loc_list): Output view list or entries too.
13588 (output_view_list_offset): New.
13589 (output_die): Handle dw_val_class_view_list.
13590 (output_dwarf_version): New.
13591 (output_compilation_unit_header): Use it.
13592 (output_skeleton_debug_sections): Likewise.
13593 (output_rnglists, output_line_info): Likewise.
13594 (output_pubnames, output_aranges): Update version comments.
13595 (output_one_line_info_table): Output view numbers in asm comments.
13596 (dw_loc_list): Determine current endview, pass it to new_loc_list.
13597 Call maybe_gen_llsym.
13598 (loc_list_from_tree_1): Adjust.
13599 (add_AT_location_description): Create view list attribute if
13600 needed, check it's absent otherwise.
13601 (convert_cfa_to_fb_loc_list): Adjust.
13602 (maybe_emit_file): Call output_asm_line_debug_info for test.
13603 (dwarf2out_var_location): Reset views as needed. Precompute
13604 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
13605 attribute. Set view.
13606 (new_line_info_table): Reset next view.
13607 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
13608 (dwarf2out_source_line): Likewise. Output view resets and labels to
13609 the assembler, or select appropriate line info opcodes.
13610 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
13611 (optimize_string_length): Catch it. Adjust.
13612 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
13613 dw_val_class_view_list, and remove it if no longer needed.
13614 (hash_loc_list): Hash view numbers.
13615 (loc_list_hasher::equal): Compare them.
13616 (optimize_location_lists): Check whether a view list symbol is
13617 needed, and whether the locview attribute is present, and
13618 whether they match. Remove the locview attribute if no longer
13620 (index_location_lists): Call skip_loc_list_entry for test.
13621 (dwarf2out_finish): Call output_asm_line_debug_info for test.
13622 Use output_dwarf_version.
13623 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
13624 (struct dw_val_node): Add val_view_list.
13625 * final.c (SEEN_NEXT_VIEW): New.
13626 (set_next_view_needed): New.
13627 (clear_next_view_needed): New.
13628 (maybe_output_next_view): New.
13629 (final_start_function): Rename to...
13630 (final_start_function_1): ... this. Take pointer to FIRST,
13631 add SEEN parameter. Emit param bindings in the initial view.
13632 (final_start_function): Reintroduce SEEN-less interface.
13633 (final): Rename to...
13634 (final_1): ... this. Take SEEN parameter. Output final pending
13635 next view at the end.
13636 (final): Reintroduce seen-less interface.
13637 (final_scan_insn): Output pending next view before switching
13638 sections or ending a block. Mark the next view as needed when
13639 outputting variable locations. Notify debug backend of section
13640 changes, and of location view changes.
13641 (rest_of_handle_final): Adjust.
13642 * toplev.c (process_options): Autodetect value for debug variable
13643 location views option. Warn on incompat5 without -gdwarf-5.
13644 * doc/invoke.texi (gvariable-location-views): New.
13645 (gvariable-location-views=incompat5): New.
13646 (gno-variable-location-views): New.
13648 2018-02-08 David Malcolm <dmalcolm@redhat.com>
13650 PR tree-optimization/84136
13651 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
13652 that the result of find_edge is non-NULL.
13654 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
13657 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
13658 storing integer register in SImode. Fix cost of 256 and 512
13659 byte aligned SSE register store.
13661 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
13663 * config/i386/i386.c (ix86_multiplication_cost): Fix
13664 multiplication cost for TARGET_AVX512DQ.
13666 2018-02-08 Marek Polacek <polacek@redhat.com>
13668 PR tree-optimization/84238
13669 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
13672 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
13674 PR tree-optimization/84265
13675 * tree-vect-stmts.c (vectorizable_store): Don't treat
13676 VMAT_CONTIGUOUS accesses as grouped.
13677 (vectorizable_load): Likewise.
13679 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
13681 PR tree-optimization/81635
13682 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
13683 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
13684 (test_round_for_mask): New functions.
13685 (wide_int_cc_tests): Call test_round_for_mask.
13686 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
13687 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
13688 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
13689 range returned by get_range_info.
13691 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
13694 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
13695 * symtab.c: Include builtins.h
13696 (symtab_node::output_to_lto_symbol_table_p): Move here
13697 from lto-streamer-out.c:output_symbol_p.
13698 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
13699 (output_symbol_p): Move all logic to symtab.c
13700 (produce_symtab): Update.
13702 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13704 * config/s390/s390-opts.h (enum indirect_branch): Define.
13705 * config/s390/s390-protos.h (s390_return_addr_from_memory)
13706 (s390_indirect_branch_via_thunk)
13707 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
13708 (enum s390_indirect_branch_type): Define.
13709 * config/s390/s390.c (struct s390_frame_layout, struct
13710 machine_function): Remove.
13711 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
13712 (indirect_branch_table_label_no, indirect_branch_table_name):
13714 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
13715 (enum s390_indirect_branch_option): Define.
13716 (s390_return_addr_from_memory): New function.
13717 (s390_handle_string_attribute): New function.
13718 (s390_attribute_table): Add new attribute handler.
13719 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
13720 (s390_indirect_branch_via_thunk): New function.
13721 (s390_indirect_branch_via_inline_thunk): New function.
13722 (s390_function_ok_for_sibcall): When jumping via thunk disallow
13723 sibling call optimization for non z10 compiles.
13724 (s390_emit_call): Force indirect branch target to be a single
13725 register. Add r1 clobber for non-z10 compiles.
13726 (s390_emit_epilogue): Emit return jump via return_use expander.
13727 (s390_reorg): Handle JUMP_INSNs as execute targets.
13728 (s390_option_override_internal): Perform validity checks for the
13729 new command line options.
13730 (s390_indirect_branch_attrvalue): New function.
13731 (s390_indirect_branch_settings): New function.
13732 (s390_set_current_function): Invoke s390_indirect_branch_settings.
13733 (s390_output_indirect_thunk_function): New function.
13734 (s390_code_end): Implement target hook.
13735 (s390_case_values_threshold): Implement target hook.
13736 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
13738 * config/s390/s390.h (struct s390_frame_layout)
13739 (struct machine_function): Move here from s390.c.
13740 (TARGET_INDIRECT_BRANCH_NOBP_RET)
13741 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
13742 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
13743 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
13744 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
13745 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
13746 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
13747 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
13748 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
13749 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
13750 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
13751 (mnemonic attribute): Add values which aren't recognized
13753 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
13754 pattern for branch conversion. Fix mnemonic attribute.
13755 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
13756 indirect branch via thunk if requested.
13757 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
13758 ("*indirect_jump"): Disable for branch conversion using out of
13760 ("indirect_jump_via_thunk<mode>_z10")
13761 ("indirect_jump_via_thunk<mode>")
13762 ("indirect_jump_via_inlinethunk<mode>_z10")
13763 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
13764 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
13765 ("casesi_jump_via_inlinethunk<mode>_z10")
13766 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
13767 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
13768 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
13769 ("*indirect2_jump"): Disable for branch conversion.
13770 ("casesi_jump"): Turn into expander and expand patterns for branch
13772 ("return_use"): New expander.
13773 ("*return"): Emit return via thunk and rename it to ...
13774 ("*return<mode>"): ... this one.
13775 * config/s390/s390.opt: Add new options and and enum for the
13778 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
13780 * lra-constraints.c (match_reload): Unconditionally use
13781 gen_lowpart_SUBREG, rather than selecting between that
13782 and equivalent gen_rtx_SUBREG code.
13784 2018-02-08 Richard Biener <rguenther@suse.de>
13786 PR tree-optimization/84233
13787 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
13788 changed flag instead of boguously re-using phi_inserted.
13790 2018-02-08 Martin Jambor <mjambor@suse.cz>
13792 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
13793 static local variables.
13795 2018-02-08 Richard Biener <rguenther@suse.de>
13797 PR tree-optimization/84278
13798 * tree-vect-stmts.c (vectorizable_store): When looking for
13799 smaller vector types to perform grouped strided loads/stores
13800 make sure the mode is supported by the target.
13801 (vectorizable_load): Likewise.
13803 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
13805 * config/aarch64/aarch64.c (aarch64_components_for_bb):
13806 Increase LDP/STP opportunities by adding adjacent callee-saves.
13808 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
13810 PR rtl-optimization/84068
13811 PR rtl-optimization/83459
13812 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
13814 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
13816 PR tree-optimization/84224
13817 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
13818 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
13819 non-zero arguments.
13821 2018-02-07 Iain Sandoe <iain@codesourcery.com>
13824 * config/rs6000/altivec.md (*restore_world): Remove LR use.
13825 * config/rs6000/predicates.md (restore_world_operation): Adjust op
13826 count, remove one USE.
13828 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13830 * doc/install.texi (Configuration): Document the
13831 --with-long-double-format={ibm,ieee} PowerPC configuration
13835 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
13836 Convert from define_expand to be define_insn_and_split. Rework
13837 float/double/_Float128 conversions to QI/HI/SImode to work with
13838 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
13839 conversions to QI/HImode types did a store and then a load to
13840 truncate the value. For conversions to VSX registers, don't split
13841 the insn, instead emit the code directly. Use the code iterator
13842 any_fix to combine signed and unsigned conversions.
13843 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
13844 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
13845 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
13846 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
13847 (fix_<mode>di2_hw): Likewise.
13848 (fixuns_<mode>di2_hw): Likewise.
13849 (fix_<mode>si2_hw): Likewise.
13850 (fixuns_<mode>si2_hw): Likewise.
13851 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
13852 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
13853 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
13854 fix<uns>_trunc<SFDF:mode>si2_p8.
13855 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
13856 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
13857 (fix<uns>_<mode>_mem): Likewise.
13858 (fctiw<u>z_<mode>_mem): Likewise.
13859 (fix<uns>_<mode>_mem): Likewise.
13860 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
13861 the register allocator from doing a direct move to the GPRs to do
13862 a store, and instead use the ISA 3.0 store byte/half-word from
13863 vector register instruction. For IEEE 128-bit floating point,
13864 also optimize stores of 32-bit ints.
13865 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
13867 2018-02-07 Alan Hayward <alan.hayward@arm.com>
13869 * genextract.c (push_pathstr_operand): New function to support
13871 (walk_rtx): Call push_pathstr_operand.
13872 (print_path): Support [a-zA-Z].
13874 2018-02-07 Richard Biener <rguenther@suse.de>
13876 PR tree-optimization/84037
13877 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
13878 (cse_and_gimplify_to_preheader): Declare.
13879 (vect_get_place_in_interleaving_chain): Likewise.
13880 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
13882 (_loop_vec_info::~_loop_vec_info): Delete it.
13883 (cse_and_gimplify_to_preheader): New function.
13884 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
13885 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
13886 (vectorizable_load): Likewise. For grouped stores always base
13887 the IV on the first element.
13888 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
13889 condition before gimplifying.
13891 2018-02-07 Jakub Jelinek <jakub@redhat.com>
13893 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
13894 *DIV_EXPR and *MOD_EXPR.
13896 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
13899 * config/i386/i386.c (ix86_option_override_internal): Mask out
13900 the CF_SET bit when checking -fcf-protection.
13902 2018-02-07 Tom de Vries <tom@codesourcery.com>
13905 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
13908 2018-02-07 Richard Biener <rguenther@suse.de>
13910 PR tree-optimization/84204
13911 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
13914 PR tree-optimization/84205
13915 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
13916 special-case isl_ast_op_zdiv_r.
13918 PR tree-optimization/84223
13919 * graphite-scop-detection.c (gather_bbs::before_dom_children):
13920 Only add conditions from within the region.
13921 (gather_bbs::after_dom_children): Adjust.
13923 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
13926 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
13927 * config/avr/avr.md: Only post-reload split REG-REG moves if
13928 either register is GENERAL_REG_P.
13930 2018-02-07 Jakub Jelinek <jakub@redhat.com>
13932 PR tree-optimization/84235
13933 * tree-ssa-scopedtables.c
13934 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
13935 if the subtraction is performed in floating point type where NaNs are
13936 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
13937 build 1. Formatting fix.
13939 2018-02-06 Jakub Jelinek <jakub@redhat.com>
13942 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
13943 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
13944 and skip it regardless of bb boundaries. Use CALL_P macro,
13945 don't test INSN_P (insn) together with CALL_P or JUMP_P check
13946 unnecessarily, formatting fix.
13948 2018-02-06 Michael Collison <michael.collison@arm.com>
13950 * config/arm/thumb2.md:
13951 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
13952 (*thumb_mov_notscc): Ditto.
13954 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
13957 * config/rs6000/rs6000.md (su code attribute): Use "u" for
13958 unsigned_fix, not "s".
13960 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13962 * configure.ac (gcc_fn_eh_frame_ro): New function.
13963 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
13964 correct .eh_frame permissions.
13965 * configure: Regenerate.
13967 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
13969 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
13970 irrelevant options.
13972 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13974 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13975 Display warning message for -mno-speculate-indirect-jumps.
13977 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
13979 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
13981 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
13983 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
13985 PR tree-optimization/84225
13986 * tree-eh.c (find_trapping_overflow): Only call
13987 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
13989 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
13992 * config/i386/i386.c: Reimplement the check of possible options
13993 -mibt/-mshstk conbination. Change error messages.
13994 * doc/invoke.texi: Fix a typo: remove extra '='.
13996 2018-02-06 Marek Polacek <polacek@redhat.com>
13998 PR tree-optimization/84228
13999 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
14001 2018-02-06 Tamar Christina <tamar.christina@arm.com>
14004 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
14005 emitted arch directives.
14006 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
14007 __ARM_FEATURE_COPROC before changing architectures.
14009 2018-02-06 Richard Biener <rguenther@suse.de>
14011 * config/i386/i386.c (print_reg): Fix typo.
14012 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
14014 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
14016 * configure: Regenerate.
14018 2018-02-05 Martin Sebor <msebor@redhat.com>
14020 PR tree-optimization/83369
14021 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
14024 2018-02-05 Martin Liska <mliska@suse.cz>
14026 * doc/invoke.texi: Cherry-pick upstream r323995.
14028 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
14030 * ira.c (ira_init_register_move_cost): Adjust comment.
14032 2018-02-05 Martin Liska <mliska@suse.cz>
14034 PR gcov-profile/84137
14035 * doc/gcov.texi: Fix typo in documentation.
14037 2018-02-05 Martin Liska <mliska@suse.cz>
14039 PR gcov-profile/83879
14040 * doc/gcov.texi: Document necessity of --dynamic-list-data when
14041 using dlopen functionality.
14043 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
14045 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
14046 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
14047 _mm_maskz_range_ss, _mm_mask_range_round_ss,
14048 _mm_maskz_range_round_ss): New intrinsics.
14049 (__builtin_ia32_rangesd128_round)
14050 (__builtin_ia32_rangess128_round): Remove.
14051 (__builtin_ia32_rangesd128_mask_round,
14052 __builtin_ia32_rangess128_mask_round): New builtins.
14053 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
14054 __builtin_ia32_rangess128_round): Remove.
14055 (__builtin_ia32_rangesd128_mask_round,
14056 __builtin_ia32_rangess128_mask_round): New builtins.
14057 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
14058 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
14059 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
14060 "<round_saeonly_constraint>")): Changed to ...
14061 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
14062 "<round_saeonly_scalar_constraint>")): ... this.
14063 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
14064 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
14065 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
14066 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
14067 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
14069 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
14071 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
14073 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
14074 Remove all values except native, 8540 and 8548.
14076 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
14078 * config/i386/i386.c (ix86_output_function_return): Pass
14079 INVALID_REGNUM, instead of -1, as invalid register number to
14080 indirect_thunk_name and output_indirect_thunk.
14082 2018-02-02 Julia Koval <julia.koval@intel.com>
14084 * config.gcc: Add -march=icelake.
14085 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
14086 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
14087 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
14088 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
14089 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
14090 (processor_target_table): Add icelake.
14091 (ix86_option_override_internal): Handle new PTAs.
14092 (get_builtin_code_for_version): Handle icelake.
14093 (M_INTEL_COREI7_ICELAKE): New.
14094 (fold_builtin_cpu): Handle icelake.
14095 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
14096 * doc/invoke.texi: Add -march=icelake.
14098 2018-02-02 Julia Koval <julia.koval@intel.com>
14100 * config/i386/i386.c (ix86_option_override_internal): Change flags type
14101 to wide_int_bitmask.
14102 * wide-int-bitmask.h: New.
14104 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
14107 * config/i386/i386.md: Replace Pmode with word_mode in
14108 builtin_setjmp_setup and builtin_longjmp to support x32.
14110 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
14114 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
14116 (rs6000_supported_cpu_names): New static variable.
14117 (linux_cpu_translation_table): Likewise.
14118 (elf_platform) <cpu>: Define new static variable and use it.
14119 Translate kernel AT_PLATFORM name to canonical name if needed.
14120 Error if platform name is unknown.
14122 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
14125 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
14127 2018-02-01 Jeff Law <law@redhat.com>
14130 * config/i386/i386.c (release_scratch_register_on_entry): Add new
14131 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
14132 the scratch if RELEASE_VIA_POP is false.
14133 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
14134 If we have to save a temporary register, decrement SIZE appropriately.
14135 Pass new arguments to release_scratch_register_on_entry.
14136 (ix86_adjust_stack_and_probe): Likewise.
14137 (ix86_emit_probe_stack_range): Pass new arguments to
14138 release_scratch_register_on_entry.
14140 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
14142 PR rtl-optimization/84157
14143 * combine.c (change_zero_ext): Use REG_P predicate in
14144 front of HARD_REGISTER_P predicate.
14146 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
14148 * config/avr/avr.c (avr_option_override): Move disabling of
14149 -fdelete-null-pointer-checks to...
14150 * common/config/avr/avr-common.c (avr_option_optimization_table):
14153 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
14155 PR tree-optimization/81635
14156 * tree-data-ref.c (split_constant_offset_1): For types that
14157 wrap on overflow, try to use range info to prove that wrapping
14160 2018-02-01 Renlin Li <renlin.li@arm.com>
14163 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
14164 TAILCALL_ADDR_REGS.
14165 (aarch64_register_move_cost): Likewise.
14166 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
14167 TAILCALL_ADDR_REGS.
14168 (REG_CLASS_NAMES): Likewise.
14169 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
14170 TAILCALL_ADDR_REGS. Remove IP registers.
14171 * config/aarch64/aarch64.md (Ucs): Update register constraint.
14173 2018-02-01 Richard Biener <rguenther@suse.de>
14175 * domwalk.h (dom_walker::dom_walker): Add additional constructor
14176 for specifying RPO order and allow NULL for that.
14177 * domwalk.c (dom_walker::dom_walker): Likewise.
14178 (dom_walker::walk): Handle NULL RPO order.
14179 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
14181 (rewrite_update_dom_walker): Likewise.
14182 (mark_def_dom_walker): Likewise.
14184 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
14186 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
14187 (aarch64_maybe_expand_sve_subreg_move): Declare.
14188 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
14189 * config/aarch64/predicates.md (aarch64_any_register_operand): New
14191 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
14192 that are semantically a reverse operation.
14193 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
14194 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
14195 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
14197 (aarch64_can_change_mode_class): For big-endian, forbid changes
14198 between two SVE modes if they have different element sizes.
14200 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
14202 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
14203 the TImode handling for big-endian targets.
14205 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
14207 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
14208 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
14210 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
14211 Remove BSWAP handing for big-endian targets and use the form of
14212 LD1RQ appropariate for the mode.
14214 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
14216 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
14217 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
14218 duplicated element.
14220 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
14223 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
14224 check for operands that need to go through aarch64_sve_reload_be.
14226 2018-02-01 Jakub Jelinek <jakub@redhat.com>
14228 PR tree-optimization/81661
14229 PR tree-optimization/84117
14230 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
14231 * tree-eh.c: Include gimplify.h.
14232 (find_trapping_overflow, replace_trapping_overflow,
14233 rewrite_to_non_trapping_overflow): New functions.
14234 * tree-vect-loop.c: Include tree-eh.h.
14235 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
14236 * tree-data-ref.c: Include tree-eh.h.
14237 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
14239 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
14241 PR rtl-optimization/84123
14242 * combine.c (change_zero_ext): Check if hard register satisfies
14243 can_change_dest_mode before calling gen_lowpart_SUBREG.
14245 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
14248 * ira.c (ira_init_register_move_cost): Remove assert.
14250 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
14252 PR rtl-optimization/84071
14253 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
14254 * doc/tm.texi: Regenerate.
14256 2018-01-31 Richard Biener <rguenther@suse.de>
14258 PR tree-optimization/84132
14259 * tree-data-ref.c (analyze_miv_subscript): Properly
14260 check whether evolution_function_is_affine_multivariate_p
14261 before calling gcd_of_steps_may_divide_p.
14263 2018-01-31 Julia Koval <julia.koval@intel.com>
14266 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
14267 * config/i386/i386.md (rdpid_rex64) New.
14268 (rdpid): Make 32bit only.
14270 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
14273 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
14274 an IDENTIFIER_NODE for FUNCTION_TYPE's.
14276 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
14279 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
14281 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
14283 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
14285 PR rtl-optimization/84071
14286 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
14287 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
14289 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
14291 * config/arc/arc.c (arc_handle_aux_attribute): New function.
14292 (arc_attribute_table): Add 'aux' attribute.
14293 (arc_in_small_data_p): Consider aux like variables.
14294 (arc_is_aux_reg_p): New function.
14295 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
14296 (arc_get_aux_arg): New function.
14297 (prepare_move_operands): Handle aux-register access.
14298 (arc_handle_aux_attribute): New function.
14299 * doc/extend.texi (ARC Variable attributes): Add subsection.
14301 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
14303 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
14304 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
14305 (arc_attribute_table): Add 'uncached' attribute.
14306 (arc_print_operand): Print '.di' flag for uncached memory
14308 (arc_in_small_data_p): Do not consider for small data the uncached
14310 (arc_is_uncached_mem_p): New function.
14311 * config/arc/predicates.md (compact_store_memory_operand): Check
14312 for uncached memory accesses.
14313 (nonvol_nonimm_operand): Likewise.
14314 * gcc/doc/extend.texi (ARC Type Attribute): New subsection.
14316 2018-01-31 Jakub Jelinek <jakub@redhat.com>
14319 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
14320 falign-loops=): Add Optimization flag.
14322 2018-01-30 Jeff Law <law@redhat.com>
14325 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
14326 INT_REGISTERS_SAVED. Check it prior to calling
14327 get_scratch_register_on_entry.
14328 (ix86_adjust_stack_and_probe): Similarly.
14329 (ix86_emit_probe_stack_range): Similarly.
14330 (ix86_expand_prologue): Corresponding changes.
14332 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14335 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
14336 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
14338 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
14341 * lra-constraints.c (curr_insn_transform): Process AND in the
14344 2018-01-30 Jakub Jelinek <jakub@redhat.com>
14346 PR rtl-optimization/83986
14347 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
14348 dependence against last_pending_memory_flush in addition to
14349 pending_jump_insns.
14351 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
14353 PR tree-optimization/81611
14354 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
14357 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14360 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
14363 2018-01-30 Richard Biener <rguenther@suse.de>
14364 Jakub Jelinek <jakub@redhat.com>
14366 PR tree-optimization/84111
14367 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
14368 inner loops added during recursion, as they don't have up-to-date
14371 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
14374 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
14375 (can_inline_edge_by_limits_p): ... here.
14376 (can_early_inline_edge_p, check_callers,
14377 update_caller_keys, update_callee_keys, recursive_inlining,
14378 add_new_edges_to_heap, speculation_useful_p,
14379 inline_small_functions,
14380 inline_small_functions, flatten_function,
14381 inline_to_all_callers_1): Update.
14383 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
14385 * profile-count.c (profile_count::combine_with_ipa_count): Handle
14388 2018-01-30 Richard Biener <rguenther@suse.de>
14390 PR tree-optimization/83008
14391 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
14392 invariant and constant vector uses in stmts when they need
14393 more than one stmt.
14395 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14398 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
14399 * configure: Regenerate.
14401 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
14403 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
14405 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
14406 Use gen_rtx_REG rather than gen_lowpart.
14408 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
14410 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
14411 rather than 0 when creating partial subregs.
14413 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
14415 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
14418 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
14421 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
14422 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
14423 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
14424 flags. This restores the settings used before the 2017-07-24.
14425 Turning off pre increment/decrement/modify allows IVOPTS to
14426 optimize DF/SF loops where the index is an int.
14428 2018-01-29 Richard Biener <rguenther@suse.de>
14429 Kelvin Nilsen <kelvin@gcc.gnu.org>
14432 * tree-vect-stmts.c (vectorizable_call): Don't call
14433 targetm.vectorize_builtin_md_vectorized_function if callee is
14436 2018-01-22 Carl Love <cel@us.ibm.com>
14438 * doc/extend.tex: Fix typo in second arg in
14439 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
14441 2018-01-29 Richard Biener <rguenther@suse.de>
14443 PR tree-optimization/84086
14444 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
14445 (flush_ssaname_freelist): When SSA names were released reset
14446 the SCEV hash table.
14448 2018-01-29 Richard Biener <rguenther@suse.de>
14450 PR tree-optimization/84057
14451 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
14452 removed paths when removing edges.
14454 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
14456 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
14457 -mfunction-return=@var{choice}.
14459 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
14461 PR diagnostic/84034
14462 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
14463 Handle CR like TAB.
14464 (layout::print_source_line): Likewise.
14465 (test_get_line_width_without_trailing_whitespace): Add test cases.
14467 2018-01-27 Jakub Jelinek <jakub@redhat.com>
14469 PR middle-end/84040
14470 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
14473 2018-01-26 Jim Wilson <jimw@sifive.com>
14475 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
14477 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
14480 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14482 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
14483 and CMP + SUB-immediate -> SUBS.
14485 2018-01-26 Martin Sebor <msebor@redhat.com>
14487 PR tree-optimization/83896
14488 * tree-ssa-strlen.c (get_string_len): Rename...
14489 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
14490 Avoid assuming length is constant.
14491 (handle_char_store): Use HOST_WIDE_INT for string length.
14493 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
14496 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
14497 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
14499 2018-01-26 Richard Biener <rguenther@suse.de>
14501 PR rtl-optimization/84003
14502 * dse.c (record_store): Only record redundant stores when
14503 the earlier store aliases at least all accesses the later one does.
14505 2018-01-26 Jakub Jelinek <jakub@redhat.com>
14507 PR rtl-optimization/83985
14508 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
14509 REG_CFA_RESTORE insns.
14510 (delete_unmarked_insns): Don't ignore separate shrink wrapping
14511 REG_CFA_RESTORE insns here.
14514 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
14515 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
14517 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
14519 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
14520 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
14521 (arc_init): Likewise.
14522 (arc_override_options): Likewise.
14523 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
14525 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
14527 * config/arc/arc.h (TARGET_DBNZ): Define.
14528 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
14529 properly set the tune attribute.
14530 (dbnz): Use TARGET_DBNZ guard.
14531 * config/arc/arc.opt (mtune): Add core3 option.
14533 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
14535 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
14536 recognize new pic like addresses.
14537 (arc_delegitimize_address): Clean up.
14539 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
14541 * config/arc/arc-arches.def: Option mrf16 valid for all
14543 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
14544 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
14545 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
14546 * config/arc/arc-tables.opt: Regenerate.
14547 * config/arc/arc.c (arc_conditional_register_usage): Handle
14548 reduced register file case.
14549 (arc_file_start): Set must have build attributes.
14550 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
14551 mrf16 option value.
14552 * config/arc/arc.opt (mrf16): Add new option.
14553 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
14554 * config/arc/genmultilib.awk: Handle new mrf16 option.
14555 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
14556 * config/arc/t-multilib: Regenerate.
14557 * doc/invoke.texi (ARC Options): Document mrf16 option.
14559 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
14561 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
14562 * config/arc/arc.c (arc_handle_secure_attribute): New function.
14563 (arc_attribute_table): Add 'secure_call' attribute.
14564 (arc_print_operand): Print secure call operand.
14565 (arc_function_ok_for_sibcall): Don't optimize tail calls when
14567 (arc_is_secure_call_p): New function. * config/arc/arc.md
14568 (call_i): Add support for sjli instruction.
14569 (call_value_i): Likewise.
14570 * config/arc/constraints.md (Csc): New constraint.
14572 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
14573 John Eric Martin <John.Martin@emmicro-us.com>
14575 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
14576 * config/arc/arc.c (_arc_jli_section): New struct.
14577 (arc_jli_section): New type.
14578 (rc_jli_sections): New static variable.
14579 (arc_handle_jli_attribute): New function.
14580 (arc_attribute_table): Add jli_always and jli_fixed attribute.
14581 (arc_file_end): New function.
14582 (TARGET_ASM_FILE_END): Define.
14583 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
14584 (arc_add_jli_section): New function.
14585 (jli_call_scan): Likewise.
14586 (arc_reorg): Call jli_call_scan.
14587 (arc_output_addsi): Remove 'S' from printing asm operand.
14588 (arc_is_jli_call_p): New function.
14589 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
14591 (movhi_insn): Likewise.
14592 (movsi_insn): Likewise.
14593 (movsi_set_cc_insn): Likewise.
14594 (loadqi_update): Likewise.
14595 (load_zeroextendqisi_update): Likewise.
14596 (load_signextendqisi_update): Likewise.
14597 (loadhi_update): Likewise.
14598 (load_zeroextendhisi_update): Likewise.
14599 (load_signextendhisi_update): Likewise.
14600 (loadsi_update): Likewise.
14601 (loadsf_update): Likewise.
14602 (movsicc_insn): Likewise.
14603 (bset_insn): Likewise.
14604 (bxor_insn): Likewise.
14605 (bclr_insn): Likewise.
14606 (bmsk_insn): Likewise.
14607 (bicsi3_insn): Likewise.
14608 (cmpsi_cc_c_insn): Likewise.
14609 (movsi_ne): Likewise.
14610 (movsi_cond_exec): Likewise.
14611 (clrsbsi2): Likewise.
14612 (norm_f): Likewise.
14622 (call_i): Remove 'S' asm letter, add jli instruction.
14623 (call_value_i): Likewise.
14624 * config/arc/arc.op (mjli-always): New option.
14625 * config/arc/constraints.md (Cji): New constraint.
14626 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
14628 (subsf3_fpx): Likewise.
14629 (mulsf3_fpx): Likewise.
14630 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
14632 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
14633 function attrbutes.
14634 * doc/invoke.texi (ARC): Document mjli-always option.
14636 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
14638 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
14639 avoid addition with 0 and use incw and decw where possible.
14641 2018-01-26 Richard Biener <rguenther@suse.de>
14643 PR tree-optimization/81082
14644 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
14645 association if it requires casting to unsigned.
14646 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
14647 from fold_plusminus_mult_expr to catch important cases late when
14648 range info is available.
14650 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14652 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
14653 * configure.ac (hidden_linkonce): New test.
14654 * configure: Regenerate.
14655 * config.in: Regenerate.
14657 2018-01-26 Julia Koval <julia.koval@intel.com>
14659 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
14660 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
14661 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
14662 _mm_mask_bitshuffle_epi64_mask): Fix type.
14663 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
14664 USI_FTYPE_V4DI_V4DI_USI): Remove.
14665 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
14666 __builtin_ia32_vpshufbitqmb256_mask,
14667 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
14668 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
14669 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
14671 2018-01-26 Alan Modra <amodra@gmail.com>
14674 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
14675 UNSPEC_VBPERMQ. Sort other unspecs.
14677 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
14679 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
14681 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
14683 PR middle-end/83055
14684 * predict.c (drop_profile): Do not push/pop cfun; update also
14686 (handle_missing_profiles): Fix logic looking for zero profiles.
14688 2018-01-25 Jakub Jelinek <jakub@redhat.com>
14690 PR middle-end/83977
14691 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
14692 on functions with #pragma omp declare simd or functions with simd
14694 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
14695 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
14696 Remove trailing \n from warning_at calls.
14698 2018-01-25 Tom de Vries <tom@codesourcery.com>
14701 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
14702 for neutered workers.
14704 2018-01-24 Joseph Myers <joseph@codesourcery.com>
14707 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
14708 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
14710 2018-01-24 Jeff Law <law@redhat.com>
14713 * i386.c (get_probe_interval): Move to earlier point.
14714 (ix86_compute_frame_layout): If -fstack-clash-protection and
14715 the frame is larger than the probe interval, then use pushes
14716 to save registers rather than reg->mem moves.
14717 (ix86_expand_prologue): Remove conditional for int_registers_saved
14720 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
14723 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
14724 min/max for never referenced object.
14726 2018-01-24 Jakub Jelinek <jakub@redhat.com>
14728 PR middle-end/83977
14729 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
14731 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
14732 attributes from DECL_ATTRIBUTES (decl) without affecting
14733 DECL_ATTRIBUTES (current_function_decl).
14734 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
14735 functions with non-NULL DECL_ABSTRACT_ORIGIN.
14737 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
14739 PR tree-optimization/83979
14740 * fold-const.c (fold_comparison): Use constant_boolean_node
14741 instead of boolean_{true,false}_node.
14743 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
14745 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
14748 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14750 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
14751 Simplify the clause that sets the length attribute.
14752 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
14753 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
14754 clause that sets the length attribute.
14755 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
14757 2018-01-24 Tom de Vries <tom@codesourcery.com>
14760 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
14761 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
14762 Add strict parameter.
14763 (prevent_branch_around_nothing): Insert dummy insn between branch to
14764 label and label with no ptx insn inbetween.
14765 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
14767 2018-01-24 Tom de Vries <tom@codesourcery.com>
14770 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
14771 for neutered threads in warp.
14772 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
14774 2018-01-24 Richard Biener <rguenther@suse.de>
14776 PR tree-optimization/83176
14777 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
14780 2018-01-24 Richard Biener <rguenther@suse.de>
14782 PR tree-optimization/82819
14783 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
14784 code generating pluses that are no-ops in the target precision.
14786 2018-01-24 Richard Biener <rguenther@suse.de>
14788 PR middle-end/84000
14789 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
14791 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
14793 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
14794 to merge probabilities.
14795 * predict.c (probably_never_executed): Also mark as cold functions
14796 with global 0 profile and guessed local profile.
14797 * profile-count.c (profile_probability::combine_with_count): New
14799 * profile-count.h (profile_probability::operator*,
14800 profile_probability::operator*=, profile_probability::operator/,
14801 profile_probability::operator/=): Reduce precision to adjusted
14802 and set value to guessed on contradictory divisions.
14803 (profile_probability::combine_with_freq): Remove.
14804 (profile_probability::combine_wiht_count): Declare.
14805 (profile_count::force_nonzero):: Set to adjusted.
14806 (profile_count::probability_in):: Set quality to adjusted.
14807 * tree-ssa-tail-merge.c (replace_block_by): Use
14808 combine_with_count.
14810 2018-01-23 Andrew Waterman <andrew@sifive.com>
14811 Jim Wilson <jimw@sifive.com>
14813 * config/riscv/riscv.c (riscv_stack_boundary): New.
14814 (riscv_option_override): Set riscv_stack_boundary. Handle
14815 riscv_preferred_stack_boundary_arg.
14816 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
14817 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
14818 (STACK_BOUNDARY): Set to riscv_stack_boundary.
14819 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
14820 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
14821 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
14823 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
14826 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
14827 of struct ix86_frame.
14828 (ix86_expand_epilogue): Likewise. Add a local variable for
14829 the reg_save_offset field in struct ix86_frame.
14831 2018-01-23 Bin Cheng <bin.cheng@arm.com>
14833 PR tree-optimization/82604
14834 * tree-loop-distribution.c (enum partition_kind): New enum item
14835 PKIND_PARTIAL_MEMSET.
14836 (partition_builtin_p): Support above new enum item.
14837 (generate_code_for_partition): Ditto.
14838 (compute_access_range): Differentiate cases that equality can be
14839 proven at all loops, the innermost loops or no loops.
14840 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
14841 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
14842 (finalize_partitions, distribute_loop): Don't fuse partition of
14843 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
14844 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
14845 parloop is enabled.
14847 2018-01-23 Martin Liska <mliska@suse.cz>
14849 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
14850 order to ignore the predictor.
14851 (PRED_POLYMORPHIC_CALL): Likewise.
14852 (PRED_RECURSIVE_CALL): Likewise.
14854 2018-01-23 Martin Liska <mliska@suse.cz>
14856 * tree-profile.c (tree_profiling): Print function header to
14857 aware reader which function we are working on.
14858 * value-prof.c (gimple_find_values_to_profile): Do not print
14859 not interesting value histograms.
14861 2018-01-23 Martin Liska <mliska@suse.cz>
14863 * profile-count.h (enum profile_quality): Add
14864 profile_uninitialized as the first value. Do not number values
14865 as they are zero based.
14866 (profile_count::verify): Update sanity check.
14867 (profile_probability::verify): Likewise.
14869 2018-01-23 Nathan Sidwell <nathan@acm.org>
14871 * doc/invoke.texi (ffor-scope): Deprecate.
14873 2018-01-23 David Malcolm <dmalcolm@redhat.com>
14875 PR tree-optimization/83510
14876 * domwalk.c (set_all_edges_as_executable): New function.
14877 (dom_walker::dom_walker): Convert bool param
14878 "skip_unreachable_blocks" to enum reachability. Move setup of
14879 edge flags to set_all_edges_as_executable and only do it when
14880 reachability is REACHABLE_BLOCKS.
14881 * domwalk.h (enum dom_walker::reachability): New enum.
14882 (dom_walker::dom_walker): Convert bool param
14883 "skip_unreachable_blocks" to enum reachability.
14884 (set_all_edges_as_executable): New decl.
14885 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
14886 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
14888 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
14889 but converting true to REACHABLE_BLOCKS.
14890 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
14892 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
14893 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
14894 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
14896 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
14897 if check_all_array_refs will be called.
14899 2018-01-23 David Malcolm <dmalcolm@redhat.com>
14901 * tree.c (selftest::test_location_wrappers): Add more test
14904 2018-01-23 David Malcolm <dmalcolm@redhat.com>
14906 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
14907 (selftest::test_bit_in_range): Likewise.
14909 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
14912 * doc/sourcebuild.texi (vect_float): Say that the selector
14913 only describes the situation when -funsafe-math-optimizations is on.
14914 (vect_float_strict): Document.
14916 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
14918 PR tree-optimization/83965
14919 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
14920 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
14921 instead of checking only for a reduction.
14922 (vect_recog_widen_sum_pattern): Likewise.
14924 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
14926 * predict.c (probably_never_executed): Only use precise profile info.
14927 (compute_function_frequency): Skip after inlining hack since we now
14928 have quality checking.
14930 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
14932 * profile-count.h (profile_probability::very_unlikely,
14933 profile_probability::unlikely, profile_probability::even): Set
14934 precision to guessed.
14936 2018-01-23 Richard Biener <rguenther@suse.de>
14938 PR tree-optimization/83963
14939 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
14940 Properly terminate dominator walk when crossing the exit edge not
14941 when visiting its source block.
14943 2018-01-23 Jakub Jelinek <jakub@redhat.com>
14946 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
14947 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
14949 2018-01-22 Jakub Jelinek <jakub@redhat.com>
14951 PR tree-optimization/83957
14952 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
14953 semicolon after for body surrounded by braces.
14955 PR tree-optimization/83081
14956 * profile-count.h (profile_probability::split): New method.
14957 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
14958 Use profile_probability::split.
14959 (do_compare_rtx_and_jump): Fix adjustment of probabilities
14960 when splitting a single conditional jump into 2.
14962 2018-01-22 David Malcolm <dmalcolm@redhat.com>
14964 PR tree-optimization/69452
14965 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
14968 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
14970 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
14971 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
14972 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
14974 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
14976 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
14978 * config/rl78/rl78.md (movdi): New define_expand.
14979 * config/rl78/rl78.c (rl78_split_movdi): New function.
14981 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
14984 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
14986 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
14987 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
14988 128-bit to produce an UNSPEC move to get the double word with the
14989 signbit and then a shift directly to do signbit.
14990 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
14991 implementation with a new version that just does either a direct
14992 move or a regular move. Move memory interface to separate insns.
14993 Move insns so they are next to the expander.
14994 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
14995 with signbit move. Split big and little endian case.
14996 (signbit<mode>2_dm_mem_le): Likewise.
14997 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
14998 (signbit<mode>2_dm2): Likewise.
15000 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
15002 * config/rl78/rl78.md (anddi3): New define_expand.
15004 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
15006 * config/rl78/rl78.md (umindi3): New define_expand.
15008 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
15010 * config/rl78/rl78.md (smindi3): New define_expand.
15012 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
15014 * config/rl78/rl78.md (smaxdi3): New define_expand.
15016 2018-01-22 Carl Love <cel@us.ibm.com>
15018 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
15019 LVX_V1TI): Add macro expansion.
15020 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
15021 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
15022 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
15023 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
15024 Change check to determine if the instruction is a byte reversing
15025 entry. Fix typo in comment.
15026 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
15027 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
15028 Add def_builtin calls for new builtins.
15029 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
15030 Add define_insn expansion.
15032 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
15034 * config/rl78/rl78.md (umaxdi3): New define_expand.
15036 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
15038 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
15039 for non-QImode registers.
15041 2018-01-22 Richard Biener <rguenther@suse.de>
15043 PR tree-optimization/83963
15044 * graphite-scop-detection.c (scop_detection::get_sese): Delay
15045 including the loop exit block.
15046 (scop_detection::merge_sese): Likewise.
15047 (scop_detection::add_scop): Do it here instead.
15049 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15051 * doc/sourcebuild.texi (arm_softfloat): Document.
15053 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
15056 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
15057 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
15058 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
15060 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15061 David Edelsohn <dje.gcc@gmail.com>
15064 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
15065 Change "crset eq" to "crset 2".
15066 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
15067 (*call_indirect_aix<mode>_nospec): Likewise.
15068 (*call_value_indirect_aix<mode>_nospec): Likewise.
15069 (*call_indirect_elfv2<mode>_nospec): Likewise.
15070 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
15071 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
15072 change assembly output from . to $.
15073 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
15074 (indirect_jump<mode>_nospec): Change assembly output from . to $.
15075 (*tablejump<mode>_internal1_nospec): Likewise.
15077 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
15080 * config/sh/sh_optimize_sett_clrt.cc:
15081 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
15083 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
15085 PR tree-optimization/83940
15086 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
15087 offset_dt to vect_constant_def rather than vect_unknown_def_type.
15088 (vect_check_load_store_mask): Add a mask_dt_out parameter and
15089 use it to pass back the definition type.
15090 (vect_check_store_rhs): Likewise rhs_dt_out.
15091 (vect_build_gather_load_calls): Add a mask_dt argument and use
15092 it instead of a call to vect_is_simple_use.
15093 (vectorizable_store): Update calls to vect_check_load_store_mask
15094 and vect_check_store_rhs. Use the dt returned by the latter instead
15095 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
15096 instead of calls to vect_is_simple_use. Pass the scalar rather
15097 than the vector operand to vect_is_simple_use when handling
15098 second and subsequent copies of an rhs value.
15099 (vectorizable_load): Update calls to vect_check_load_store_mask
15100 and vect_build_gather_load_calls. Use the cached mask_dt and
15101 gs_info.offset_dt instead of calls to vect_is_simple_use.
15103 2018-01-20 Jakub Jelinek <jakub@redhat.com>
15105 PR middle-end/83945
15106 * tree-emutls.c: Include gimplify.h.
15107 (lower_emutls_2): New function.
15108 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
15109 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
15110 it before further processing.
15113 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
15114 UINTVAL (trueop1) instead of INTVAL (op1).
15116 2018-01-19 Jakub Jelinek <jakub@redhat.com>
15120 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
15121 INCOMING_FRAME_SP_OFFSET if not defined.
15122 (scan_trace): Add ENTRY argument. If true and
15123 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
15124 emit a note to adjust the CFA offset.
15125 (create_cfi_notes): Adjust scan_trace callers.
15126 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
15127 INCOMING_FRAME_SP_OFFSET in the CIE.
15128 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
15129 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
15131 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
15132 * doc/tm.texi: Regenerated.
15134 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15136 PR rtl-optimization/83147
15137 * lra-constraints.c (remove_inheritance_pseudos): Use
15138 lra_substitute_pseudo_within_insn.
15140 2018-01-19 Tom de Vries <tom@codesourcery.com>
15141 Cesar Philippidis <cesar@codesourcery.com>
15144 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
15146 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
15149 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
15150 spaces for function labels.
15152 2018-01-19 Martin Liska <mliska@suse.cz>
15154 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
15155 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
15156 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
15157 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
15158 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
15159 (PRED_CONST_RETURN): Change from 69 to 65.
15160 (PRED_NULL_RETURN): Change from 91 to 71.
15161 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
15162 (PRED_LOOP_GUARD): Change from 66 to 73.
15164 2018-01-19 Martin Liska <mliska@suse.cz>
15166 * predict.c (predict_insn_def): Add new assert.
15167 (struct branch_predictor): Change type to signed integer.
15168 (test_prediction_value_range): Amend test to cover
15169 PROB_UNINITIALIZED.
15170 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
15171 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
15172 (PRED_LOOP_ITERATIONS_MAX): Likewise.
15173 (PRED_LOOP_IV_COMPARE): Likewise.
15174 * predict.h (PROB_UNINITIALIZED): Define new constant.
15176 2018-01-19 Martin Liska <mliska@suse.cz>
15178 * predict.c (dump_prediction): Add new format for
15179 analyze_brprob.py script which is enabled with -details
15181 * profile-count.h (precise_p): New function.
15183 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
15185 PR tree-optimization/83922
15186 * tree-vect-loop.c (vect_verify_full_masking): Return false if
15187 there are no statements that need masking.
15188 (vect_active_double_reduction_p): New function.
15189 (vect_analyze_loop_operations): Use it when handling phis that
15190 are not in the loop header.
15192 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
15194 PR tree-optimization/83914
15195 * tree-vect-loop.c (vectorizable_induction): Don't convert
15196 init_expr or apply the peeling adjustment for inductions
15197 that are nested within the vectorized loop.
15199 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15201 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
15204 2018-01-18 Jakub Jelinek <jakub@redhat.com>
15208 * function.h (gimplify_parameters): Add gimple_seq * argument.
15209 * function.c: Include gimple.h and options.h.
15210 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
15211 for the added local temporaries if needed.
15212 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
15213 if there are any parameter cleanups, wrap whole body into a
15214 try/finally with the cleanups.
15216 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
15219 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
15220 Use GET_MODE_CLASS for scalar floating point.
15222 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
15226 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
15227 Fix call of call_cgraph_insertion_hooks.
15229 2018-01-18 Martin Sebor <msebor@redhat.com>
15231 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
15233 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
15236 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
15239 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
15242 * common.opt: (-ffile-prefix-map): New option.
15243 * opts.c (common_handle_option): Defer it.
15244 * opts-global.c (handle_common_deferred_options): Handle it.
15245 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
15246 * file-prefix-map.h: New file.
15247 (remap_debug_filename, add_debug_prefix_map): ...here.
15248 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
15249 * final.c (debug_prefix_map, add_debug_prefix_map
15250 remap_debug_filename): Move to...
15251 * file-prefix-map.c: New file.
15252 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
15253 generalize, get rid of alloca(), use strrchr() instead of strchr().
15254 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
15255 Implement in terms of add_prefix_map().
15256 (remap_macro_filename, remap_debug_filename): Implement in term of
15258 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
15259 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
15260 * dbxout.c: Include file-prefix-map.h.
15261 * varasm.c: Likewise.
15262 * vmsdbgout.c: Likewise.
15263 * xcoffout.c: Likewise.
15264 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
15265 * doc/cppopts.texi (-fmacro-prefix-map): Document.
15266 * doc/invoke.texi (-ffile-prefix-map): Document.
15267 (-fdebug-prefix-map): Update description.
15269 2018-01-18 Martin Liska <mliska@suse.cz>
15271 * config/i386/i386.c (indirect_thunk_name): Document that also
15273 (output_indirect_thunk): Document why both instructions
15274 (pause and lfence) are generated.
15276 2018-01-18 Richard Biener <rguenther@suse.de>
15278 PR tree-optimization/83887
15279 * graphite-scop-detection.c
15280 (scop_detection::get_nearest_dom_with_single_entry): Remove.
15281 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
15282 (scop_detection::merge_sese): Re-implement with a flood-fill
15283 algorithm that properly finds a SESE region if it exists.
15285 2018-01-18 Jakub Jelinek <jakub@redhat.com>
15288 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
15289 pointer_diff optimizations use view_convert instead of convert.
15291 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15293 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
15294 Generate different code for -mno-speculate-indirect-jumps.
15295 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
15296 (*call_indirect_aix<mode>): Disable for
15297 -mno-speculate-indirect-jumps.
15298 (*call_indirect_aix<mode>_nospec): New define_insn.
15299 (*call_value_indirect_aix<mode>): Disable for
15300 -mno-speculate-indirect-jumps.
15301 (*call_value_indirect_aix<mode>_nospec): New define_insn.
15302 (*sibcall_nonlocal_sysv<mode>): Generate different code for
15303 -mno-speculate-indirect-jumps.
15304 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
15306 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
15308 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
15309 long double type, set the flags for noting the default long double
15310 type, even if we don't pass or return a long double type.
15312 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
15315 * ipa-inline.c (flatten_function): Do not overwrite final inlining
15318 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
15320 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
15321 support for merge[hl].
15322 (fold_mergehl_helper): New helper function.
15323 (tree-vector-builder.h): New #include for tree_vector_builder usage.
15324 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
15325 (altivec_vmrglw_direct): Add xxmrglw insn.
15327 2018-01-17 Andrew Waterman <andrew@sifive.com>
15329 * config/riscv/riscv.c (riscv_conditional_register_usage): If
15330 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
15332 2018-01-17 David Malcolm <dmalcolm@redhat.com>
15335 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
15336 call the lto_location_cache before reading the
15337 DECL_SOURCE_LOCATION of the types.
15339 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
15340 Richard Sandiford <richard.sandiford@linaro.org>
15342 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
15343 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
15344 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
15345 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
15346 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
15348 * config/aarch64/constraints.md (aarch64_movti_operand):
15350 * config/aarch64/predicates.md (Uti): Add new constraint.
15352 2018-01-17 Carl Love <cel@us.ibm.com>
15354 * config/rs6000/vsx.md (define_expand xl_len_r,
15355 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
15356 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
15358 (define_expand, define_insn): Move the shift left from the
15359 define_insn to the define_expand for lxvl and stxvl instructions.
15360 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
15361 and XL_LEN_R definitions to PURE.
15363 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
15365 * config/i386/i386.c (indirect_thunk_name): Declare regno
15366 as unsigned int. Compare regno with INVALID_REGNUM.
15367 (output_indirect_thunk): Ditto.
15368 (output_indirect_thunk_function): Ditto.
15369 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
15370 in the call to output_indirect_thunk_function.
15372 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
15374 PR middle-end/83884
15375 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
15376 rather than the size of inner_type to determine the stack slot size
15377 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
15379 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
15382 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
15385 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
15387 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
15388 endian Linux systems to optionally enable multilibs for selecting
15389 the long double type if the user configured an explicit type.
15390 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
15391 have no long double multilibs if not defined.
15392 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
15393 warn if the user used -mabi={ieee,ibm}longdouble and we built
15394 multilibs for long double.
15395 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
15396 appropriate multilib option.
15397 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
15399 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
15400 for building long double multilibs.
15401 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
15403 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
15405 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
15408 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
15410 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
15413 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
15416 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
15419 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
15421 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
15422 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
15424 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
15426 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
15427 different rtl trees depending on TARGET_64BIT.
15428 (rs6000_gen_lvx): Likewise.
15430 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
15432 * config/visium/visium.md (nop): Tweak comment.
15433 (hazard_nop): Likewise.
15435 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15437 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
15438 -mspeculate-indirect-jumps.
15439 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
15440 for -mno-speculate-indirect-jumps.
15441 (*call_indirect_elfv2<mode>_nospec): New define_insn.
15442 (*call_value_indirect_elfv2<mode>): Disable for
15443 -mno-speculate-indirect-jumps.
15444 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
15445 (indirect_jump): Emit different RTL for
15446 -mno-speculate-indirect-jumps.
15447 (*indirect_jump<mode>): Disable for
15448 -mno-speculate-indirect-jumps.
15449 (*indirect_jump<mode>_nospec): New define_insn.
15450 (tablejump): Emit different RTL for
15451 -mno-speculate-indirect-jumps.
15452 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
15453 (tablejumpsi_nospec): New define_expand.
15454 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
15455 (tablejumpdi_nospec): New define_expand.
15456 (*tablejump<mode>_internal1): Disable for
15457 -mno-speculate-indirect-jumps.
15458 (*tablejump<mode>_internal1_nospec): New define_insn.
15459 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
15462 2018-01-16 Artyom Skrobov tyomitch@gmail.com
15464 * caller-save.c (insert_save): Drop unnecessary parameter. All
15467 2018-01-16 Jakub Jelinek <jakub@redhat.com>
15468 Richard Biener <rguenth@suse.de>
15471 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
15472 return early, inline manually is_gimple_sizepos. Make sure if we
15473 call gimplify_expr we don't end up with a gimple constant.
15474 * tree.c (variably_modified_type_p): Don't return true for
15475 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
15476 * gimplify.h (is_gimple_sizepos): Remove.
15478 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
15480 PR tree-optimization/83857
15481 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
15482 vectorizable_live_operation for pure SLP statements.
15483 (vectorizable_live_operation): Handle PHIs.
15485 2018-01-16 Richard Biener <rguenther@suse.de>
15487 PR tree-optimization/83867
15488 * tree-vect-stmts.c (vect_transform_stmt): Precompute
15489 nested_in_vect_loop_p since the scalar stmt may get invalidated.
15491 2018-01-16 Jakub Jelinek <jakub@redhat.com>
15494 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
15495 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
15496 If off is not INTEGER_CST, issue a may not be aligned warning
15497 rather than isn't aligned. Use isn%'t rather than isn't.
15498 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
15500 <case MULT_EXPR>: Improve the case when bottom and one of the
15501 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
15502 operand, in that case check if the other operand is multiple of
15503 bottom divided by the INTEGER_CST operand.
15505 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
15508 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
15509 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
15510 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
15511 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
15512 * config/pa/pa.c (pa_function_arg_advance): Likewise.
15513 (pa_function_arg, pa_arg_partial_bytes): Likewise.
15514 (pa_function_arg_size): New function.
15516 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
15518 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
15519 in a separate statement.
15521 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
15523 PR tree-optimization/83847
15524 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
15525 group gathers and scatters.
15527 2018-01-16 Jakub Jelinek <jakub@redhat.com>
15529 PR rtl-optimization/86620
15530 * params.def (max-sched-ready-insns): Bump minimum value to 1.
15532 PR rtl-optimization/83213
15533 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
15534 to last if both are JUMP_INSNs.
15536 PR tree-optimization/83843
15537 * gimple-ssa-store-merging.c
15538 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
15539 store_immediate_info for bswap/nop orig_stores.
15541 2018-01-15 Andrew Waterman <andrew@sifive.com>
15543 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
15545 <UDIV>: Increase cost if !TARGET_DIV.
15547 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
15549 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
15550 (define_attr "cr_logical_3op"): New.
15551 (cceq_ior_compare): Adjust.
15552 (cceq_ior_compare_complement): Adjust.
15553 (*cceq_rev_compare): Adjust.
15554 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
15555 (is_cracked_insn): Adjust.
15556 (insn_must_be_first_in_group): Adjust.
15557 * config/rs6000/40x.md: Adjust.
15558 * config/rs6000/440.md: Adjust.
15559 * config/rs6000/476.md: Adjust.
15560 * config/rs6000/601.md: Adjust.
15561 * config/rs6000/603.md: Adjust.
15562 * config/rs6000/6xx.md: Adjust.
15563 * config/rs6000/7450.md: Adjust.
15564 * config/rs6000/7xx.md: Adjust.
15565 * config/rs6000/8540.md: Adjust.
15566 * config/rs6000/cell.md: Adjust.
15567 * config/rs6000/e300c2c3.md: Adjust.
15568 * config/rs6000/e500mc.md: Adjust.
15569 * config/rs6000/e500mc64.md: Adjust.
15570 * config/rs6000/e5500.md: Adjust.
15571 * config/rs6000/e6500.md: Adjust.
15572 * config/rs6000/mpc.md: Adjust.
15573 * config/rs6000/power4.md: Adjust.
15574 * config/rs6000/power5.md: Adjust.
15575 * config/rs6000/power6.md: Adjust.
15576 * config/rs6000/power7.md: Adjust.
15577 * config/rs6000/power8.md: Adjust.
15578 * config/rs6000/power9.md: Adjust.
15579 * config/rs6000/rs64.md: Adjust.
15580 * config/rs6000/titan.md: Adjust.
15582 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
15584 * config/i386/predicates.md (indirect_branch_operand): Rewrite
15585 ix86_indirect_branch_register logic.
15587 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
15589 * config/i386/constraints.md (Bs): Update
15590 ix86_indirect_branch_register check. Don't check
15591 ix86_indirect_branch_register with GOT_memory_operand.
15593 * config/i386/predicates.md (GOT_memory_operand): Don't check
15594 ix86_indirect_branch_register here.
15595 (GOT32_symbol_operand): Likewise.
15597 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
15599 * config/i386/predicates.md (constant_call_address_operand):
15600 Rewrite ix86_indirect_branch_register logic.
15601 (sibcall_insn_operand): Likewise.
15603 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
15605 * config/i386/constraints.md (Bs): Replace
15606 ix86_indirect_branch_thunk_register with
15607 ix86_indirect_branch_register.
15609 * config/i386/i386.md (indirect_jump): Likewise.
15610 (tablejump): Likewise.
15611 (*sibcall_memory): Likewise.
15612 (*sibcall_value_memory): Likewise.
15613 Peepholes of indirect call and jump via memory: Likewise.
15614 * config/i386/i386.opt: Likewise.
15615 * config/i386/predicates.md (indirect_branch_operand): Likewise.
15616 (GOT_memory_operand): Likewise.
15617 (call_insn_operand): Likewise.
15618 (sibcall_insn_operand): Likewise.
15619 (GOT32_symbol_operand): Likewise.
15621 2018-01-15 Jakub Jelinek <jakub@redhat.com>
15623 PR middle-end/83837
15624 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
15625 type rather than type addr's type points to.
15626 (expand_omp_atomic_mutex): Likewise.
15627 (expand_omp_atomic): Likewise.
15629 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
15632 * config/i386/i386.c (output_indirect_thunk_function): Use
15633 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
15634 for __x86_return_thunk.
15636 2018-01-15 Richard Biener <rguenther@suse.de>
15638 PR middle-end/83850
15639 * expmed.c (extract_bit_field_1): Fix typo.
15641 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15644 * config/arm/iterators.md (VF): New mode iterator.
15645 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
15646 Remove integer-related logic from pattern.
15647 (neon_vabd<mode>_3): Likewise.
15649 2018-01-15 Jakub Jelinek <jakub@redhat.com>
15651 PR middle-end/82694
15652 * common.opt (fstrict-overflow): No longer an alias.
15653 (fwrapv-pointer): New option.
15654 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
15655 also for pointer types based on flag_wrapv_pointer.
15656 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
15657 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
15658 opts->x_flag_wrapv got set.
15659 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
15660 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
15661 POINTER_TYPE_OVERFLOW_UNDEFINED.
15662 * match.pd: Likewise in address comparison pattern.
15663 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
15665 2018-01-15 Richard Biener <rguenther@suse.de>
15668 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
15669 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
15670 Reset type names to their identifier if their TYPE_DECL doesn't
15671 have linkage (and thus is used for ODR and devirt).
15672 (save_debug_info_for_decl): Remove.
15673 (save_debug_info_for_type): Likewise.
15674 (add_tree_to_fld_list): Adjust.
15675 * tree-pretty-print.c (dump_generic_node): Make dumping of
15676 type names more robust.
15678 2018-01-15 Richard Biener <rguenther@suse.de>
15680 * BASE-VER: Bump to 8.0.1.
15682 2018-01-14 Martin Sebor <msebor@redhat.com>
15685 * builtins.c (check_access): Avoid warning when the no-warning bit
15688 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
15690 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
15691 * ira-color (allocno_hard_regs_compare): Likewise.
15693 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
15696 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
15697 Use .pushsection/.popsection.
15699 2018-01-14 Martin Sebor <msebor@redhat.com>
15702 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
15704 2018-01-14 Jakub Jelinek <jakub@redhat.com>
15706 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
15707 entry from extra_headers.
15708 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
15709 extra_headers, make the list bitwise identical to the i?86-*-* one.
15711 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
15713 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
15714 -mcmodel=large with -mindirect-branch=thunk,
15715 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
15716 -mfunction-return=thunk-extern.
15717 * doc/invoke.texi: Document -mcmodel=large is incompatible with
15718 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
15719 -mfunction-return=thunk and -mfunction-return=thunk-extern.
15721 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
15723 * config/i386/i386.c (print_reg): Print the name of the full
15724 integer register without '%'.
15725 (ix86_print_operand): Handle 'V'.
15726 * doc/extend.texi: Document 'V' modifier.
15728 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
15730 * config/i386/constraints.md (Bs): Disallow memory operand for
15731 -mindirect-branch-register.
15733 * config/i386/predicates.md (indirect_branch_operand): Likewise.
15734 (GOT_memory_operand): Likewise.
15735 (call_insn_operand): Likewise.
15736 (sibcall_insn_operand): Likewise.
15737 (GOT32_symbol_operand): Likewise.
15738 * config/i386/i386.md (indirect_jump): Call convert_memory_address
15739 for -mindirect-branch-register.
15740 (tablejump): Likewise.
15741 (*sibcall_memory): Likewise.
15742 (*sibcall_value_memory): Likewise.
15743 Disallow peepholes of indirect call and jump via memory for
15744 -mindirect-branch-register.
15745 (*call_pop): Replace m with Bw.
15746 (*call_value_pop): Likewise.
15747 (*sibcall_pop_memory): Replace m with Bs.
15748 * config/i386/i386.opt (mindirect-branch-register): New option.
15749 * doc/invoke.texi: Document -mindirect-branch-register option.
15751 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
15753 * config/i386/i386-protos.h (ix86_output_function_return): New.
15754 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
15755 set function_return_type.
15756 (indirect_thunk_name): Add ret_p to indicate thunk for function
15758 (output_indirect_thunk_function): Pass false to
15759 indirect_thunk_name.
15760 (ix86_output_indirect_branch_via_reg): Likewise.
15761 (ix86_output_indirect_branch_via_push): Likewise.
15762 (output_indirect_thunk_function): Create alias for function
15763 return thunk if regno < 0.
15764 (ix86_output_function_return): New function.
15765 (ix86_handle_fndecl_attribute): Handle function_return.
15766 (ix86_attribute_table): Add function_return.
15767 * config/i386/i386.h (machine_function): Add
15768 function_return_type.
15769 * config/i386/i386.md (simple_return_internal): Use
15770 ix86_output_function_return.
15771 (simple_return_internal_long): Likewise.
15772 * config/i386/i386.opt (mfunction-return=): New option.
15773 (indirect_branch): Mention -mfunction-return=.
15774 * doc/extend.texi: Document function_return function attribute.
15775 * doc/invoke.texi: Document -mfunction-return= option.
15777 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
15779 * config/i386/i386-opts.h (indirect_branch): New.
15780 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
15781 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
15782 with local indirect jump when converting indirect call and jump.
15783 (ix86_set_indirect_branch_type): New.
15784 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
15785 (indirectlabelno): New.
15786 (indirect_thunk_needed): Likewise.
15787 (indirect_thunk_bnd_needed): Likewise.
15788 (indirect_thunks_used): Likewise.
15789 (indirect_thunks_bnd_used): Likewise.
15790 (INDIRECT_LABEL): Likewise.
15791 (indirect_thunk_name): Likewise.
15792 (output_indirect_thunk): Likewise.
15793 (output_indirect_thunk_function): Likewise.
15794 (ix86_output_indirect_branch_via_reg): Likewise.
15795 (ix86_output_indirect_branch_via_push): Likewise.
15796 (ix86_output_indirect_branch): Likewise.
15797 (ix86_output_indirect_jmp): Likewise.
15798 (ix86_code_end): Call output_indirect_thunk_function if needed.
15799 (ix86_output_call_insn): Call ix86_output_indirect_branch if
15801 (ix86_handle_fndecl_attribute): Handle indirect_branch.
15802 (ix86_attribute_table): Add indirect_branch.
15803 * config/i386/i386.h (machine_function): Add indirect_branch_type
15804 and has_local_indirect_jump.
15805 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
15807 (tablejump): Likewise.
15808 (*indirect_jump): Use ix86_output_indirect_jmp.
15809 (*tablejump_1): Likewise.
15810 (simple_return_indirect_internal): Likewise.
15811 * config/i386/i386.opt (mindirect-branch=): New option.
15812 (indirect_branch): New.
15815 (thunk-inline): Likewise.
15816 (thunk-extern): Likewise.
15817 * doc/extend.texi: Document indirect_branch function attribute.
15818 * doc/invoke.texi: Document -mindirect-branch= option.
15820 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
15823 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
15825 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
15827 * ipa-inline.c (want_inline_small_function_p): Return false if
15828 inlining has already failed with CIF_FINAL_ERROR.
15829 (update_caller_keys): Call want_inline_small_function_p before
15831 (update_callee_keys): Likewise.
15833 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
15835 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
15837 (rs6000_quadword_masked_address_p): Likewise.
15838 (quad_aligned_load_p): Likewise.
15839 (quad_aligned_store_p): Likewise.
15840 (const_load_sequence_p): Add comment to describe the outer-most loop.
15841 (mimic_memory_attributes_and_flags): New function.
15842 (rs6000_gen_stvx): Likewise.
15843 (replace_swapped_aligned_store): Likewise.
15844 (rs6000_gen_lvx): Likewise.
15845 (replace_swapped_aligned_load): Likewise.
15846 (replace_swapped_load_constant): Capitalize argument name in
15847 comment describing this function.
15848 (rs6000_analyze_swaps): Add a third pass to search for vector loads
15849 and stores that access quad-word aligned addresses and replace
15850 with stvx or lvx instructions when appropriate.
15851 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
15852 New function prototype.
15853 (rs6000_quadword_masked_address_p): Likewise.
15854 (rs6000_gen_lvx): Likewise.
15855 (rs6000_gen_stvx): Likewise.
15856 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
15857 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
15858 when memory address is aligned.
15859 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
15860 this split to select lvx instruction when memory address is aligned.
15861 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
15862 instruction when memory address is aligned.
15863 (*vsx_le_perm_load_v16qi): Likewise.
15864 (four unnamed splitters): Modify to select the stvx instruction
15865 when memory is aligned.
15867 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
15869 * predict.c (determine_unlikely_bbs): Handle correctly BBs
15870 which appears in the queue multiple times.
15872 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
15873 Alan Hayward <alan.hayward@arm.com>
15874 David Sherwood <david.sherwood@arm.com>
15876 * tree-vectorizer.h (vec_lower_bound): New structure.
15877 (_loop_vec_info): Add check_nonzero and lower_bounds.
15878 (LOOP_VINFO_CHECK_NONZERO): New macro.
15879 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
15880 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
15881 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
15882 fields. Make seg_len the distance travelled, not including the
15884 (dr_direction_indicator): Declare.
15885 (dr_zero_step_indicator): Likewise.
15886 (dr_known_forward_stride_p): Likewise.
15887 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
15889 (runtime_alias_check_p): Allow runtime alias checks with
15891 (operator ==): Compare access_size and align.
15892 (prune_runtime_alias_test_list): Rework for new distinction between
15893 the access_size and seg_len.
15894 (create_intersect_range_checks_index): Likewise. Cope with polynomial
15896 (get_segment_min_max): New function.
15897 (create_intersect_range_checks): Use it.
15898 (dr_step_indicator): New function.
15899 (dr_direction_indicator): Likewise.
15900 (dr_zero_step_indicator): Likewise.
15901 (dr_known_forward_stride_p): Likewise.
15902 * tree-loop-distribution.c (data_ref_segment_size): Return
15903 DR_STEP * (niters - 1).
15904 (compute_alias_check_pairs): Update call to the dr_with_seg_len
15906 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
15907 (vect_preserves_scalar_order_p): New function, split out from...
15908 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
15909 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
15910 (vect_vfa_access_size): New function.
15911 (vect_vfa_align): Likewise.
15912 (vect_compile_time_alias): Take access_size_a and access_b arguments.
15913 (dump_lower_bound): New function.
15914 (vect_check_lower_bound): Likewise.
15915 (vect_small_gap_p): Likewise.
15916 (vectorizable_with_step_bound_p): Likewise.
15917 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
15918 depencies if the vectorization factor is 1. Convert the checks
15919 for nonzero steps into checks on the bounds of DR_STEP. Try using
15920 a bunds check for variable steps if the minimum required step is
15921 relatively small. Update calls to the dr_with_seg_len
15922 constructor and to vect_compile_time_alias.
15923 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
15925 (vect_loop_versioning): Call it.
15926 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
15928 (vect_estimate_min_profitable_iters): Account for any bounds checks.
15930 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
15931 Alan Hayward <alan.hayward@arm.com>
15932 David Sherwood <david.sherwood@arm.com>
15934 * doc/sourcebuild.texi (vect_scatter_store): Document.
15935 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
15937 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
15939 * genopinit.c (main): Add supports_vec_scatter_store and
15940 supports_vec_scatter_store_cached to target_optabs.
15941 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
15942 IFN_MASK_SCATTER_STORE.
15943 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
15945 * internal-fn.h (internal_store_fn_p): Declare.
15946 (internal_fn_stored_value_index): Likewise.
15947 * internal-fn.c (scatter_store_direct): New macro.
15948 (expand_scatter_store_optab_fn): New function.
15949 (direct_scatter_store_optab_supported_p): New macro.
15950 (internal_store_fn_p): New function.
15951 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
15952 IFN_MASK_SCATTER_STORE.
15953 (internal_fn_mask_index): Likewise.
15954 (internal_fn_stored_value_index): New function.
15955 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
15956 for scatter stores.
15957 * optabs-query.h (supports_vec_scatter_store_p): Declare.
15958 * optabs-query.c (supports_vec_scatter_store_p): New function.
15959 * tree-vectorizer.h (vect_get_store_rhs): Declare.
15960 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
15961 true for scatter stores.
15962 (vect_gather_scatter_fn_p): Handle scatter stores too.
15963 (vect_check_gather_scatter): Consider using scatter stores if
15964 supports_vec_scatter_store_p.
15965 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
15966 scatter stores too.
15967 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
15968 internal_fn_stored_value_index.
15969 (check_load_store_masking): Handle scatter stores too.
15970 (vect_get_store_rhs): Make public.
15971 (vectorizable_call): Use internal_store_fn_p.
15972 (vectorizable_store): Handle scatter store internal functions.
15973 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
15974 when deciding whether the end of the group has been reached.
15975 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
15976 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
15977 (mask_scatter_store<mode>): New insns.
15979 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
15980 Alan Hayward <alan.hayward@arm.com>
15981 David Sherwood <david.sherwood@arm.com>
15983 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
15984 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
15985 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
15987 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
15988 Use vect_truncate_gather_scatter_offset if we can't treat the
15989 operation as a normal gather load or scatter store.
15990 (get_group_load_store_type): Take the gather_scatter_info
15991 as argument. Try using a gather load or scatter store for
15992 single-element groups.
15993 (get_load_store_type): Update calls to get_group_load_store_type
15994 and vect_use_strided_gather_scatters_p.
15996 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
15997 Alan Hayward <alan.hayward@arm.com>
15998 David Sherwood <david.sherwood@arm.com>
16000 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
16001 optional tree argument.
16002 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
16004 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
16005 but continue to use the current value as a fallback.
16006 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
16007 to compare the updates.
16008 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
16009 (get_load_store_type): Use it when handling a strided access.
16010 (vect_get_strided_load_store_ops): New function.
16011 (vect_get_data_ptr_increment): Likewise.
16012 (vectorizable_load): Handle strided gather loads. Always pass
16013 a step to vect_create_data_ref_ptr and bump_vector_ptr.
16015 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16016 Alan Hayward <alan.hayward@arm.com>
16017 David Sherwood <david.sherwood@arm.com>
16019 * doc/md.texi (gather_load@var{m}): Document.
16020 (mask_gather_load@var{m}): Likewise.
16021 * genopinit.c (main): Add supports_vec_gather_load and
16022 supports_vec_gather_load_cached to target_optabs.
16023 * optabs-tree.c (init_tree_optimization_optabs): Use
16024 ggc_cleared_alloc to allocate target_optabs.
16025 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
16026 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
16028 * internal-fn.h (internal_load_fn_p): Declare.
16029 (internal_gather_scatter_fn_p): Likewise.
16030 (internal_fn_mask_index): Likewise.
16031 (internal_gather_scatter_fn_supported_p): Likewise.
16032 * internal-fn.c (gather_load_direct): New macro.
16033 (expand_gather_load_optab_fn): New function.
16034 (direct_gather_load_optab_supported_p): New macro.
16035 (direct_internal_fn_optab): New function.
16036 (internal_load_fn_p): Likewise.
16037 (internal_gather_scatter_fn_p): Likewise.
16038 (internal_fn_mask_index): Likewise.
16039 (internal_gather_scatter_fn_supported_p): Likewise.
16040 * optabs-query.c (supports_at_least_one_mode_p): New function.
16041 (supports_vec_gather_load_p): Likewise.
16042 * optabs-query.h (supports_vec_gather_load_p): Declare.
16043 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
16044 and memory_type field.
16045 (NUM_PATTERNS): Bump to 15.
16046 * tree-vect-data-refs.c: Include internal-fn.h.
16047 (vect_gather_scatter_fn_p): New function.
16048 (vect_describe_gather_scatter_call): Likewise.
16049 (vect_check_gather_scatter): Try using internal functions for
16050 gather loads. Recognize existing calls to a gather load function.
16051 (vect_analyze_data_refs): Consider using gather loads if
16052 supports_vec_gather_load_p.
16053 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
16054 (vect_get_gather_scatter_offset_type): Likewise.
16055 (vect_convert_mask_for_vectype): Likewise.
16056 (vect_add_conversion_to_patterm): Likewise.
16057 (vect_try_gather_scatter_pattern): Likewise.
16058 (vect_recog_gather_scatter_pattern): New pattern recognizer.
16059 (vect_vect_recog_func_ptrs): Add it.
16060 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
16061 internal_fn_mask_index and internal_gather_scatter_fn_p.
16062 (check_load_store_masking): Take the gather_scatter_info as an
16063 argument and handle gather loads.
16064 (vect_get_gather_scatter_ops): New function.
16065 (vectorizable_call): Check internal_load_fn_p.
16066 (vectorizable_load): Likewise. Handle gather load internal
16068 (vectorizable_store): Update call to check_load_store_masking.
16069 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
16070 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
16071 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
16072 (aarch64_gather_scale_operand_d): New predicates.
16073 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
16074 (mask_gather_load<mode>): New insns.
16076 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16077 Alan Hayward <alan.hayward@arm.com>
16078 David Sherwood <david.sherwood@arm.com>
16080 * optabs.def (fold_left_plus_optab): New optab.
16081 * doc/md.texi (fold_left_plus_@var{m}): Document.
16082 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
16083 * internal-fn.c (fold_left_direct): Define.
16084 (expand_fold_left_optab_fn): Likewise.
16085 (direct_fold_left_optab_supported_p): Likewise.
16086 * fold-const-call.c (fold_const_fold_left): New function.
16087 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
16088 * tree-parloops.c (valid_reduction_p): New function.
16089 (gather_scalar_reductions): Use it.
16090 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
16091 (vect_finish_replace_stmt): Declare.
16092 * tree-vect-loop.c (fold_left_reduction_fn): New function.
16093 (needs_fold_left_reduction_p): New function, split out from...
16094 (vect_is_simple_reduction): ...here. Accept reductions that
16095 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
16096 (vect_force_simple_reduction): Also store the reduction type in
16097 the assignment's STMT_VINFO_REDUC_TYPE.
16098 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
16099 (merge_with_identity): New function.
16100 (vect_expand_fold_left): Likewise.
16101 (vectorize_fold_left_reduction): Likewise.
16102 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
16103 scalar phi in place for it. Check for target support and reject
16104 cases that would reassociate the operation. Defer the transform
16105 phase to vectorize_fold_left_reduction.
16106 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
16107 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
16108 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
16110 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16112 * tree-if-conv.c (predicate_mem_writes): Remove redundant
16113 call to ifc_temp_var.
16115 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16116 Alan Hayward <alan.hayward@arm.com>
16117 David Sherwood <david.sherwood@arm.com>
16119 * target.def (legitimize_address_displacement): Take the original
16120 offset as a poly_int.
16121 * targhooks.h (default_legitimize_address_displacement): Update
16123 * targhooks.c (default_legitimize_address_displacement): Likewise.
16124 * doc/tm.texi: Regenerate.
16125 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
16126 as an argument, moving assert of ad->disp == ad->disp_term to...
16127 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
16128 Try calling targetm.legitimize_address_displacement before expanding
16129 the address rather than afterwards, and adjust for the new interface.
16130 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
16131 Match the new hook interface. Handle SVE addresses.
16132 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
16133 new hook interface.
16135 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16137 * Makefile.in (OBJS): Add early-remat.o.
16138 * target.def (select_early_remat_modes): New hook.
16139 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
16140 * doc/tm.texi: Regenerate.
16141 * targhooks.h (default_select_early_remat_modes): Declare.
16142 * targhooks.c (default_select_early_remat_modes): New function.
16143 * timevar.def (TV_EARLY_REMAT): New timevar.
16144 * passes.def (pass_early_remat): New pass.
16145 * tree-pass.h (make_pass_early_remat): Declare.
16146 * early-remat.c: New file.
16147 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
16149 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
16151 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16152 Alan Hayward <alan.hayward@arm.com>
16153 David Sherwood <david.sherwood@arm.com>
16155 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
16156 vfm1 with a bound_epilog parameter.
16157 (vect_do_peeling): Update calls accordingly, and move the prologue
16158 call earlier in the function. Treat the base bound_epilog as 0 for
16159 fully-masked loops and retain vf - 1 for other loops. Add 1 to
16160 this base when peeling for gaps.
16161 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
16162 with fully-masked loops.
16163 (vect_estimate_min_profitable_iters): Handle the single peeled
16164 iteration in that case.
16166 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16167 Alan Hayward <alan.hayward@arm.com>
16168 David Sherwood <david.sherwood@arm.com>
16170 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
16171 single-element interleaving even if the size is not a power of 2.
16172 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
16173 accesses for single-element interleaving if the group size is
16176 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16177 Alan Hayward <alan.hayward@arm.com>
16178 David Sherwood <david.sherwood@arm.com>
16180 * doc/md.texi (fold_extract_last_@var{m}): Document.
16181 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
16182 * optabs.def (fold_extract_last_optab): New optab.
16183 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
16184 * internal-fn.c (fold_extract_direct): New macro.
16185 (expand_fold_extract_optab_fn): Likewise.
16186 (direct_fold_extract_optab_supported_p): Likewise.
16187 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
16188 * tree-vect-loop.c (vect_model_reduction_cost): Handle
16189 EXTRACT_LAST_REDUCTION.
16190 (get_initial_def_for_reduction): Do not create an initial vector
16191 for EXTRACT_LAST_REDUCTION reductions.
16192 (vectorizable_reduction): Leave the scalar phi in place for
16193 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
16194 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
16195 epilogue code for EXTRACT_LAST_REDUCTION and defer the
16196 transform phase to vectorizable_condition.
16197 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
16199 (vect_finish_stmt_generation): ...here.
16200 (vect_finish_replace_stmt): New function.
16201 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
16202 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
16204 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
16206 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16207 Alan Hayward <alan.hayward@arm.com>
16208 David Sherwood <david.sherwood@arm.com>
16210 * doc/md.texi (extract_last_@var{m}): Document.
16211 * optabs.def (extract_last_optab): New optab.
16212 * internal-fn.def (EXTRACT_LAST): New internal function.
16213 * internal-fn.c (cond_unary_direct): New macro.
16214 (expand_cond_unary_optab_fn): Likewise.
16215 (direct_cond_unary_optab_supported_p): Likewise.
16216 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
16217 loops using EXTRACT_LAST.
16218 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
16219 (extract_last_<mode>): ...this optab.
16220 (vec_extract<mode><Vel>): Update accordingly.
16222 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16223 Alan Hayward <alan.hayward@arm.com>
16224 David Sherwood <david.sherwood@arm.com>
16226 * target.def (empty_mask_is_expensive): New hook.
16227 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
16228 * doc/tm.texi: Regenerate.
16229 * targhooks.h (default_empty_mask_is_expensive): Declare.
16230 * targhooks.c (default_empty_mask_is_expensive): New function.
16231 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
16232 if the target says that empty masks are expensive.
16233 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
16235 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
16237 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16238 Alan Hayward <alan.hayward@arm.com>
16239 David Sherwood <david.sherwood@arm.com>
16241 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
16242 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
16243 (vect_use_loop_mask_for_alignment_p): New function.
16244 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
16245 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
16246 niters_skip argument. Make sure that the first niters_skip elements
16247 of the first iteration are inactive.
16248 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
16249 Update call to vect_set_loop_masks_directly.
16250 (get_misalign_in_elems): New function, split out from...
16251 (vect_gen_prolog_loop_niters): ...here.
16252 (vect_update_init_of_dr): Take a code argument that specifies whether
16253 the adjustment should be added or subtracted.
16254 (vect_update_init_of_drs): Likewise.
16255 (vect_prepare_for_masked_peels): New function.
16256 (vect_do_peeling): Skip prologue peeling if we're using a mask
16257 instead. Update call to vect_update_inits_of_drs.
16258 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
16260 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
16261 alignment. Do not include the number of peeled iterations in
16262 the minimum threshold in that case.
16263 (vectorizable_induction): Adjust the start value down by
16264 LOOP_VINFO_MASK_SKIP_NITERS iterations.
16265 (vect_transform_loop): Call vect_prepare_for_masked_peels.
16266 Take the number of skipped iterations into account when calculating
16268 * tree-vect-stmts.c (vect_gen_while_not): New function.
16270 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16271 Alan Hayward <alan.hayward@arm.com>
16272 David Sherwood <david.sherwood@arm.com>
16274 * doc/sourcebuild.texi (vect_fully_masked): Document.
16275 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
16276 default value to 0.
16277 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
16279 (vect_analyze_loop_2): ...here. Don't check the vectorization
16280 factor against the number of loop iterations if the loop is
16283 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16284 Alan Hayward <alan.hayward@arm.com>
16285 David Sherwood <david.sherwood@arm.com>
16287 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
16288 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
16289 (dump_groups): Update accordingly.
16290 (iv_use::mem_type): New member variable.
16291 (address_p): New function.
16292 (record_use): Add a mem_type argument and initialize the new
16294 (record_group_use): Add a mem_type argument. Use address_p.
16295 Remove obsolete null checks of base_object. Update call to record_use.
16296 (find_interesting_uses_op): Update call to record_group_use.
16297 (find_interesting_uses_cond): Likewise.
16298 (find_interesting_uses_address): Likewise.
16299 (get_mem_type_for_internal_fn): New function.
16300 (find_address_like_use): Likewise.
16301 (find_interesting_uses_stmt): Try find_address_like_use before
16302 calling find_interesting_uses_op.
16303 (addr_offset_valid_p): Use the iv mem_type field as the type
16304 of the addressed memory.
16305 (add_autoinc_candidates): Likewise.
16306 (get_address_cost): Likewise.
16307 (split_small_address_groups_p): Use address_p.
16308 (split_address_groups): Likewise.
16309 (add_iv_candidate_for_use): Likewise.
16310 (autoinc_possible_for_pair): Likewise.
16311 (rewrite_groups): Likewise.
16312 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
16313 (determine_group_iv_cost): Update after split of USE_ADDRESS.
16314 (get_alias_ptr_type_for_ptr_address): New function.
16315 (rewrite_use_address): Rewrite address uses in calls that were
16316 identified by find_address_like_use.
16318 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16319 Alan Hayward <alan.hayward@arm.com>
16320 David Sherwood <david.sherwood@arm.com>
16322 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
16324 * gimple-expr.h (is_gimple_addressable: Likewise.
16325 * gimple-expr.c (is_gimple_address): Likewise.
16326 * internal-fn.c (expand_call_mem_ref): New function.
16327 (expand_mask_load_optab_fn): Use it.
16328 (expand_mask_store_optab_fn): Likewise.
16330 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16331 Alan Hayward <alan.hayward@arm.com>
16332 David Sherwood <david.sherwood@arm.com>
16334 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
16335 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
16336 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
16337 (cond_umax@var{mode}): Document.
16338 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
16339 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
16340 (cond_umin_optab, cond_umax_optab): New optabs.
16341 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
16342 (COND_IOR, COND_XOR): New internal functions.
16343 * internal-fn.h (get_conditional_internal_fn): Declare.
16344 * internal-fn.c (cond_binary_direct): New macro.
16345 (expand_cond_binary_optab_fn): Likewise.
16346 (direct_cond_binary_optab_supported_p): Likewise.
16347 (get_conditional_internal_fn): New function.
16348 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
16349 Cope with reduction statements that are vectorized as calls rather
16351 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
16352 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
16353 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
16354 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
16355 (UNSPEC_COND_EOR): New unspecs.
16356 (optab): Add mappings for them.
16357 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
16358 (sve_int_op, sve_fp_op): New int attributes.
16360 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16361 Alan Hayward <alan.hayward@arm.com>
16362 David Sherwood <david.sherwood@arm.com>
16364 * optabs.def (while_ult_optab): New optab.
16365 * doc/md.texi (while_ult@var{m}@var{n}): Document.
16366 * internal-fn.def (WHILE_ULT): New internal function.
16367 * internal-fn.h (direct_internal_fn_supported_p): New override
16368 that takes two types as argument.
16369 * internal-fn.c (while_direct): New macro.
16370 (expand_while_optab_fn): New function.
16371 (convert_optab_supported_p): Likewise.
16372 (direct_while_optab_supported_p): New macro.
16373 * wide-int.h (wi::udiv_ceil): New function.
16374 * tree-vectorizer.h (rgroup_masks): New structure.
16375 (vec_loop_masks): New typedef.
16376 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
16377 and fully_masked_p.
16378 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
16379 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
16380 (vect_max_vf): New function.
16381 (slpeel_make_loop_iterate_ntimes): Delete.
16382 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
16383 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
16384 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
16385 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
16386 internal-fn.h, stor-layout.h and optabs-query.h.
16387 (vect_set_loop_mask): New function.
16388 (add_preheader_seq): Likewise.
16389 (add_header_seq): Likewise.
16390 (interleave_supported_p): Likewise.
16391 (vect_maybe_permute_loop_masks): Likewise.
16392 (vect_set_loop_masks_directly): Likewise.
16393 (vect_set_loop_condition_masked): Likewise.
16394 (vect_set_loop_condition_unmasked): New function, split out from
16395 slpeel_make_loop_iterate_ntimes.
16396 (slpeel_make_loop_iterate_ntimes): Rename to..
16397 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
16398 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
16399 (vect_do_peeling): Update call accordingly.
16400 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
16402 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
16403 mask_compare_type, can_fully_mask_p and fully_masked_p.
16404 (release_vec_loop_masks): New function.
16405 (_loop_vec_info): Use it to free the loop masks.
16406 (can_produce_all_loop_masks_p): New function.
16407 (vect_get_max_nscalars_per_iter): Likewise.
16408 (vect_verify_full_masking): Likewise.
16409 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
16410 retries, and free the mask rgroups before retrying. Check loop-wide
16411 reasons for disallowing fully-masked loops. Make the final decision
16412 about whether use a fully-masked loop or not.
16413 (vect_estimate_min_profitable_iters): Do not assume that peeling
16414 for the number of iterations will be needed for fully-masked loops.
16415 (vectorizable_reduction): Disable fully-masked loops.
16416 (vectorizable_live_operation): Likewise.
16417 (vect_halve_mask_nunits): New function.
16418 (vect_double_mask_nunits): Likewise.
16419 (vect_record_loop_mask): Likewise.
16420 (vect_get_loop_mask): Likewise.
16421 (vect_transform_loop): Handle the case in which the final loop
16422 iteration might handle a partial vector. Call vect_set_loop_condition
16423 instead of slpeel_make_loop_iterate_ntimes.
16424 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
16425 (check_load_store_masking): New function.
16426 (prepare_load_store_mask): Likewise.
16427 (vectorizable_store): Handle fully-masked loops.
16428 (vectorizable_load): Likewise.
16429 (supportable_widening_operation): Use vect_halve_mask_nunits for
16431 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
16432 (vect_gen_while): New function.
16433 * config/aarch64/aarch64.md (umax<mode>3): New expander.
16434 (aarch64_uqdec<mode>): New insn.
16436 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16437 Alan Hayward <alan.hayward@arm.com>
16438 David Sherwood <david.sherwood@arm.com>
16440 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
16441 (reduc_xor_scal_optab): New optabs.
16442 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
16443 (reduc_xor_scal_@var{m}): Document.
16444 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
16445 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
16446 internal functions.
16447 * fold-const-call.c (fold_const_call): Handle them.
16448 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
16449 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
16450 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
16451 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
16452 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
16453 (UNSPEC_XORV): New unspecs.
16454 (optab): Add entries for them.
16455 (BITWISEV): New int iterator.
16456 (bit_reduc_op): New int attributes.
16458 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16459 Alan Hayward <alan.hayward@arm.com>
16460 David Sherwood <david.sherwood@arm.com>
16462 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
16463 * internal-fn.def (VEC_SHL_INSERT): New internal function.
16464 * optabs.def (vec_shl_insert_optab): New optab.
16465 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
16466 (duplicate_and_interleave): Likewise.
16467 * tree-vect-loop.c: Include internal-fn.h.
16468 (neutral_op_for_slp_reduction): New function, split out from
16469 get_initial_defs_for_reduction.
16470 (get_initial_def_for_reduction): Handle option 2 for variable-length
16471 vectors by loading the neutral value into a vector and then shifting
16472 the initial value into element 0.
16473 (get_initial_defs_for_reduction): Replace the code argument with
16474 the neutral value calculated by neutral_op_for_slp_reduction.
16475 Use gimple_build_vector for constant-length vectors.
16476 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
16477 but the first group_size elements have a neutral value.
16478 Use duplicate_and_interleave otherwise.
16479 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
16480 Update call to get_initial_defs_for_reduction. Handle SLP
16481 reductions for variable-length vectors by creating one vector
16482 result for each scalar result, with the elements associated
16483 with other scalar results stubbed out with the neutral value.
16484 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
16485 Require IFN_VEC_SHL_INSERT for double reductions on
16486 variable-length vectors, or SLP reductions that have
16487 a neutral value. Require can_duplicate_and_interleave_p
16488 support for variable-length unchained SLP reductions if there
16489 is no neutral value, such as for MIN/MAX reductions. Also require
16490 the number of vector elements to be a multiple of the number of
16491 SLP statements when doing variable-length unchained SLP reductions.
16492 Update call to vect_create_epilog_for_reduction.
16493 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
16494 and remove initial values.
16495 (duplicate_and_interleave): Make public.
16496 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
16497 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
16499 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16500 Alan Hayward <alan.hayward@arm.com>
16501 David Sherwood <david.sherwood@arm.com>
16503 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
16504 (can_duplicate_and_interleave_p): New function.
16505 (vect_get_and_check_slp_defs): Take the vector of statements
16506 rather than just the current one. Remove excess parentheses.
16507 Restriction rejectinon of vect_constant_def and vect_external_def
16508 for variable-length vectors to boolean types, or types for which
16509 can_duplicate_and_interleave_p is false.
16510 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
16511 (duplicate_and_interleave): New function.
16512 (vect_get_constant_vectors): Use gimple_build_vector for
16513 constant-length vectors and suitable variable-length constant
16514 vectors. Use duplicate_and_interleave for other variable-length
16515 vectors. Don't defer the update when inserting new statements.
16517 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16518 Alan Hayward <alan.hayward@arm.com>
16519 David Sherwood <david.sherwood@arm.com>
16521 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
16522 min_profitable_iters doesn't go negative.
16524 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16525 Alan Hayward <alan.hayward@arm.com>
16526 David Sherwood <david.sherwood@arm.com>
16528 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
16529 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
16530 * optabs.def (vec_mask_load_lanes_optab): New optab.
16531 (vec_mask_store_lanes_optab): Likewise.
16532 * internal-fn.def (MASK_LOAD_LANES): New internal function.
16533 (MASK_STORE_LANES): Likewise.
16534 * internal-fn.c (mask_load_lanes_direct): New macro.
16535 (mask_store_lanes_direct): Likewise.
16536 (expand_mask_load_optab_fn): Handle masked operations.
16537 (expand_mask_load_lanes_optab_fn): New macro.
16538 (expand_mask_store_optab_fn): Handle masked operations.
16539 (expand_mask_store_lanes_optab_fn): New macro.
16540 (direct_mask_load_lanes_optab_supported_p): Likewise.
16541 (direct_mask_store_lanes_optab_supported_p): Likewise.
16542 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
16544 (vect_load_lanes_supported): Likewise.
16545 * tree-vect-data-refs.c (strip_conversion): New function.
16546 (can_group_stmts_p): Likewise.
16547 (vect_analyze_data_ref_accesses): Use it instead of checking
16548 for a pair of assignments.
16549 (vect_store_lanes_supported): Take a masked_p parameter.
16550 (vect_load_lanes_supported): Likewise.
16551 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
16552 vect_store_lanes_supported and vect_load_lanes_supported.
16553 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
16554 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
16555 parameter. Don't allow gaps for masked accesses.
16556 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
16557 and vect_load_lanes_supported.
16558 (get_load_store_type): Take a masked_p parameter and update
16559 call to get_group_load_store_type.
16560 (vectorizable_store): Update call to get_load_store_type.
16561 Handle IFN_MASK_STORE_LANES.
16562 (vectorizable_load): Update call to get_load_store_type.
16563 Handle IFN_MASK_LOAD_LANES.
16565 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16566 Alan Hayward <alan.hayward@arm.com>
16567 David Sherwood <david.sherwood@arm.com>
16569 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
16571 * config/aarch64/aarch64-protos.h
16572 (aarch64_sve_struct_memory_operand_p): Declare.
16573 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
16574 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
16575 (VPRED, vpred): Handle SVE structure modes.
16576 * config/aarch64/constraints.md (Utx): New constraint.
16577 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
16578 (aarch64_sve_struct_nonimmediate_operand): New predicates.
16579 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
16580 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
16581 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
16582 structure modes. Split into pieces after RA.
16583 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
16584 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
16586 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
16587 SVE structure modes.
16588 (aarch64_classify_address): Likewise.
16589 (sizetochar): Move earlier in file.
16590 (aarch64_print_operand): Handle SVE register lists.
16591 (aarch64_array_mode): New function.
16592 (aarch64_sve_struct_memory_operand_p): Likewise.
16593 (TARGET_ARRAY_MODE): Redefine.
16595 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16596 Alan Hayward <alan.hayward@arm.com>
16597 David Sherwood <david.sherwood@arm.com>
16599 * target.def (array_mode): New target hook.
16600 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
16601 * doc/tm.texi: Regenerate.
16602 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
16603 * hooks.c (hook_optmode_mode_uhwi_none): New function.
16604 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
16605 targetm.array_mode.
16606 * stor-layout.c (mode_for_array): Likewise. Support polynomial
16609 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16610 Alan Hayward <alan.hayward@arm.com>
16611 David Sherwood <david.sherwood@arm.com>
16613 * fold-const.c (fold_binary_loc): Check the argument types
16614 rather than the result type when testing for a vector operation.
16616 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16618 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
16619 * doc/tm.texi: Regenerate.
16621 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16622 Alan Hayward <alan.hayward@arm.com>
16623 David Sherwood <david.sherwood@arm.com>
16625 * doc/invoke.texi (-msve-vector-bits=): Document new option.
16626 (sve): Document new AArch64 extension.
16627 * doc/md.texi (w): Extend the description of the AArch64
16628 constraint to include SVE vectors.
16629 (Upl, Upa): Document new AArch64 predicate constraints.
16630 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
16632 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
16633 (msve-vector-bits=): New option.
16634 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
16635 SVE when these are disabled.
16636 (sve): New extension.
16637 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
16638 modes. Adjust their number of units based on aarch64_sve_vg.
16639 (MAX_BITSIZE_MODE_ANY_MODE): Define.
16640 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
16641 aarch64_addr_query_type.
16642 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
16643 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
16644 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
16645 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
16646 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
16647 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
16648 (aarch64_simd_imm_zero_p): Delete.
16649 (aarch64_check_zero_based_sve_index_immediate): Declare.
16650 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
16651 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
16652 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
16653 (aarch64_sve_float_mul_immediate_p): Likewise.
16654 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
16655 rather than an rtx.
16656 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
16657 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
16658 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
16659 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
16660 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
16661 (aarch64_regmode_natural_size): Likewise.
16662 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
16663 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
16665 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
16666 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
16667 for VG and the SVE predicate registers.
16668 (V_ALIASES): Add a "z"-prefixed alias.
16669 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
16670 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
16671 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
16672 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
16673 (REG_CLASS_NAMES): Add entries for them.
16674 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
16675 and the predicate registers.
16676 (aarch64_sve_vg): Declare.
16677 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
16678 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
16679 (REGMODE_NATURAL_SIZE): Define.
16680 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
16682 * config/aarch64/aarch64.c: Include cfgrtl.h.
16683 (simd_immediate_info): Add a constructor for series vectors,
16684 and an associated step field.
16685 (aarch64_sve_vg): New variable.
16686 (aarch64_dbx_register_number): Handle VG and the predicate registers.
16687 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
16688 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
16689 (VEC_ANY_DATA, VEC_STRUCT): New constants.
16690 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
16691 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
16692 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
16693 (aarch64_get_mask_mode): New functions.
16694 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
16695 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
16696 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
16697 predicate modes and predicate registers. Explicitly restrict
16698 GPRs to modes of 16 bytes or smaller. Only allow FP registers
16699 to store a vector mode if it is recognized by
16700 aarch64_classify_vector_mode.
16701 (aarch64_regmode_natural_size): New function.
16702 (aarch64_hard_regno_caller_save_mode): Return the original mode
16704 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
16705 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
16706 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
16707 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
16709 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
16710 does not overlap dest if the function is frame-related. Handle
16712 (aarch64_split_add_offset): New function.
16713 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
16714 them aarch64_add_offset.
16715 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
16716 and update call to aarch64_sub_sp.
16717 (aarch64_add_cfa_expression): New function.
16718 (aarch64_expand_prologue): Pass extra temporary registers to the
16719 functions above. Handle the case in which we need to emit new
16720 DW_CFA_expressions for registers that were originally saved
16721 relative to the stack pointer, but now have to be expressed
16722 relative to the frame pointer.
16723 (aarch64_output_mi_thunk): Pass extra temporary registers to the
16725 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
16726 IP0 and IP1 values for SVE frames.
16727 (aarch64_expand_vec_series): New function.
16728 (aarch64_expand_sve_widened_duplicate): Likewise.
16729 (aarch64_expand_sve_const_vector): Likewise.
16730 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
16731 Handle SVE constants. Use emit_move_insn to move a force_const_mem
16732 into the register, rather than emitting a SET directly.
16733 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
16734 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
16735 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
16736 (offset_9bit_signed_scaled_p): New functions.
16737 (aarch64_replicate_bitmask_imm): New function.
16738 (aarch64_bitmask_imm): Use it.
16739 (aarch64_cannot_force_const_mem): Reject expressions involving
16740 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
16741 (aarch64_classify_index): Handle SVE indices, by requiring
16742 a plain register index with a scale that matches the element size.
16743 (aarch64_classify_address): Handle SVE addresses. Assert that
16744 the mode of the address is VOIDmode or an integer mode.
16745 Update call to aarch64_classify_symbol.
16746 (aarch64_classify_symbolic_expression): Update call to
16747 aarch64_classify_symbol.
16748 (aarch64_const_vec_all_in_range_p): New function.
16749 (aarch64_print_vector_float_operand): Likewise.
16750 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
16751 "vN" for FP registers with SVE modes. Handle (const ...) vectors
16752 and the FP immediates 1.0 and 0.5.
16753 (aarch64_print_address_internal): Handle SVE addresses.
16754 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
16755 (aarch64_regno_regclass): Handle predicate registers.
16756 (aarch64_secondary_reload): Handle big-endian reloads of SVE
16758 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
16759 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
16760 (aarch64_convert_sve_vector_bits): New function.
16761 (aarch64_override_options): Use it to handle -msve-vector-bits=.
16762 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
16763 rather than an rtx.
16764 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
16765 Handle SVE vector and predicate modes. Accept VL-based constants
16766 that need only one temporary register, and VL offsets that require
16767 no temporary registers.
16768 (aarch64_conditional_register_usage): Mark the predicate registers
16769 as fixed if SVE isn't available.
16770 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
16771 Return true for SVE vector and predicate modes.
16772 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
16773 rather than an unsigned int. Handle SVE modes.
16774 (aarch64_preferred_simd_mode): Update call accordingly. Handle
16776 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
16778 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
16779 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
16780 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
16781 (aarch64_sve_float_mul_immediate_p): New functions.
16782 (aarch64_sve_valid_immediate): New function.
16783 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
16784 Explicitly reject structure modes. Check for INDEX constants.
16785 Handle PTRUE and PFALSE constants.
16786 (aarch64_check_zero_based_sve_index_immediate): New function.
16787 (aarch64_simd_imm_zero_p): Delete.
16788 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
16789 vector modes. Accept constants in the range of CNT[BHWD].
16790 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
16791 ask for an Advanced SIMD mode.
16792 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
16793 (aarch64_simd_vector_alignment): Handle SVE predicates.
16794 (aarch64_vectorize_preferred_vector_alignment): New function.
16795 (aarch64_simd_vector_alignment_reachable): Use it instead of
16797 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
16798 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
16800 (MAX_VECT_LEN): Delete.
16801 (expand_vec_perm_d): Add a vec_flags field.
16802 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
16803 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
16804 (aarch64_evpc_ext): Don't apply a big-endian lane correction
16806 (aarch64_evpc_rev): Rename to...
16807 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
16808 (aarch64_evpc_rev_global): New function.
16809 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
16810 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
16812 (aarch64_evpc_sve_tbl): New function.
16813 (aarch64_expand_vec_perm_const_1): Update after rename of
16814 aarch64_evpc_rev. Handle SVE permutes too, trying
16815 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
16816 than aarch64_evpc_tbl.
16817 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
16818 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
16819 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
16820 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
16821 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
16822 (aarch64_expand_sve_vcond): New functions.
16823 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
16824 of aarch64_vector_mode_p.
16825 (aarch64_dwarf_poly_indeterminate_value): New function.
16826 (aarch64_compute_pressure_classes): Likewise.
16827 (aarch64_can_change_mode_class): Likewise.
16828 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
16829 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
16830 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
16831 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
16832 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
16833 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
16834 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
16835 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
16837 (Dn, Dl, Dr): Accept const as well as const_vector.
16838 (Dz): Likewise. Compare against CONST0_RTX.
16839 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
16840 of "vector" where appropriate.
16841 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
16842 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
16843 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
16844 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
16845 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
16846 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
16847 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
16848 (v_int_equiv): Extend to SVE modes.
16849 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
16851 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
16852 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
16853 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
16854 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
16855 (SVE_COND_FP_CMP): New int iterators.
16856 (perm_hilo): Handle the new unpack unspecs.
16857 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
16859 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
16860 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
16861 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
16862 (aarch64_equality_operator, aarch64_constant_vector_operand)
16863 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
16864 (aarch64_sve_nonimmediate_operand): Likewise.
16865 (aarch64_sve_general_operand): Likewise.
16866 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
16867 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
16868 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
16869 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
16870 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
16871 (aarch64_sve_float_arith_immediate): Likewise.
16872 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
16873 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
16874 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
16875 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
16876 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
16877 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
16878 (aarch64_sve_float_arith_operand): Likewise.
16879 (aarch64_sve_float_arith_with_sub_operand): Likewise.
16880 (aarch64_sve_float_mul_operand): Likewise.
16881 (aarch64_sve_vec_perm_operand): Likewise.
16882 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
16883 (aarch64_mov_operand): Accept const_poly_int and const_vector.
16884 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
16885 as well as const_vector.
16886 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
16887 in file. Use CONST0_RTX and CONSTM1_RTX.
16888 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
16889 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
16890 Use aarch64_simd_imm_zero.
16891 * config/aarch64/aarch64-sve.md: New file.
16892 * config/aarch64/aarch64.md: Include it.
16893 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
16894 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
16895 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
16896 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
16897 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
16898 (sve): New attribute.
16899 (enabled): Disable instructions with the sve attribute unless
16901 (movqi, movhi): Pass CONST_POLY_INT operaneds through
16902 aarch64_expand_mov_immediate.
16903 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
16904 CNT[BHSD] immediates.
16905 (movti): Split CONST_POLY_INT moves into two halves.
16906 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
16907 Split additions that need a temporary here if the destination
16908 is the stack pointer.
16909 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
16910 (*add<mode>3_poly_1): New instruction.
16911 (set_clobber_cc): New expander.
16913 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16915 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
16916 parameter and use it instead of GET_MODE_SIZE (innermode). Use
16917 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
16918 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
16919 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
16920 Change innermode from fixed_mode_size to machine_mode.
16921 (simplify_subreg): Update call accordingly. Handle a constant-sized
16922 subreg of a variable-length CONST_VECTOR.
16924 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16925 Alan Hayward <alan.hayward@arm.com>
16926 David Sherwood <david.sherwood@arm.com>
16928 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
16929 (add_offset_to_base): New function, split out from...
16930 (create_mem_ref): ...here. When handling a scale other than 1,
16931 check first whether the address is valid without the offset.
16932 Add it into the base if so, leaving the index and scale as-is.
16934 2018-01-12 Jakub Jelinek <jakub@redhat.com>
16937 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
16938 fold_for_warn before checking if arg2 is INTEGER_CST.
16940 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
16942 * config/rs6000/predicates.md (load_multiple_operation): Delete.
16943 (store_multiple_operation): Delete.
16944 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
16945 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
16946 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
16947 guarded by TARGET_STRING.
16948 (rs6000_output_load_multiple): Delete.
16949 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
16950 OPTION_MASK_STRING / TARGET_STRING handling.
16951 (print_operand) <'N', 'O'>: Add comment that these are unused now.
16952 (const rs6000_opt_masks) <"string">: Change mask to 0.
16953 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
16954 (MASK_STRING): Delete.
16955 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
16957 (load_multiple): Delete.
16964 (store_multiple): Delete.
16971 (movmemsi_8reg): Delete.
16972 (corresponding unnamed define_insn): Delete.
16973 (movmemsi_6reg): Delete.
16974 (corresponding unnamed define_insn): Delete.
16975 (movmemsi_4reg): Delete.
16976 (corresponding unnamed define_insn): Delete.
16977 (movmemsi_2reg): Delete.
16978 (corresponding unnamed define_insn): Delete.
16979 (movmemsi_1reg): Delete.
16980 (corresponding unnamed define_insn): Delete.
16981 * config/rs6000/rs6000.opt (mno-string): New.
16982 (mstring): Replace by deprecation warning stub.
16983 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
16985 2018-01-12 Jakub Jelinek <jakub@redhat.com>
16987 * regrename.c (regrename_do_replace): If replacing the same
16988 reg multiple times, try to reuse last created gen_raw_REG.
16991 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
16992 main to workaround a bug in GDB.
16994 2018-01-12 Tom de Vries <tom@codesourcery.com>
16997 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
16999 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
17001 PR rtl-optimization/80481
17002 * ira-color.c (get_cap_member): New function.
17003 (allocnos_conflict_by_live_ranges_p): Use it.
17004 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
17005 (setup_slot_coalesced_allocno_live_ranges): Ditto.
17007 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
17010 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
17011 (*saddl_se_1): Ditto.
17012 (*ssubsi_1): Ditto.
17013 (*ssubl_se_1): Ditto.
17015 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
17017 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
17018 rather than wi::to_widest for DR_INITs.
17019 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
17020 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
17021 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
17023 (vect_analyze_group_access_1): Note that here.
17025 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
17027 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
17028 polynomial type sizes.
17030 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
17032 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
17033 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
17034 (gimple_add_tmp_var): Likewise.
17036 2018-01-12 Martin Liska <mliska@suse.cz>
17038 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
17039 (gimple_alloc_sizes): Likewise.
17040 (dump_gimple_statistics): Use PRIu64 in printf format.
17041 * gimple.h: Change uint64_t to int.
17043 2018-01-12 Martin Liska <mliska@suse.cz>
17045 * tree-core.h: Use uint64_t instead of int.
17046 * tree.c (tree_node_counts): Likewise.
17047 (tree_node_sizes): Likewise.
17048 (dump_tree_statistics): Use PRIu64 in printf format.
17050 2018-01-12 Martin Liska <mliska@suse.cz>
17052 * Makefile.in: As qsort_chk is implemented in vec.c, add
17053 vec.o to linkage of gencfn-macros.
17054 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
17055 passing the info to record_node_allocation_statistics.
17056 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
17058 * ggc-common.c (struct ggc_usage): Add operator== and use
17059 it in operator< and compare function.
17060 * mem-stats.h (struct mem_usage): Likewise.
17061 * vec.c (struct vec_usage): Remove operator< and compare
17062 function. Can be simply inherited.
17064 2018-01-12 Martin Jambor <mjambor@suse.cz>
17067 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
17068 * tree-ssa-math-opts.c: Include domwalk.h.
17069 (convert_mult_to_fma_1): New function.
17070 (fma_transformation_info): New type.
17071 (fma_deferring_state): Likewise.
17072 (cancel_fma_deferring): New function.
17073 (result_of_phi): Likewise.
17074 (last_fma_candidate_feeds_initial_phi): Likewise.
17075 (convert_mult_to_fma): Added deferring logic, split actual
17076 transformation to convert_mult_to_fma_1.
17077 (math_opts_dom_walker): New type.
17078 (math_opts_dom_walker::after_dom_children): New method, body moved
17079 here from pass_optimize_widening_mul::execute, added deferring logic
17081 (pass_optimize_widening_mul::execute): Moved most of code to
17082 math_opts_dom_walker::after_dom_children.
17083 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
17084 * config/i386/i386.c (ix86_option_override_internal): Added
17085 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
17087 2018-01-12 Richard Biener <rguenther@suse.de>
17090 * dwarf2out.c (gen_variable_die): Do not reset old_die for
17091 inline instance vars.
17093 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
17096 * config/rx/rx.c (rx_is_restricted_memory_address):
17097 Handle SUBREG case.
17099 2018-01-12 Richard Biener <rguenther@suse.de>
17101 PR tree-optimization/80846
17102 * target.def (split_reduction): New target hook.
17103 * targhooks.c (default_split_reduction): New function.
17104 * targhooks.h (default_split_reduction): Declare.
17105 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
17106 target requests first reduce vectors by combining low and high
17108 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
17109 (get_vectype_for_scalar_type_and_size): Export.
17110 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
17111 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
17112 * doc/tm.texi: Regenerate.
17113 * config/i386/i386.c (ix86_split_reduction): Implement
17114 TARGET_VECTORIZE_SPLIT_REDUCTION.
17116 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
17119 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
17120 in PIC mode except for TARGET_VXWORKS_RTP.
17121 * config/sparc/sparc.c: Include cfgrtl.h.
17122 (TARGET_INIT_PIC_REG): Define.
17123 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
17124 (sparc_pic_register_p): New predicate.
17125 (sparc_legitimate_address_p): Use it.
17126 (sparc_legitimize_pic_address): Likewise.
17127 (sparc_delegitimize_address): Likewise.
17128 (sparc_mode_dependent_address_p): Likewise.
17129 (gen_load_pcrel_sym): Remove 4th parameter.
17130 (load_got_register): Adjust call to above. Remove obsolete stuff.
17131 (sparc_expand_prologue): Do not call load_got_register here.
17132 (sparc_flat_expand_prologue): Likewise.
17133 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
17134 (sparc_use_pseudo_pic_reg): New function.
17135 (sparc_init_pic_reg): Likewise.
17136 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
17137 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
17139 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
17141 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
17142 Add item for branch_cost.
17144 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
17146 PR rtl-optimization/83565
17147 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
17148 not extend the result to a larger mode for rotate operations.
17149 (num_sign_bit_copies1): Likewise.
17151 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17154 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
17156 Use values-Xc.o for -pedantic.
17157 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
17159 2018-01-12 Martin Liska <mliska@suse.cz>
17162 * ipa-devirt.c (final_warning_record::grow_type_warnings):
17164 (possible_polymorphic_call_targets): Use it.
17165 (ipa_devirt): Likewise.
17167 2018-01-12 Martin Liska <mliska@suse.cz>
17169 * profile-count.h (enum profile_quality): Use 0 as invalid
17170 enum value of profile_quality.
17172 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
17174 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
17175 -mext-string options.
17177 2018-01-12 Richard Biener <rguenther@suse.de>
17179 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
17180 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
17181 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
17183 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
17185 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
17187 * configure.ac (--with-long-double-format): Add support for the
17188 configuration option to change the default long double format on
17190 * config.gcc (powerpc*-linux*-*): Likewise.
17191 * configure: Regenerate.
17192 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
17193 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
17194 used without modification.
17196 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17198 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
17199 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
17200 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
17201 MISC_BUILTIN_SPEC_BARRIER.
17202 (rs6000_init_builtins): Likewise.
17203 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
17205 (speculation_barrier): New define_insn.
17206 * doc/extend.texi: Document __builtin_speculation_barrier.
17208 2018-01-11 Jakub Jelinek <jakub@redhat.com>
17211 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
17212 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
17213 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
17215 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
17216 integral modes instead of "ss" and "sd".
17217 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
17218 vectors with 32-bit and 64-bit elements.
17219 (vecdupssescalarmodesuffix): New mode attribute.
17220 (vec_dup<mode>): Use it.
17222 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
17225 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
17226 frame if argument is passed on stack.
17228 2018-01-11 Jakub Jelinek <jakub@redhat.com>
17231 * ree.c (combine_reaching_defs): Optimize also
17232 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
17233 reg2=any_extend(exp); reg1=reg2;, formatting fix.
17235 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
17237 PR middle-end/83189
17238 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
17240 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
17242 PR middle-end/83718
17243 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
17244 after they are computed.
17246 2018-01-11 Bin Cheng <bin.cheng@arm.com>
17248 PR tree-optimization/83695
17249 * gimple-loop-linterchange.cc
17250 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
17251 reset cached scev information after interchange.
17252 (pass_linterchange::execute): Remove call to scev_reset_htab.
17254 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17256 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
17257 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
17258 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
17259 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
17260 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
17261 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
17262 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
17263 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
17264 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
17265 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
17266 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
17267 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
17268 (V_lane_reg): Likewise.
17269 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
17271 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
17272 (vfmal_lane_low<mode>_intrinsic,
17273 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
17274 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
17275 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
17276 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
17277 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
17278 vfmsl_lane_high<mode>_intrinsic): New define_insns.
17280 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17282 * config/arm/arm-cpus.in (fp16fml): New feature.
17283 (ALL_SIMD): Add fp16fml.
17284 (armv8.2-a): Add fp16fml as an option.
17285 (armv8.3-a): Likewise.
17286 (armv8.4-a): Add fp16fml as part of fp16.
17287 * config/arm/arm.h (TARGET_FP16FML): Define.
17288 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
17290 * config/arm/arm-modes.def (V2HF): Define.
17291 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
17292 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
17293 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
17294 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
17295 vfmsl_low, vfmsl_high): New set of builtins.
17296 * config/arm/iterators.md (PLUSMINUS): New code iterator.
17297 (vfml_op): New code attribute.
17298 (VFMLHALVES): New int iterator.
17299 (VFML, VFMLSEL): New mode attributes.
17300 (V_reg): Define mapping for V2HF.
17301 (V_hi, V_lo): New mode attributes.
17302 (VF_constraint): Likewise.
17303 (vfml_half, vfml_half_selector): New int attributes.
17304 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
17306 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
17307 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
17309 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
17310 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
17311 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
17312 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
17314 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
17315 Document new effective target and option set.
17317 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17319 * config/arm/arm-cpus.in (armv8_4): New feature.
17320 (ARMv8_4a): New fgroup.
17321 (armv8.4-a): New arch.
17322 * config/arm/arm-tables.opt: Regenerate.
17323 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
17324 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
17325 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
17326 Add matching rules for -march=armv8.4-a and extensions.
17327 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
17329 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
17332 * config/rx/rx.md (BW): New mode attribute.
17333 (sync_lock_test_and_setsi): Add mode suffix to insn output.
17335 2018-01-11 Richard Biener <rguenther@suse.de>
17337 PR tree-optimization/83435
17338 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
17339 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
17340 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
17342 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
17343 Alan Hayward <alan.hayward@arm.com>
17344 David Sherwood <david.sherwood@arm.com>
17346 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
17348 (aarch64_classify_address): Initialize it. Track polynomial offsets.
17349 (aarch64_print_address_internal): Use it to check for a zero offset.
17351 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
17352 Alan Hayward <alan.hayward@arm.com>
17353 David Sherwood <david.sherwood@arm.com>
17355 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
17356 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
17357 Return a poly_int64 rather than a HOST_WIDE_INT.
17358 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
17359 rather than a HOST_WIDE_INT.
17360 * config/aarch64/aarch64.h (aarch64_frame): Protect with
17361 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
17362 hard_fp_offset, frame_size, initial_adjust, callee_offset and
17363 final_offset from HOST_WIDE_INT to poly_int64.
17364 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
17365 to_constant when getting the number of units in an Advanced SIMD
17367 (aarch64_builtin_vectorized_function): Check for a constant number
17369 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
17371 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
17372 attribute instead of GET_MODE_NUNITS.
17373 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
17374 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
17375 GET_MODE_SIZE for fixed-size registers.
17376 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
17377 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
17378 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
17379 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
17380 (aarch64_print_operand, aarch64_print_address_internal)
17381 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
17382 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
17383 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
17384 Handle polynomial GET_MODE_SIZE.
17385 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
17386 wider than SImode without modification.
17387 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
17388 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
17389 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
17390 passing and returning SVE modes.
17391 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
17392 rather than GEN_INT.
17393 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
17394 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
17395 (aarch64_allocate_and_probe_stack_space): Likewise.
17396 (aarch64_layout_frame): Cope with polynomial offsets.
17397 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
17398 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
17399 polynomial offsets.
17400 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
17401 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
17402 poly_int64 rather than a HOST_WIDE_INT.
17403 (aarch64_get_separate_components, aarch64_process_components)
17404 (aarch64_expand_prologue, aarch64_expand_epilogue)
17405 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
17406 (aarch64_anchor_offset): New function, split out from...
17407 (aarch64_legitimize_address): ...here.
17408 (aarch64_builtin_vectorization_cost): Handle polynomial
17409 TYPE_VECTOR_SUBPARTS.
17410 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
17412 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
17413 number of elements from the PARALLEL rather than the mode.
17414 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
17415 rather than GET_MODE_BITSIZE.
17416 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
17417 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
17418 (aarch64_expand_vec_perm_const_1): Handle polynomial
17419 d->perm.length () and d->perm elements.
17420 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
17421 Apply to_constant to d->perm elements.
17422 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
17423 polynomial CONST_VECTOR_NUNITS.
17424 (aarch64_move_pointer): Take amount as a poly_int64 rather
17426 (aarch64_progress_pointer): Avoid temporary variable.
17427 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
17428 the mode attribute instead of GET_MODE.
17430 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
17431 Alan Hayward <alan.hayward@arm.com>
17432 David Sherwood <david.sherwood@arm.com>
17434 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
17435 x exists before using it.
17436 (aarch64_add_constant_internal): Rename to...
17437 (aarch64_add_offset_1): ...this. Replace regnum with separate
17438 src and dest rtxes. Handle the case in which they're different,
17439 including when the offset is zero. Replace scratchreg with an rtx.
17440 Use 2 additions if there is no spare register into which we can
17441 move a 16-bit constant.
17442 (aarch64_add_constant): Delete.
17443 (aarch64_add_offset): Replace reg with separate src and dest
17444 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
17445 Use aarch64_add_offset_1.
17446 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
17447 an rtx rather than an int. Take the delta as a poly_int64
17448 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
17449 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
17450 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
17451 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
17452 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
17453 and aarch64_add_sp.
17454 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
17455 aarch64_add_constant.
17457 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
17459 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
17460 Use scalar_float_mode.
17462 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
17464 * config/aarch64/aarch64-simd.md
17465 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
17466 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
17467 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
17468 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
17469 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
17470 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
17471 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
17472 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
17473 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
17474 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
17476 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17479 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
17480 targ_options->x_arm_arch_string is non NULL.
17482 2018-01-11 Tamar Christina <tamar.christina@arm.com>
17484 * config/aarch64/aarch64.h
17485 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
17487 2018-01-11 Sudakshina Das <sudi.das@arm.com>
17490 * expmed.c (emit_store_flag_force): Swap if const op0
17491 and change VOIDmode to mode of op0.
17493 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
17495 PR rtl-optimization/83761
17496 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
17497 than bytes to mode_for_size.
17499 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
17501 PR middle-end/83189
17502 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
17503 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
17506 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
17508 PR middle-end/83575
17509 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
17510 when in layout mode.
17511 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
17512 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
17515 2018-01-10 Michael Collison <michael.collison@arm.com>
17517 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
17518 * config/aarch64/aarch64-option-extension.def: Add
17519 AARCH64_OPT_EXTENSION of 'fp16fml'.
17520 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
17521 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
17522 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
17523 * config/aarch64/constraints.md (Ui7): New constraint.
17524 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
17525 (VFMLA_SEL_W): Ditto.
17528 (VFMLA16_LOW): New int iterator.
17529 (VFMLA16_HIGH): Ditto.
17530 (UNSPEC_FMLAL): New unspec.
17531 (UNSPEC_FMLSL): Ditto.
17532 (UNSPEC_FMLAL2): Ditto.
17533 (UNSPEC_FMLSL2): Ditto.
17534 (f16mac): New code attribute.
17535 * config/aarch64/aarch64-simd-builtins.def
17536 (aarch64_fmlal_lowv2sf): Ditto.
17537 (aarch64_fmlsl_lowv2sf): Ditto.
17538 (aarch64_fmlalq_lowv4sf): Ditto.
17539 (aarch64_fmlslq_lowv4sf): Ditto.
17540 (aarch64_fmlal_highv2sf): Ditto.
17541 (aarch64_fmlsl_highv2sf): Ditto.
17542 (aarch64_fmlalq_highv4sf): Ditto.
17543 (aarch64_fmlslq_highv4sf): Ditto.
17544 (aarch64_fmlal_lane_lowv2sf): Ditto.
17545 (aarch64_fmlsl_lane_lowv2sf): Ditto.
17546 (aarch64_fmlal_laneq_lowv2sf): Ditto.
17547 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
17548 (aarch64_fmlalq_lane_lowv4sf): Ditto.
17549 (aarch64_fmlsl_lane_lowv4sf): Ditto.
17550 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
17551 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
17552 (aarch64_fmlal_lane_highv2sf): Ditto.
17553 (aarch64_fmlsl_lane_highv2sf): Ditto.
17554 (aarch64_fmlal_laneq_highv2sf): Ditto.
17555 (aarch64_fmlsl_laneq_highv2sf): Ditto.
17556 (aarch64_fmlalq_lane_highv4sf): Ditto.
17557 (aarch64_fmlsl_lane_highv4sf): Ditto.
17558 (aarch64_fmlalq_laneq_highv4sf): Ditto.
17559 (aarch64_fmlsl_laneq_highv4sf): Ditto.
17560 * config/aarch64/aarch64-simd.md:
17561 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
17562 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
17563 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
17564 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
17565 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
17566 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
17567 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
17568 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
17569 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
17570 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
17571 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
17572 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
17573 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
17574 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
17575 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
17576 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
17577 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
17578 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
17579 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
17580 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
17581 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
17582 (vfmlsl_low_u32): Ditto.
17583 (vfmlalq_low_u32): Ditto.
17584 (vfmlslq_low_u32): Ditto.
17585 (vfmlal_high_u32): Ditto.
17586 (vfmlsl_high_u32): Ditto.
17587 (vfmlalq_high_u32): Ditto.
17588 (vfmlslq_high_u32): Ditto.
17589 (vfmlal_lane_low_u32): Ditto.
17590 (vfmlsl_lane_low_u32): Ditto.
17591 (vfmlal_laneq_low_u32): Ditto.
17592 (vfmlsl_laneq_low_u32): Ditto.
17593 (vfmlalq_lane_low_u32): Ditto.
17594 (vfmlslq_lane_low_u32): Ditto.
17595 (vfmlalq_laneq_low_u32): Ditto.
17596 (vfmlslq_laneq_low_u32): Ditto.
17597 (vfmlal_lane_high_u32): Ditto.
17598 (vfmlsl_lane_high_u32): Ditto.
17599 (vfmlal_laneq_high_u32): Ditto.
17600 (vfmlsl_laneq_high_u32): Ditto.
17601 (vfmlalq_lane_high_u32): Ditto.
17602 (vfmlslq_lane_high_u32): Ditto.
17603 (vfmlalq_laneq_high_u32): Ditto.
17604 (vfmlslq_laneq_high_u32): Ditto.
17605 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
17606 (AARCH64_FL_FOR_ARCH8_4): New.
17607 (AARCH64_ISA_F16FML): New ISA flag.
17608 (TARGET_F16FML): New feature flag for fp16fml.
17609 (doc/invoke.texi): Document new fp16fml option.
17611 2018-01-10 Michael Collison <michael.collison@arm.com>
17613 * config/aarch64/aarch64-builtins.c:
17614 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
17615 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
17616 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
17617 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
17618 (AARCH64_ISA_SHA3): New ISA flag.
17619 (TARGET_SHA3): New feature flag for sha3.
17620 * config/aarch64/iterators.md (sha512_op): New int attribute.
17621 (CRYPTO_SHA512): New int iterator.
17622 (UNSPEC_SHA512H): New unspec.
17623 (UNSPEC_SHA512H2): Ditto.
17624 (UNSPEC_SHA512SU0): Ditto.
17625 (UNSPEC_SHA512SU1): Ditto.
17626 * config/aarch64/aarch64-simd-builtins.def
17627 (aarch64_crypto_sha512hqv2di): New builtin.
17628 (aarch64_crypto_sha512h2qv2di): Ditto.
17629 (aarch64_crypto_sha512su0qv2di): Ditto.
17630 (aarch64_crypto_sha512su1qv2di): Ditto.
17631 (aarch64_eor3qv8hi): Ditto.
17632 (aarch64_rax1qv2di): Ditto.
17633 (aarch64_xarqv2di): Ditto.
17634 (aarch64_bcaxqv8hi): Ditto.
17635 * config/aarch64/aarch64-simd.md:
17636 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
17637 (aarch64_crypto_sha512su0qv2di): Ditto.
17638 (aarch64_crypto_sha512su1qv2di): Ditto.
17639 (aarch64_eor3qv8hi): Ditto.
17640 (aarch64_rax1qv2di): Ditto.
17641 (aarch64_xarqv2di): Ditto.
17642 (aarch64_bcaxqv8hi): Ditto.
17643 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
17644 (vsha512h2q_u64): Ditto.
17645 (vsha512su0q_u64): Ditto.
17646 (vsha512su1q_u64): Ditto.
17647 (veor3q_u16): Ditto.
17648 (vrax1q_u64): Ditto.
17649 (vxarq_u64): Ditto.
17650 (vbcaxq_u16): Ditto.
17651 * config/arm/types.md (crypto_sha512): New type attribute.
17652 (crypto_sha3): Ditto.
17653 (doc/invoke.texi): Document new sha3 option.
17655 2018-01-10 Michael Collison <michael.collison@arm.com>
17657 * config/aarch64/aarch64-builtins.c:
17658 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
17659 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
17660 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
17661 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
17662 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
17663 (AARCH64_ISA_SM4): New ISA flag.
17664 (TARGET_SM4): New feature flag for sm4.
17665 * config/aarch64/aarch64-simd-builtins.def
17666 (aarch64_sm3ss1qv4si): Ditto.
17667 (aarch64_sm3tt1aq4si): Ditto.
17668 (aarch64_sm3tt1bq4si): Ditto.
17669 (aarch64_sm3tt2aq4si): Ditto.
17670 (aarch64_sm3tt2bq4si): Ditto.
17671 (aarch64_sm3partw1qv4si): Ditto.
17672 (aarch64_sm3partw2qv4si): Ditto.
17673 (aarch64_sm4eqv4si): Ditto.
17674 (aarch64_sm4ekeyqv4si): Ditto.
17675 * config/aarch64/aarch64-simd.md:
17676 (aarch64_sm3ss1qv4si): Ditto.
17677 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
17678 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
17679 (aarch64_sm4eqv4si): Ditto.
17680 (aarch64_sm4ekeyqv4si): Ditto.
17681 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
17682 (sm3part_op): Ditto.
17683 (CRYPTO_SM3TT): Ditto.
17684 (CRYPTO_SM3PART): Ditto.
17685 (UNSPEC_SM3SS1): New unspec.
17686 (UNSPEC_SM3TT1A): Ditto.
17687 (UNSPEC_SM3TT1B): Ditto.
17688 (UNSPEC_SM3TT2A): Ditto.
17689 (UNSPEC_SM3TT2B): Ditto.
17690 (UNSPEC_SM3PARTW1): Ditto.
17691 (UNSPEC_SM3PARTW2): Ditto.
17692 (UNSPEC_SM4E): Ditto.
17693 (UNSPEC_SM4EKEY): Ditto.
17694 * config/aarch64/constraints.md (Ui2): New constraint.
17695 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
17696 * config/arm/types.md (crypto_sm3): New type attribute.
17697 (crypto_sm4): Ditto.
17698 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
17699 (vsm3tt1aq_u32): Ditto.
17700 (vsm3tt1bq_u32): Ditto.
17701 (vsm3tt2aq_u32): Ditto.
17702 (vsm3tt2bq_u32): Ditto.
17703 (vsm3partw1q_u32): Ditto.
17704 (vsm3partw2q_u32): Ditto.
17705 (vsm4eq_u32): Ditto.
17706 (vsm4ekeyq_u32): Ditto.
17707 (doc/invoke.texi): Document new sm4 option.
17709 2018-01-10 Michael Collison <michael.collison@arm.com>
17711 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
17712 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
17713 (AARCH64_FL_FOR_ARCH8_4): New.
17714 (AARCH64_FL_V8_4): New flag.
17715 (doc/invoke.texi): Document new armv8.4-a option.
17717 2018-01-10 Michael Collison <michael.collison@arm.com>
17719 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
17720 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
17721 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
17722 * config/aarch64/aarch64-option-extension.def: Add
17723 AARCH64_OPT_EXTENSION of 'sha2'.
17724 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
17725 (crypto): Disable sha2 and aes if crypto disabled.
17726 (crypto): Enable aes and sha2 if enabled.
17727 (simd): Disable sha2 and aes if simd disabled.
17728 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
17730 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
17731 (TARGET_SHA2): New feature flag for sha2.
17732 (TARGET_AES): New feature flag for aes.
17733 * config/aarch64/aarch64-simd.md:
17734 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
17735 conditional on TARGET_AES.
17736 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
17737 (aarch64_crypto_sha1hsi): Make pattern conditional
17739 (aarch64_crypto_sha1hv4si): Ditto.
17740 (aarch64_be_crypto_sha1hv4si): Ditto.
17741 (aarch64_crypto_sha1su1v4si): Ditto.
17742 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
17743 (aarch64_crypto_sha1su0v4si): Ditto.
17744 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
17745 (aarch64_crypto_sha256su0v4si): Ditto.
17746 (aarch64_crypto_sha256su1v4si): Ditto.
17747 (doc/invoke.texi): Document new aes and sha2 options.
17749 2018-01-10 Martin Sebor <msebor@redhat.com>
17751 PR tree-optimization/83781
17752 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
17755 2018-01-11 Martin Sebor <msebor@gmail.com>
17756 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17758 PR tree-optimization/83501
17759 PR tree-optimization/81703
17761 * tree-ssa-strlen.c (get_string_cst): Rename...
17762 (get_string_len): ...to this. Handle global constants.
17763 (handle_char_store): Adjust.
17765 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
17766 Jim Wilson <jimw@sifive.com>
17768 * config/riscv/riscv-protos.h (riscv_output_return): New.
17769 * config/riscv/riscv.c (struct machine_function): New naked_p field.
17770 (riscv_attribute_table, riscv_output_return),
17771 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
17772 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
17773 (riscv_compute_frame_info): Only compute frame->mask if not a naked
17775 (riscv_expand_prologue): Add early return for naked function.
17776 (riscv_expand_epilogue): Likewise.
17777 (riscv_function_ok_for_sibcall): Return false for naked function.
17778 (riscv_set_current_function): New.
17779 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
17780 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
17781 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
17782 * doc/extend.texi (RISC-V Function Attributes): New.
17784 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
17786 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
17787 check for 128-bit long double before checking TCmode.
17788 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
17789 128-bit long doubles before checking TFmode or TCmode.
17790 (FLOAT128_IBM_P): Likewise.
17792 2018-01-10 Martin Sebor <msebor@redhat.com>
17794 PR tree-optimization/83671
17795 * builtins.c (c_strlen): Unconditionally return zero for the empty
17797 Use -Warray-bounds for warnings.
17798 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
17799 for non-constant array indices with COMPONENT_REF, arrays of
17800 arrays, and pointers to arrays.
17801 (gimple_fold_builtin_strlen): Determine and set length range for
17802 non-constant character arrays.
17804 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
17806 PR middle-end/81897
17807 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
17810 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
17812 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
17814 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
17817 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
17818 VECTOR_MEM_ALTIVEC_OR_VSX_P.
17819 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
17820 indexed_or_indirect_operand predicate.
17821 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
17822 (*vsx_le_perm_load_v8hi): Likewise.
17823 (*vsx_le_perm_load_v16qi): Likewise.
17824 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
17825 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
17826 (*vsx_le_perm_store_v8hi): Likewise.
17827 (*vsx_le_perm_store_v16qi): Likewise.
17828 (eight unnamed splitters): Likewise.
17830 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
17832 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
17833 * config/rs6000/emmintrin.h: Likewise.
17834 * config/rs6000/mmintrin.h: Likewise.
17835 * config/rs6000/xmmintrin.h: Likewise.
17837 2018-01-10 David Malcolm <dmalcolm@redhat.com>
17840 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
17842 * tree.c (tree_nop_conversion): Return true for location wrapper
17844 (maybe_wrap_with_location): New function.
17845 (selftest::check_strip_nops): New function.
17846 (selftest::test_location_wrappers): New function.
17847 (selftest::tree_c_tests): Call it.
17848 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
17849 (maybe_wrap_with_location): New decl.
17850 (EXPR_LOCATION_WRAPPER_P): New macro.
17851 (location_wrapper_p): New inline function.
17852 (tree_strip_any_location_wrapper): New inline function.
17854 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
17857 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
17858 stack_realign_offset for the largest alignment of stack slot
17860 (ix86_find_max_used_stack_alignment): New function.
17861 (ix86_finalize_stack_frame_flags): Use it. Set
17862 max_used_stack_alignment if we don't realign stack.
17863 * config/i386/i386.h (machine_function): Add
17864 max_used_stack_alignment.
17866 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
17868 * config/arm/arm.opt (-mbranch-cost): New option.
17869 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
17872 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
17875 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
17876 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
17878 2018-01-10 Richard Biener <rguenther@suse.de>
17881 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
17882 early out so it also covers the case where we have a non-NULL
17885 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
17887 PR tree-optimization/83753
17888 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
17889 for non-strided grouped accesses if the number of elements is 1.
17891 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
17894 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
17895 * i386.h (TARGET_USE_GATHER): Define.
17896 * x86-tune.def (X86_TUNE_USE_GATHER): New.
17898 2018-01-10 Martin Liska <mliska@suse.cz>
17901 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
17902 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
17904 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
17905 CLEANUP_NO_PARTITIONING is not set.
17907 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
17909 * doc/rtl.texi: Remove documentation of (const ...) wrappers
17910 for vectors, as a partial revert of r254296.
17911 * rtl.h (const_vec_p): Delete.
17912 (const_vec_duplicate_p): Don't test for vector CONSTs.
17913 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
17914 * expmed.c (make_tree): Likewise.
17917 * common.md (E, F): Use CONSTANT_P instead of checking for
17919 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
17920 checking for CONST_VECTOR.
17922 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
17924 PR middle-end/83575
17925 * predict.c (force_edge_cold): Handle in more sane way edges
17926 with no prediction.
17928 2018-01-09 Carl Love <cel@us.ibm.com>
17930 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
17932 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
17933 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
17934 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
17935 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
17936 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
17937 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
17938 * config/rs6000/rs6000-protos.h: Add extern defition for
17939 rs6000_generate_float2_double_code.
17940 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
17942 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
17943 (float2_v2df): Add define_expand.
17945 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
17948 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
17949 op_mode in the force_to_mode call.
17951 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
17953 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
17954 instead of checking each element individually.
17955 (aarch64_evpc_uzp): Likewise.
17956 (aarch64_evpc_zip): Likewise.
17957 (aarch64_evpc_ext): Likewise.
17958 (aarch64_evpc_rev): Likewise.
17959 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
17960 instead of checking each element individually. Return true without
17962 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
17963 whether all selected elements come from the same input, instead of
17964 checking each element individually. Remove calls to gen_rtx_REG,
17965 start_sequence and end_sequence and instead assert that no rtl is
17968 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
17970 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
17971 order of HIGH and CONST checks.
17973 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
17975 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
17976 if the destination isn't an SSA_NAME.
17978 2018-01-09 Richard Biener <rguenther@suse.de>
17980 PR tree-optimization/83668
17981 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
17983 (canonicalize_loop_form): ... here, renamed from ...
17984 (canonicalize_loop_closed_ssa_form): ... this and amended to
17985 swap successor edges for loop exit blocks to make us use
17986 the RPO order we need for initial schedule generation.
17988 2018-01-09 Joseph Myers <joseph@codesourcery.com>
17990 PR tree-optimization/64811
17991 * match.pd: When optimizing comparisons with Inf, avoid
17992 introducing or losing exceptions from comparisons with NaN.
17994 2018-01-09 Martin Liska <mliska@suse.cz>
17997 * asan.c (shadow_mem_size): Add gcc_assert.
17999 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
18001 Don't save registers in main().
18004 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
18005 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
18006 * config/avr/avr.c (avr_set_current_function): Don't error if
18007 naked, OS_task or OS_main are specified at the same time.
18008 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
18010 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
18012 * common/config/avr/avr-common.c (avr_option_optimization_table):
18013 Switch on -mmain-is-OS_task for optimizing compilations.
18015 2018-01-09 Richard Biener <rguenther@suse.de>
18017 PR tree-optimization/83572
18018 * graphite.c: Include cfganal.h.
18019 (graphite_transform_loops): Connect infinite loops to exit
18020 and remove fake edges at the end.
18022 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
18024 * ipa-inline.c (edge_badness): Revert accidental checkin.
18026 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
18029 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
18030 symbols; not inline clones.
18032 2018-01-09 Jakub Jelinek <jakub@redhat.com>
18035 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
18036 hard registers. Formatting fixes.
18038 PR preprocessor/83722
18039 * gcc.c (try_generate_repro): Pass
18040 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
18041 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
18044 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
18045 Kito Cheng <kito.cheng@gmail.com>
18047 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
18048 (riscv_leaf_function_p): Delete.
18049 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
18051 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18053 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
18055 (do_ifelse): New function.
18056 (do_isel): New function.
18057 (do_sub3): New function.
18058 (do_add3): New function.
18059 (do_load_mask_compare): New function.
18060 (do_overlap_load_compare): New function.
18061 (expand_compare_loop): New function.
18062 (expand_block_compare): Call expand_compare_loop() when appropriate.
18063 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
18064 option description.
18065 (-mblock-compare-inline-loop-limit): New option.
18067 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18070 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
18071 Reverse order of second and third operands in first alternative.
18072 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
18073 of first and second elements in UNSPEC_VPERMR vector.
18074 (altivec_expand_vec_perm_le): Likewise.
18076 2018-01-08 Jeff Law <law@redhat.com>
18078 PR rtl-optimizatin/81308
18079 * tree-switch-conversion.c (cfg_altered): New file scoped static.
18080 (process_switch): If group_case_labels makes a change, then set
18082 (pass_convert_switch::execute): If a switch is converted, then
18083 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
18085 PR rtl-optimization/81308
18086 * recog.c (split_all_insns): Conditionally cleanup the CFG after
18089 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
18091 PR target/83663 - Revert r255946
18092 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
18093 generation for cases where splatting a value is not useful.
18094 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
18095 across a vec_duplicate and a paradoxical subreg forming a vector
18096 mode to a vec_concat.
18098 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18100 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
18101 -march=armv8.3-a variants.
18102 * config/arm/t-multilib: Likewise.
18103 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
18105 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18107 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
18109 (cceq_ior_compare_complement): Give it a name so I can use it, and
18110 change boolean_or_operator predicate to boolean_operator so it can
18111 be used to generate a crand.
18112 (eqne): New code iterator.
18113 (bd/bd_neg): New code_attrs.
18114 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
18115 a single define_insn.
18116 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
18117 decrement (bdnzt/bdnzf/bdzt/bdzf).
18118 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
18119 with the new names of the branch decrement patterns, and added the
18120 names of the branch decrement conditional patterns.
18122 2018-01-08 Richard Biener <rguenther@suse.de>
18124 PR tree-optimization/83563
18125 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
18128 2018-01-08 Richard Biener <rguenther@suse.de>
18130 PR middle-end/83713
18131 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
18133 2018-01-08 Richard Biener <rguenther@suse.de>
18135 PR tree-optimization/83685
18136 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
18137 references to abnormals.
18139 2018-01-08 Richard Biener <rguenther@suse.de>
18142 * dwarf2out.c (output_indirect_strings): Handle empty
18143 skeleton_debug_str_hash.
18144 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
18146 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
18148 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
18149 (emit_store_direct): Likewise.
18150 (arc_trampoline_adjust_address): Likewise.
18151 (arc_asm_trampoline_template): New function.
18152 (arc_initialize_trampoline): Use asm_trampoline_template.
18153 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
18154 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
18155 * config/arc/arc.md (flush_icache): Delete pattern.
18157 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
18159 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
18160 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
18163 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
18166 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
18167 by not USED_FOR_TARGET.
18168 (make_pass_resolve_sw_modes): Likewise.
18170 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
18172 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
18175 2018-01-08 Richard Biener <rguenther@suse.de>
18177 PR middle-end/83580
18178 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
18180 2018-01-08 Richard Biener <rguenther@suse.de>
18182 PR middle-end/83517
18183 * match.pd ((t * 2) / 2) -> t): Add missing :c.
18185 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
18187 PR middle-end/81897
18188 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
18189 basic blocks with a small number of successors.
18190 (convert_control_dep_chain_into_preds): Improve handling of
18192 (dump_predicates): Split apart into...
18193 (dump_pred_chain): ...here...
18194 (dump_pred_info): ...and here.
18195 (can_one_predicate_be_invalidated_p): Add debugging printfs.
18196 (can_chain_union_be_invalidated_p): Improve check for invalidation
18198 (uninit_uses_cannot_happen): Avoid unnecessary if
18199 convert_control_dep_chain_into_preds yielded nothing.
18201 2018-01-06 Martin Sebor <msebor@redhat.com>
18203 PR tree-optimization/83640
18204 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
18205 subtracting negative offset from size.
18206 (builtin_access::overlap): Adjust offset bounds of the access to fall
18207 within the size of the object if possible.
18209 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
18211 PR rtl-optimization/83699
18212 * expmed.c (extract_bit_field_1): Restrict the vector usage of
18213 extract_bit_field_as_subreg to cases in which the extracted
18214 value is also a vector.
18216 * lra-constraints.c (process_alt_operands): Test for the equivalence
18217 substitutions when detecting a possible reload cycle.
18219 2018-01-06 Jakub Jelinek <jakub@redhat.com>
18222 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
18223 by default if flag_selective_schedling{,2}. Formatting fixes.
18225 PR rtl-optimization/83682
18226 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
18227 if it has non-VECTOR_MODE element mode.
18228 (vec_duplicate_p): Likewise.
18230 PR middle-end/83694
18231 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
18232 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
18234 2018-01-05 Jakub Jelinek <jakub@redhat.com>
18237 * config/i386/i386-builtin.def
18238 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
18239 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
18240 Require also OPTION_MASK_ISA_AVX512F in addition to
18241 OPTION_MASK_ISA_GFNI.
18242 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
18243 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
18244 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
18245 to OPTION_MASK_ISA_GFNI.
18246 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
18247 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
18248 OPTION_MASK_ISA_AVX512BW.
18249 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
18250 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
18251 addition to OPTION_MASK_ISA_GFNI.
18252 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
18253 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
18254 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
18255 to OPTION_MASK_ISA_GFNI.
18256 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
18257 a requirement for all ISAs rather than any of them with a few
18259 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
18261 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
18262 bitmasks to be enabled with 3 exceptions, instead of requiring any
18263 enabled ISA with lots of exceptions.
18264 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
18265 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
18266 Change avx512bw in isa attribute to avx512f.
18267 * config/i386/sgxintrin.h: Add license boilerplate.
18268 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
18269 to __AVX512F__ and __AVX512VL to __AVX512VL__.
18270 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
18271 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
18273 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
18274 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
18275 temporarily sse2 rather than sse if not enabled already.
18278 * config/i386/sse.md (VI248_VLBW): Rename to ...
18279 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
18280 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
18281 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
18282 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
18283 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
18284 mode iterator instead of VI248_VLBW.
18286 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
18288 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
18289 (record_modified): Skip clobbers; add debug output.
18290 (param_change_prob): Use sreal frequencies.
18292 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
18294 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
18295 punt for user-aligned variables.
18297 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
18299 * tree-chrec.c (chrec_contains_symbols): Return true for
18302 2018-01-05 Sudakshina Das <sudi.das@arm.com>
18305 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
18306 of (x|y) == x for BICS pattern.
18308 2018-01-05 Jakub Jelinek <jakub@redhat.com>
18310 PR tree-optimization/83605
18311 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
18312 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
18315 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
18317 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
18318 * config/epiphany/rtems.h: New file.
18320 2018-01-04 Jakub Jelinek <jakub@redhat.com>
18321 Uros Bizjak <ubizjak@gmail.com>
18324 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
18325 QIreg_operand instead of register_operand predicate.
18326 * config/i386/i386.c (ix86_rop_should_change_byte_p,
18327 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
18328 comments instead of -fmitigate[-_]rop.
18330 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18333 * cgraphunit.c (symbol_table::compile): Switch to text_section
18334 before calling assembly_start debug hook.
18335 * run-rtl-passes.c (run_rtl_passes): Likewise.
18338 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
18340 * tree-vrp.c (extract_range_from_binary_expr_1): Check
18341 range_int_cst_p rather than !symbolic_range_p before calling
18342 extract_range_from_multiplicative_op_1.
18344 2018-01-04 Jeff Law <law@redhat.com>
18346 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
18347 redundant test in assertion.
18349 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
18351 * doc/rtl.texi: Document machine_mode wrapper classes.
18353 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
18355 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
18356 using tree_to_uhwi.
18358 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
18360 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
18361 the VEC_PERM_EXPR fold to fail.
18363 2018-01-04 Jakub Jelinek <jakub@redhat.com>
18366 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
18367 to switched_sections.
18369 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
18372 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
18373 test for d.testing.
18375 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
18378 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
18379 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
18381 2018-01-04 Jakub Jelinek <jakub@redhat.com>
18384 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
18385 is BLKmode and bitpos not zero or mode change is needed.
18387 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
18390 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
18393 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
18396 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
18397 instead of MULT rtx. Update all corresponding splitters.
18398 (*saddl_se): Ditto.
18399 (*ssub<modesuffix>): Ditto.
18400 (*ssubl_se): Ditto.
18401 (*cmp_sadd_di): Update split patterns.
18402 (*cmp_sadd_si): Ditto.
18403 (*cmp_sadd_sidi): Ditto.
18404 (*cmp_ssub_di): Ditto.
18405 (*cmp_ssub_si): Ditto.
18406 (*cmp_ssub_sidi): Ditto.
18407 * config/alpha/predicates.md (const23_operand): New predicate.
18408 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
18409 Look for ASHIFT, not MULT inner operand.
18410 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
18412 2018-01-04 Martin Liska <mliska@suse.cz>
18414 PR gcov-profile/83669
18415 * gcov.c (output_intermediate_file): Add version to intermediate
18417 * doc/gcov.texi: Document new field 'version' in intermediate
18418 file format. Fix location of '-k' option of gcov command.
18420 2018-01-04 Martin Liska <mliska@suse.cz>
18423 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
18425 2018-01-04 Jakub Jelinek <jakub@redhat.com>
18427 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
18429 2018-01-03 Martin Sebor <msebor@redhat.com>
18431 PR tree-optimization/83655
18432 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
18433 checking calls with invalid arguments.
18435 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18437 * tree-vect-stmts.c (vect_get_store_rhs): New function.
18438 (vectorizable_mask_load_store): Delete.
18439 (vectorizable_call): Return false for masked loads and stores.
18440 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
18441 instead of gimple_assign_rhs1.
18442 (vectorizable_load): Handle IFN_MASK_LOAD.
18443 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
18445 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18447 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
18449 (vectorizable_mask_load_store): ...here.
18450 (vectorizable_load): ...and here.
18452 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18454 * tree-vect-stmts.c (vect_build_all_ones_mask)
18455 (vect_build_zero_merge_argument): New functions, split out from...
18456 (vectorizable_load): ...here.
18458 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18460 * tree-vect-stmts.c (vect_check_store_rhs): New function,
18462 (vectorizable_mask_load_store): ...here.
18463 (vectorizable_store): ...and here.
18465 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18467 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
18469 (vectorizable_mask_load_store): ...here.
18471 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18473 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
18474 (vect_model_store_cost): Take a vec_load_store_type instead of a
18476 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
18477 (vect_model_store_cost): Take a vec_load_store_type instead of a
18479 (vectorizable_mask_load_store): Update accordingly.
18480 (vectorizable_store): Likewise.
18481 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
18483 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18485 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
18486 IFN_MASK_LOAD calls here rather than...
18487 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
18489 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18490 Alan Hayward <alan.hayward@arm.com>
18491 David Sherwood <david.sherwood@arm.com>
18493 * expmed.c (extract_bit_field_1): For vector extracts,
18494 fall back to extract_bit_field_as_subreg if vec_extract
18497 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18498 Alan Hayward <alan.hayward@arm.com>
18499 David Sherwood <david.sherwood@arm.com>
18501 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
18502 they are variable or constant sized.
18503 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
18504 slots for constant-sized data.
18506 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18507 Alan Hayward <alan.hayward@arm.com>
18508 David Sherwood <david.sherwood@arm.com>
18510 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
18511 handling COND_EXPRs with boolean comparisons, try to find a better
18512 basis for the mask type than the boolean itself.
18514 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18516 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
18517 is calculated and how it can be overridden.
18518 * genmodes.c (max_bitsize_mode_any_mode): New variable.
18519 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
18521 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
18524 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18525 Alan Hayward <alan.hayward@arm.com>
18526 David Sherwood <david.sherwood@arm.com>
18528 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
18529 Remove the mode argument.
18530 (aarch64_simd_valid_immediate): Remove the mode and inverse
18532 * config/aarch64/iterators.md (bitsize): New iterator.
18533 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
18534 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
18535 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
18536 aarch64_simd_valid_immediate.
18537 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
18538 (aarch64_reg_or_bic_imm): Likewise.
18539 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
18540 with an insn_type enum and msl with a modifier_type enum.
18541 Replace element_width with a scalar_mode. Change the shift
18542 to unsigned int. Add constructors for scalar_float_mode and
18543 scalar_int_mode elements.
18544 (aarch64_vect_float_const_representable_p): Delete.
18545 (aarch64_can_const_movi_rtx_p)
18546 (aarch64_simd_scalar_immediate_valid_for_move)
18547 (aarch64_simd_make_constant): Update call to
18548 aarch64_simd_valid_immediate.
18549 (aarch64_advsimd_valid_immediate_hs): New function.
18550 (aarch64_advsimd_valid_immediate): Likewise.
18551 (aarch64_simd_valid_immediate): Remove mode and inverse
18552 arguments. Rewrite to use the above. Use const_vec_duplicate_p
18553 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
18554 and aarch64_float_const_representable_p on the result.
18555 (aarch64_output_simd_mov_immediate): Remove mode argument.
18556 Update call to aarch64_simd_valid_immediate and use of
18557 simd_immediate_info.
18558 (aarch64_output_scalar_simd_mov_immediate): Update call
18561 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18562 Alan Hayward <alan.hayward@arm.com>
18563 David Sherwood <david.sherwood@arm.com>
18565 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
18566 (mode_nunits): Likewise CONST_MODE_NUNITS.
18567 * machmode.def (ADJUST_NUNITS): Document.
18568 * genmodes.c (mode_data::need_nunits_adj): New field.
18569 (blank_mode): Update accordingly.
18570 (adj_nunits): New variable.
18571 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
18573 (emit_mode_size_inline): Set need_bytesize_adj for all modes
18574 listed in adj_nunits.
18575 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
18576 listed in adj_nunits. Don't emit case statements for such modes.
18577 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
18578 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
18579 nothing if adj_nunits is nonnull.
18580 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
18581 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
18582 (emit_mode_fbit): Update use of print_maybe_const_decl.
18583 (emit_move_size): Likewise. Treat the array as non-const
18585 (emit_mode_adjustments): Handle adj_nunits.
18587 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18589 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
18590 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
18591 (VECTOR_MODES): Use it.
18592 (make_vector_modes): Take the prefix as an argument.
18594 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18595 Alan Hayward <alan.hayward@arm.com>
18596 David Sherwood <david.sherwood@arm.com>
18598 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
18599 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
18600 for MODE_VECTOR_BOOL.
18601 * machmode.def (VECTOR_BOOL_MODE): Document.
18602 * genmodes.c (VECTOR_BOOL_MODE): New macro.
18603 (make_vector_bool_mode): New function.
18604 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
18606 * lto-streamer-in.c (lto_input_mode_table): Likewise.
18607 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
18609 * stor-layout.c (int_mode_for_mode): Likewise.
18610 * tree.c (build_vector_type_for_mode): Likewise.
18611 * varasm.c (output_constant_pool_2): Likewise.
18612 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
18613 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
18614 for MODE_VECTOR_BOOL.
18615 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
18616 of mode class checks.
18617 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
18618 instead of a list of mode class checks.
18619 (expand_vector_scalar_condition): Likewise.
18620 (type_for_widest_vector_mode): Handle BImode as an inner mode.
18622 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18623 Alan Hayward <alan.hayward@arm.com>
18624 David Sherwood <david.sherwood@arm.com>
18626 * machmode.h (mode_size): Change from unsigned short to
18628 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
18629 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
18630 or if measurement_type is not polynomial.
18631 (fixed_size_mode::includes_p): Check for constant-sized modes.
18632 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
18633 return a poly_uint16 rather than an unsigned short.
18634 (emit_mode_size): Change the type of mode_size from unsigned short
18635 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
18636 (emit_mode_adjustments): Cope with polynomial vector sizes.
18637 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
18639 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
18641 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
18642 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
18643 * caller-save.c (setup_save_areas): Likewise.
18644 (replace_reg_with_saved_mem): Likewise.
18645 * calls.c (emit_library_call_value_1): Likewise.
18646 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
18647 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
18648 (gen_lowpart_for_combine): Likewise.
18649 * convert.c (convert_to_integer_1): Likewise.
18650 * cse.c (equiv_constant, cse_insn): Likewise.
18651 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
18652 (cselib_subst_to_values): Likewise.
18653 * dce.c (word_dce_process_block): Likewise.
18654 * df-problems.c (df_word_lr_mark_ref): Likewise.
18655 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
18656 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
18657 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
18658 (rtl_for_decl_location): Likewise.
18659 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
18660 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
18661 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
18662 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
18663 (expand_expr_real_1): Likewise.
18664 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
18665 (pad_below): Likewise.
18666 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
18667 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
18668 * ira.c (get_subreg_tracking_sizes): Likewise.
18669 * ira-build.c (ira_create_allocno_objects): Likewise.
18670 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
18671 (ira_sort_regnos_for_alter_reg): Likewise.
18672 * ira-costs.c (record_operand_costs): Likewise.
18673 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
18674 (resolve_simple_move): Likewise.
18675 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
18676 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
18677 (lra_constraints): Likewise.
18678 (CONST_POOL_OK_P): Reject variable-sized modes.
18679 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
18680 (add_pseudo_to_slot, lra_spill): Likewise.
18681 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18682 * optabs-query.c (get_best_extraction_insn): Likewise.
18683 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
18684 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
18685 (expand_mult_highpart, valid_multiword_target_p): Likewise.
18686 * recog.c (offsettable_address_addr_space_p): Likewise.
18687 * regcprop.c (maybe_mode_change): Likewise.
18688 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
18689 * regrename.c (build_def_use): Likewise.
18690 * regstat.c (dump_reg_info): Likewise.
18691 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
18692 (find_reloads, find_reloads_subreg_address): Likewise.
18693 * reload1.c (eliminate_regs_1): Likewise.
18694 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
18695 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
18696 (simplify_binary_operation_1, simplify_subreg): Likewise.
18697 * targhooks.c (default_function_arg_padding): Likewise.
18698 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
18699 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
18700 (verify_gimple_assign_ternary): Likewise.
18701 * tree-inline.c (estimate_move_cost): Likewise.
18702 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
18703 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
18704 (get_address_cost_ainc): Likewise.
18705 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
18706 (vect_supportable_dr_alignment): Likewise.
18707 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
18708 (vectorizable_reduction): Likewise.
18709 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
18710 (vectorizable_operation, vectorizable_load): Likewise.
18711 * tree.c (build_same_sized_truth_vector_type): Likewise.
18712 * valtrack.c (cleanup_auto_inc_dec): Likewise.
18713 * var-tracking.c (emit_note_insn_var_location): Likewise.
18714 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
18715 (ADDR_VEC_ALIGN): Likewise.
18717 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18718 Alan Hayward <alan.hayward@arm.com>
18719 David Sherwood <david.sherwood@arm.com>
18721 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
18723 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
18724 or if measurement_type is polynomial.
18725 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
18726 * combine.c (make_extraction): Likewise.
18727 * dse.c (find_shift_sequence): Likewise.
18728 * dwarf2out.c (mem_loc_descriptor): Likewise.
18729 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
18730 (extract_bit_field, extract_low_bits): Likewise.
18731 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
18732 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
18733 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
18734 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
18735 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
18736 * reload.c (find_reloads): Likewise.
18737 * reload1.c (alter_reg): Likewise.
18738 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
18739 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
18740 * tree-if-conv.c (predicate_mem_writes): Likewise.
18741 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
18742 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
18743 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
18744 * valtrack.c (dead_debug_insert_temp): Likewise.
18745 * varasm.c (mergeable_constant_section): Likewise.
18746 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
18748 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18749 Alan Hayward <alan.hayward@arm.com>
18750 David Sherwood <david.sherwood@arm.com>
18752 * expr.c (expand_assignment): Cope with polynomial mode sizes
18753 when assigning to a CONCAT.
18755 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18756 Alan Hayward <alan.hayward@arm.com>
18757 David Sherwood <david.sherwood@arm.com>
18759 * machmode.h (mode_precision): Change from unsigned short to
18761 (mode_to_precision): Return a poly_uint16 rather than an unsigned
18763 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
18764 or if measurement_type is not polynomial.
18765 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
18766 in which the mode is already known to be a scalar_int_mode.
18767 * genmodes.c (emit_mode_precision): Change the type of mode_precision
18768 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
18770 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
18771 for GET_MODE_PRECISION.
18772 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
18773 for GET_MODE_PRECISION.
18774 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
18776 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
18777 (expand_field_assignment, make_extraction): Likewise.
18778 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
18779 (get_last_value): Likewise.
18780 * convert.c (convert_to_integer_1): Likewise.
18781 * cse.c (cse_insn): Likewise.
18782 * expr.c (expand_expr_real_1): Likewise.
18783 * lra-constraints.c (simplify_operand_subreg): Likewise.
18784 * optabs-query.c (can_atomic_load_p): Likewise.
18785 * optabs.c (expand_atomic_load): Likewise.
18786 (expand_atomic_store): Likewise.
18787 * ree.c (combine_reaching_defs): Likewise.
18788 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
18789 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
18790 * tree.h (type_has_mode_precision_p): Likewise.
18791 * ubsan.c (instrument_si_overflow): Likewise.
18793 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18794 Alan Hayward <alan.hayward@arm.com>
18795 David Sherwood <david.sherwood@arm.com>
18797 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
18798 polynomial numbers of units.
18799 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
18800 (valid_vector_subparts_p): New function.
18801 (build_vector_type): Remove temporary shim and take the number
18802 of units as a poly_uint64 rather than an int.
18803 (build_opaque_vector_type): Take the number of units as a
18804 poly_uint64 rather than an int.
18805 * tree.c (build_vector_from_ctor): Handle polynomial
18806 TYPE_VECTOR_SUBPARTS.
18807 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
18808 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
18809 (build_vector_from_val): If the number of units is variable,
18810 use build_vec_duplicate_cst for constant operands and
18811 VEC_DUPLICATE_EXPR otherwise.
18812 (make_vector_type): Remove temporary is_constant ().
18813 (build_vector_type, build_opaque_vector_type): Take the number of
18814 units as a poly_uint64 rather than an int.
18815 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
18817 * cfgexpand.c (expand_debug_expr): Likewise.
18818 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
18819 (store_constructor, expand_expr_real_1): Likewise.
18820 (const_scalar_mask_from_tree): Likewise.
18821 * fold-const-call.c (fold_const_reduction): Likewise.
18822 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
18823 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
18824 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
18825 (fold_relational_const): Likewise.
18826 (native_interpret_vector): Likewise. Change the size from an
18827 int to an unsigned int.
18828 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
18829 TYPE_VECTOR_SUBPARTS.
18830 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
18831 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
18832 duplicating a non-constant operand into a variable-length vector.
18833 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
18834 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
18835 * ipa-icf.c (sem_variable::equals): Likewise.
18836 * match.pd: Likewise.
18837 * omp-simd-clone.c (simd_clone_subparts): Likewise.
18838 * print-tree.c (print_node): Likewise.
18839 * stor-layout.c (layout_type): Likewise.
18840 * targhooks.c (default_builtin_vectorization_cost): Likewise.
18841 * tree-cfg.c (verify_gimple_comparison): Likewise.
18842 (verify_gimple_assign_binary): Likewise.
18843 (verify_gimple_assign_ternary): Likewise.
18844 (verify_gimple_assign_single): Likewise.
18845 * tree-pretty-print.c (dump_generic_node): Likewise.
18846 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
18847 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
18848 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
18849 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
18850 (vect_shift_permute_load_chain): Likewise.
18851 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
18852 (expand_vector_condition, optimize_vector_constructor): Likewise.
18853 (lower_vec_perm, get_compute_type): Likewise.
18854 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
18855 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
18856 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
18857 (vect_recog_mask_conversion_pattern): Likewise.
18858 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
18859 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
18860 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
18861 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
18862 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
18863 (vectorizable_shift, vectorizable_operation, vectorizable_store)
18864 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
18865 (supportable_widening_operation): Likewise.
18866 (supportable_narrowing_operation): Likewise.
18867 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
18869 * varasm.c (output_constant): Likewise.
18871 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18872 Alan Hayward <alan.hayward@arm.com>
18873 David Sherwood <david.sherwood@arm.com>
18875 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
18876 so that both the length == 3 and length != 3 cases set up their
18877 own permute vectors. Add comments explaining why we know the
18878 number of elements is constant.
18879 (vect_permute_load_chain): Likewise.
18881 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18882 Alan Hayward <alan.hayward@arm.com>
18883 David Sherwood <david.sherwood@arm.com>
18885 * machmode.h (mode_nunits): Change from unsigned char to
18887 (ONLY_FIXED_SIZE_MODES): New macro.
18888 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
18889 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
18890 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
18892 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
18893 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
18894 or if measurement_type is not polynomial.
18895 * genmodes.c (ZERO_COEFFS): New macro.
18896 (emit_mode_nunits_inline): Make mode_nunits_inline return a
18898 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
18899 Use ZERO_COEFFS when emitting initializers.
18900 * data-streamer.h (bp_pack_poly_value): New function.
18901 (bp_unpack_poly_value): Likewise.
18902 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
18903 for GET_MODE_NUNITS.
18904 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
18905 for GET_MODE_NUNITS.
18906 * tree.c (make_vector_type): Remove temporary shim and make
18907 the real function take the number of units as a poly_uint64
18908 rather than an int.
18909 (build_vector_type_for_mode): Handle polynomial nunits.
18910 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
18911 * emit-rtl.c (const_vec_series_p_1): Likewise.
18912 (gen_rtx_CONST_VECTOR): Likewise.
18913 * fold-const.c (test_vec_duplicate_folding): Likewise.
18914 * genrecog.c (validate_pattern): Likewise.
18915 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
18916 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
18917 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
18918 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
18919 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
18920 * rtlanal.c (subreg_get_info): Likewise.
18921 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
18922 (vect_grouped_load_supported): Likewise.
18923 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
18924 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
18925 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18926 (simplify_const_unary_operation, simplify_binary_operation_1)
18927 (simplify_const_binary_operation, simplify_ternary_operation)
18928 (test_vector_ops_duplicate, test_vector_ops): Likewise.
18929 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
18930 instead of CONST_VECTOR_NUNITS.
18931 * varasm.c (output_constant_pool_2): Likewise.
18932 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
18933 explicit-encoded elements in the XVEC for variable-length vectors.
18935 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18937 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
18939 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18940 Alan Hayward <alan.hayward@arm.com>
18941 David Sherwood <david.sherwood@arm.com>
18943 * coretypes.h (fixed_size_mode): Declare.
18944 (fixed_size_mode_pod): New typedef.
18945 * builtins.h (target_builtins::x_apply_args_mode)
18946 (target_builtins::x_apply_result_mode): Change type to
18947 fixed_size_mode_pod.
18948 * builtins.c (apply_args_size, apply_result_size, result_vector)
18949 (expand_builtin_apply_args_1, expand_builtin_apply)
18950 (expand_builtin_return): Update accordingly.
18952 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18954 * cse.c (hash_rtx_cb): Hash only the encoded elements.
18955 * cselib.c (cselib_hash_rtx): Likewise.
18956 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
18957 CONST_VECTOR encoding.
18959 2018-01-03 Jakub Jelinek <jakub@redhat.com>
18960 Jeff Law <law@redhat.com>
18963 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
18964 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
18965 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
18966 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
18969 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
18970 explicitly probe *sp in a noreturn function if there were any callee
18971 register saves or frame pointer is needed.
18973 2018-01-03 Jakub Jelinek <jakub@redhat.com>
18976 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
18977 BLKmode for ternary, binary or unary expressions.
18980 * var-tracking.c (delete_vta_debug_insn): New inline function.
18981 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
18982 insns from get_insns () to NULL instead of each bb separately.
18983 Use delete_vta_debug_insn. No longer static.
18984 (vt_debug_insns_local, variable_tracking_main_1): Adjust
18985 delete_vta_debug_insns callers.
18986 * rtl.h (delete_vta_debug_insns): Declare.
18987 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
18988 instead of variable_tracking_main.
18990 2018-01-03 Martin Sebor <msebor@redhat.com>
18992 PR tree-optimization/83603
18993 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
18994 arguments past the endof the argument list in functions declared
18995 without a prototype.
18996 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
18997 Avoid checking when arguments are null.
18999 2018-01-03 Martin Sebor <msebor@redhat.com>
19002 * doc/extend.texi (attribute const): Fix a typo.
19003 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
19004 issuing -Wsuggest-attribute for void functions.
19006 2018-01-03 Martin Sebor <msebor@redhat.com>
19008 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
19009 offset_int::from instead of wide_int::to_shwi.
19010 (maybe_diag_overlap): Remove assertion.
19011 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
19012 * gimple-ssa-sprintf.c (format_directive): Same.
19013 (parse_directive): Same.
19014 (sprintf_dom_walker::compute_format_length): Same.
19015 (try_substitute_return_value): Same.
19017 2018-01-03 Jeff Law <law@redhat.com>
19019 PR middle-end/83654
19020 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
19021 non-constant residual for zero at runtime and avoid probing in
19022 that case. Reorganize code for trailing problem to mirror handling
19025 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19027 PR tree-optimization/83501
19028 * tree-ssa-strlen.c (get_string_cst): New.
19029 (handle_char_store): Call get_string_cst.
19031 2018-01-03 Martin Liska <mliska@suse.cz>
19033 PR tree-optimization/83593
19034 * tree-ssa-strlen.c: Include tree-cfg.h.
19035 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
19036 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
19037 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
19039 (strlen_dom_walker::before_dom_children): Call
19040 gimple_purge_dead_eh_edges. Dump tranformation with details
19042 (strlen_dom_walker::before_dom_children): Update call by adding
19043 new argument cleanup_eh.
19044 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
19046 2018-01-03 Martin Liska <mliska@suse.cz>
19049 * cif-code.def (VARIADIC_THUNK): New enum value.
19050 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
19053 2018-01-03 Jan Beulich <jbeulich@suse.com>
19055 * sse.md (mov<mode>_internal): Tighten condition for when to use
19056 vmovdqu<ssescalarsize> for TI and OI modes.
19058 2018-01-03 Jakub Jelinek <jakub@redhat.com>
19060 Update copyright years.
19062 2018-01-03 Martin Liska <mliska@suse.cz>
19065 * ipa-visibility.c (function_and_variable_visibility): Skip
19066 functions with noipa attribure.
19068 2018-01-03 Jakub Jelinek <jakub@redhat.com>
19070 * gcc.c (process_command): Update copyright notice dates.
19071 * gcov-dump.c (print_version): Ditto.
19072 * gcov.c (print_version): Ditto.
19073 * gcov-tool.c (print_version): Ditto.
19074 * gengtype.c (create_file): Ditto.
19075 * doc/cpp.texi: Bump @copying's copyright year.
19076 * doc/cppinternals.texi: Ditto.
19077 * doc/gcc.texi: Ditto.
19078 * doc/gccint.texi: Ditto.
19079 * doc/gcov.texi: Ditto.
19080 * doc/install.texi: Ditto.
19081 * doc/invoke.texi: Ditto.
19083 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19085 * vector-builder.h (vector_builder::m_full_nelts): Change from
19086 unsigned int to poly_uint64.
19087 (vector_builder::full_nelts): Update prototype accordingly.
19088 (vector_builder::new_vector): Likewise.
19089 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
19090 (vector_builder::operator ==): Likewise.
19091 (vector_builder::finalize): Likewise.
19092 * int-vector-builder.h (int_vector_builder::int_vector_builder):
19093 Take the number of elements as a poly_uint64 rather than an
19095 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
19096 from unsigned int to poly_uint64.
19097 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
19098 (vec_perm_indices::new_vector): Likewise.
19099 (vec_perm_indices::length): Likewise.
19100 (vec_perm_indices::nelts_per_input): Likewise.
19101 (vec_perm_indices::input_nelts): Likewise.
19102 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
19103 number of elements per input as a poly_uint64 rather than an
19104 unsigned int. Use the original encoding for variable-length
19105 vectors, rather than clamping each individual element.
19106 For the second and subsequent elements in each pattern,
19107 clamp the step and base before clamping their sum.
19108 (vec_perm_indices::series_p): Handle polynomial element counts.
19109 (vec_perm_indices::all_in_range_p): Likewise.
19110 (vec_perm_indices_to_tree): Likewise.
19111 (vec_perm_indices_to_rtx): Likewise.
19112 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
19113 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
19114 (tree_vector_builder::new_binary_operation): Handle polynomial
19115 element counts. Return false if we need to know the number
19116 of elements at compile time.
19117 * fold-const.c (fold_vec_perm): Punt if the number of elements
19118 isn't known at compile time.
19120 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19122 * vec-perm-indices.h (vec_perm_builder): Change element type
19123 from HOST_WIDE_INT to poly_int64.
19124 (vec_perm_indices::element_type): Update accordingly.
19125 (vec_perm_indices::clamp): Handle polynomial element_types.
19126 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
19127 (vec_perm_indices::all_in_range_p): Likewise.
19128 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
19130 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
19131 polynomial vec_perm_indices element types.
19132 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
19133 * fold-const.c (fold_vec_perm): Likewise.
19134 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
19135 * tree-vect-generic.c (lower_vec_perm): Likewise.
19136 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
19137 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
19138 element type to HOST_WIDE_INT.
19140 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19141 Alan Hayward <alan.hayward@arm.com>
19142 David Sherwood <david.sherwood@arm.com>
19144 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
19145 rather than an int. Use plus_constant.
19146 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
19147 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
19149 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19150 Alan Hayward <alan.hayward@arm.com>
19151 David Sherwood <david.sherwood@arm.com>
19153 * calls.c (emit_call_1, expand_call): Change struct_value_size from
19154 a HOST_WIDE_INT to a poly_int64.
19156 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19157 Alan Hayward <alan.hayward@arm.com>
19158 David Sherwood <david.sherwood@arm.com>
19160 * calls.c (load_register_parameters): Cope with polynomial
19161 mode sizes. Require a constant size for BLKmode parameters
19162 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
19163 forces a parameter to be padded at the lsb end in order to
19164 fill a complete number of words, require the parameter size
19165 to be ordered wrt UNITS_PER_WORD.
19167 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19168 Alan Hayward <alan.hayward@arm.com>
19169 David Sherwood <david.sherwood@arm.com>
19171 * reload1.c (spill_stack_slot_width): Change element type
19172 from unsigned int to poly_uint64_pod.
19173 (alter_reg): Treat mode sizes as polynomial.
19175 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19176 Alan Hayward <alan.hayward@arm.com>
19177 David Sherwood <david.sherwood@arm.com>
19179 * reload.c (complex_word_subreg_p): New function.
19180 (reload_inner_reg_of_subreg, push_reload): Use it.
19182 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19183 Alan Hayward <alan.hayward@arm.com>
19184 David Sherwood <david.sherwood@arm.com>
19186 * lra-constraints.c (process_alt_operands): Reject matched
19187 operands whose sizes aren't ordered.
19188 (match_reload): Refer to this check here.
19190 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19191 Alan Hayward <alan.hayward@arm.com>
19192 David Sherwood <david.sherwood@arm.com>
19194 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
19195 that the mode size is in the set {1, 2, 4, 8, 16}.
19197 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19198 Alan Hayward <alan.hayward@arm.com>
19199 David Sherwood <david.sherwood@arm.com>
19201 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
19202 Use plus_constant instead of gen_rtx_PLUS.
19204 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19205 Alan Hayward <alan.hayward@arm.com>
19206 David Sherwood <david.sherwood@arm.com>
19208 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
19209 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
19210 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
19211 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
19212 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
19213 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
19214 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
19215 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
19216 * config/i386/i386.c (ix86_push_rounding): ...this new function.
19217 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
19219 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
19220 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
19221 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
19222 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
19223 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
19224 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
19225 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
19226 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
19227 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
19228 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
19230 * expr.c (emit_move_resolve_push): Treat the input and result
19231 of PUSH_ROUNDING as a poly_int64.
19232 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
19233 (emit_push_insn): Likewise.
19234 * lra-eliminations.c (mark_not_eliminable): Likewise.
19235 * recog.c (push_operand): Likewise.
19236 * reload1.c (elimination_effects): Likewise.
19237 * rtlanal.c (nonzero_bits1): Likewise.
19238 * calls.c (store_one_arg): Likewise. Require the padding to be
19239 known at compile time.
19241 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19242 Alan Hayward <alan.hayward@arm.com>
19243 David Sherwood <david.sherwood@arm.com>
19245 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
19246 Use plus_constant instead of gen_rtx_PLUS.
19248 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19249 Alan Hayward <alan.hayward@arm.com>
19250 David Sherwood <david.sherwood@arm.com>
19252 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
19253 rather than an int.
19255 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19256 Alan Hayward <alan.hayward@arm.com>
19257 David Sherwood <david.sherwood@arm.com>
19259 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
19260 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
19261 via stack temporaries. Treat the mode size as polynomial too.
19263 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19264 Alan Hayward <alan.hayward@arm.com>
19265 David Sherwood <david.sherwood@arm.com>
19267 * expr.c (expand_expr_real_2): When handling conversions involving
19268 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
19269 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
19270 as a poly_uint64 too.
19272 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19273 Alan Hayward <alan.hayward@arm.com>
19274 David Sherwood <david.sherwood@arm.com>
19276 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
19278 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19279 Alan Hayward <alan.hayward@arm.com>
19280 David Sherwood <david.sherwood@arm.com>
19282 * combine.c (can_change_dest_mode): Handle polynomial
19283 REGMODE_NATURAL_SIZE.
19284 * expmed.c (store_bit_field_1): Likewise.
19285 * expr.c (store_constructor): Likewise.
19286 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
19287 and polynomial REGMODE_NATURAL_SIZE.
19288 (gen_lowpart_common): Likewise.
19289 * reginfo.c (record_subregs_of_mode): Likewise.
19290 * rtlanal.c (read_modify_subreg_p): Likewise.
19292 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19293 Alan Hayward <alan.hayward@arm.com>
19294 David Sherwood <david.sherwood@arm.com>
19296 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
19297 numbers of elements.
19299 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19300 Alan Hayward <alan.hayward@arm.com>
19301 David Sherwood <david.sherwood@arm.com>
19303 * match.pd: Cope with polynomial numbers of vector elements.
19305 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19306 Alan Hayward <alan.hayward@arm.com>
19307 David Sherwood <david.sherwood@arm.com>
19309 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
19310 in a POINTER_PLUS_EXPR.
19312 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19313 Alan Hayward <alan.hayward@arm.com>
19314 David Sherwood <david.sherwood@arm.com>
19316 * omp-simd-clone.c (simd_clone_subparts): New function.
19317 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
19318 (ipa_simd_modify_function_body): Likewise.
19320 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19321 Alan Hayward <alan.hayward@arm.com>
19322 David Sherwood <david.sherwood@arm.com>
19324 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
19325 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
19326 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
19327 (expand_vector_condition, vector_element): Likewise.
19328 (subparts_gt): New function.
19329 (get_compute_type): Use subparts_gt.
19330 (count_type_subparts): Delete.
19331 (expand_vector_operations_1): Use subparts_gt instead of
19332 count_type_subparts.
19334 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19335 Alan Hayward <alan.hayward@arm.com>
19336 David Sherwood <david.sherwood@arm.com>
19338 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
19339 (vect_compile_time_alias): ...this new function. Do the calculation
19340 on poly_ints rather than trees.
19341 (vect_prune_runtime_alias_test_list): Update call accordingly.
19343 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19344 Alan Hayward <alan.hayward@arm.com>
19345 David Sherwood <david.sherwood@arm.com>
19347 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
19349 (vect_schedule_slp_instance): Likewise.
19351 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19352 Alan Hayward <alan.hayward@arm.com>
19353 David Sherwood <david.sherwood@arm.com>
19355 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
19356 constant and extern definitions for variable-length vectors.
19357 (vect_get_constant_vectors): Note that the number of units
19358 is known to be constant.
19360 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19361 Alan Hayward <alan.hayward@arm.com>
19362 David Sherwood <david.sherwood@arm.com>
19364 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
19365 of units as polynomial. Choose between WIDE and NARROW based
19368 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19369 Alan Hayward <alan.hayward@arm.com>
19370 David Sherwood <david.sherwood@arm.com>
19372 * tree-vect-stmts.c (simd_clone_subparts): New function.
19373 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
19375 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19376 Alan Hayward <alan.hayward@arm.com>
19377 David Sherwood <david.sherwood@arm.com>
19379 * tree-vect-stmts.c (vectorizable_call): Treat the number of
19380 vectors as polynomial. Use build_index_vector for
19381 IFN_GOMP_SIMD_LANE.
19383 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19384 Alan Hayward <alan.hayward@arm.com>
19385 David Sherwood <david.sherwood@arm.com>
19387 * tree-vect-stmts.c (get_load_store_type): Treat the number of
19388 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
19389 for variable-length vectors.
19390 (vectorizable_mask_load_store): Treat the number of units as
19391 polynomial, asserting that it is constant if the condition has
19392 already been enforced.
19393 (vectorizable_store, vectorizable_load): Likewise.
19395 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19396 Alan Hayward <alan.hayward@arm.com>
19397 David Sherwood <david.sherwood@arm.com>
19399 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
19400 of units as polynomial. Punt if we can't tell at compile time
19401 which vector contains the final result.
19403 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19404 Alan Hayward <alan.hayward@arm.com>
19405 David Sherwood <david.sherwood@arm.com>
19407 * tree-vect-loop.c (vectorizable_induction): Treat the number
19408 of units as polynomial. Punt on SLP inductions. Use an integer
19409 VEC_SERIES_EXPR for variable-length integer reductions. Use a
19410 cast of such a series for variable-length floating-point
19413 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19414 Alan Hayward <alan.hayward@arm.com>
19415 David Sherwood <david.sherwood@arm.com>
19417 * tree.h (build_index_vector): Declare.
19418 * tree.c (build_index_vector): New function.
19419 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
19420 of units as polynomial, forcibly converting it to a constant if
19421 vectorizable_reduction has already enforced the condition.
19422 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
19423 to create a {1,2,3,...} vector.
19424 (vectorizable_reduction): Treat the number of units as polynomial.
19425 Choose vectype_in based on the largest scalar element size rather
19426 than the smallest number of units. Enforce the restrictions
19429 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19430 Alan Hayward <alan.hayward@arm.com>
19431 David Sherwood <david.sherwood@arm.com>
19433 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
19434 number of units as polynomial.
19436 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19437 Alan Hayward <alan.hayward@arm.com>
19438 David Sherwood <david.sherwood@arm.com>
19440 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
19441 * target.def (autovectorize_vector_sizes): Return the vector sizes
19442 by pointer, using vector_sizes rather than a bitmask.
19443 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
19444 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
19445 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
19447 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
19448 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
19449 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
19450 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
19451 * omp-general.c (omp_max_vf): Likewise.
19452 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19453 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
19454 * tree-vect-loop.c (vect_analyze_loop): Likewise.
19455 * tree-vect-slp.c (vect_slp_bb): Likewise.
19456 * doc/tm.texi: Regenerate.
19457 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
19459 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
19460 the vector size as a poly_uint64 rather than an unsigned int.
19461 (current_vector_size): Change from an unsigned int to a poly_uint64.
19462 (get_vectype_for_scalar_type): Update accordingly.
19463 * tree.h (build_truth_vector_type): Take the size and number of
19464 units as a poly_uint64 rather than an unsigned int.
19465 (build_vector_type): Add a temporary overload that takes
19466 the number of units as a poly_uint64 rather than an unsigned int.
19467 * tree.c (make_vector_type): Likewise.
19468 (build_truth_vector_type): Take the number of units as a poly_uint64
19469 rather than an unsigned int.
19471 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19472 Alan Hayward <alan.hayward@arm.com>
19473 David Sherwood <david.sherwood@arm.com>
19475 * target.def (get_mask_mode): Take the number of units and length
19476 as poly_uint64s rather than unsigned ints.
19477 * targhooks.h (default_get_mask_mode): Update accordingly.
19478 * targhooks.c (default_get_mask_mode): Likewise.
19479 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
19480 * doc/tm.texi: Regenerate.
19482 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19483 Alan Hayward <alan.hayward@arm.com>
19484 David Sherwood <david.sherwood@arm.com>
19486 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
19487 * omp-general.c (omp_max_vf): Likewise.
19488 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
19489 (expand_omp_simd): Handle polynomial safelen.
19490 * omp-low.c (omplow_simd_context): Add a default constructor.
19491 (omplow_simd_context::max_vf): Change from int to poly_uint64.
19492 (lower_rec_simd_input_clauses): Update accordingly.
19493 (lower_rec_input_clauses): Likewise.
19495 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19496 Alan Hayward <alan.hayward@arm.com>
19497 David Sherwood <david.sherwood@arm.com>
19499 * tree-vectorizer.h (vect_nunits_for_cost): New function.
19500 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
19501 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
19502 (vect_analyze_slp_cost): Likewise.
19503 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
19504 (vect_model_load_cost): Likewise.
19506 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19507 Alan Hayward <alan.hayward@arm.com>
19508 David Sherwood <david.sherwood@arm.com>
19510 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
19511 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
19512 from an unsigned int * to a poly_uint64_pod *.
19513 (calculate_unrolling_factor): New function.
19514 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
19516 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19517 Alan Hayward <alan.hayward@arm.com>
19518 David Sherwood <david.sherwood@arm.com>
19520 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
19521 from an unsigned int to a poly_uint64.
19522 (_loop_vec_info::slp_unrolling_factor): Likewise.
19523 (_loop_vec_info::vectorization_factor): Change from an int
19525 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
19526 (vect_get_num_vectors): New function.
19527 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
19528 (vect_get_num_copies): Use vect_get_num_vectors.
19529 (vect_analyze_data_ref_dependences): Change max_vf from an int *
19530 to an unsigned int *.
19531 (vect_analyze_data_refs): Change min_vf from an int * to a
19533 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
19534 than an unsigned HOST_WIDE_INT.
19535 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
19536 (vect_analyze_data_ref_dependence): Change max_vf from an int *
19537 to an unsigned int *.
19538 (vect_analyze_data_ref_dependences): Likewise.
19539 (vect_compute_data_ref_alignment): Handle polynomial vf.
19540 (vect_enhance_data_refs_alignment): Likewise.
19541 (vect_prune_runtime_alias_test_list): Likewise.
19542 (vect_shift_permute_load_chain): Likewise.
19543 (vect_supportable_dr_alignment): Likewise.
19544 (dependence_distance_ge_vf): Take the vectorization factor as a
19545 poly_uint64 rather than an unsigned HOST_WIDE_INT.
19546 (vect_analyze_data_refs): Change min_vf from an int * to a
19548 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
19549 vfm1 as a poly_uint64 rather than an int. Make the same change
19550 for the returned bound_scalar.
19551 (vect_gen_vector_loop_niters): Handle polynomial vf.
19552 (vect_do_peeling): Likewise. Update call to
19553 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
19554 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
19556 * tree-vect-loop.c (vect_determine_vectorization_factor)
19557 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
19558 (vect_get_known_peeling_cost): Likewise.
19559 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
19560 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
19561 (vect_transform_loop): Likewise. Use the lowest possible VF when
19562 updating the upper bounds of the loop.
19563 (vect_min_worthwhile_factor): Make static. Return an unsigned int
19564 rather than an int.
19565 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
19566 polynomial unroll factors.
19567 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
19568 (vect_make_slp_decision): Likewise.
19569 (vect_supported_load_permutation_p): Likewise, and polynomial
19571 (vect_analyze_slp_cost): Handle polynomial vf.
19572 (vect_slp_analyze_node_operations): Likewise.
19573 (vect_slp_analyze_bb_1): Likewise.
19574 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
19575 than an unsigned HOST_WIDE_INT.
19576 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
19577 (vectorizable_load): Handle polynomial vf.
19578 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
19580 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
19582 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19583 Alan Hayward <alan.hayward@arm.com>
19584 David Sherwood <david.sherwood@arm.com>
19586 * match.pd: Handle bit operations involving three constants
19587 and try to fold one pair.
19589 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19591 * tree-vect-loop-manip.c: Include gimple-fold.h.
19592 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
19593 niters_maybe_zero parameters. Handle other cases besides a step of 1.
19594 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
19595 Add a path that uses a step of VF instead of 1, but disable it
19597 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
19598 and niters_no_overflow parameters. Update calls to
19599 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
19600 Create a new SSA name if the latter choses to use a ste other
19601 than zero, and return it via niters_vector_mult_vf_var.
19602 * tree-vect-loop.c (vect_transform_loop): Update calls to
19603 vect_do_peeling, vect_gen_vector_loop_niters and
19604 slpeel_make_loop_iterate_ntimes.
19605 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
19606 (vect_gen_vector_loop_niters): Update declarations after above changes.
19608 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
19610 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
19611 128-bit round to integer instructions.
19612 (ceil<mode>2): Likewise.
19613 (btrunc<mode>2): Likewise.
19614 (round<mode>2): Likewise.
19616 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19618 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
19619 unaligned VSX load/store on P8/P9.
19620 (expand_block_clear): Allow the use of unaligned VSX
19621 load/store on P8/P9.
19623 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19625 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
19627 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
19628 swap associated with both a load and a store.
19630 2018-01-02 Andrew Waterman <andrew@sifive.com>
19632 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
19633 * config/riscv/riscv.md (clear_cache): Use it.
19635 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
19637 * web.c: Remove out-of-date comment.
19639 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19641 * expr.c (fixup_args_size_notes): Check that any existing
19642 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
19643 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
19644 (emit_single_push_insn): ...here.
19646 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19648 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
19649 (const_vector_encoded_nelts): New function.
19650 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
19651 (const_vector_int_elt, const_vector_elt): Declare.
19652 * emit-rtl.c (const_vector_int_elt_1): New function.
19653 (const_vector_elt): Likewise.
19654 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
19655 of CONST_VECTOR_ELT.
19657 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19659 * expr.c: Include rtx-vector-builder.h.
19660 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
19661 directly on the tree encoding.
19662 (const_vector_from_tree): Likewise.
19663 * optabs.c: Include rtx-vector-builder.h.
19664 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
19665 sequence of "u" values.
19666 * vec-perm-indices.c: Include rtx-vector-builder.h.
19667 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
19668 directly on the vec_perm_indices encoding.
19670 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19672 * doc/rtl.texi (const_vector): Describe new encoding scheme.
19673 * Makefile.in (OBJS): Add rtx-vector-builder.o.
19674 * rtx-vector-builder.h: New file.
19675 * rtx-vector-builder.c: Likewise.
19676 * rtl.h (rtx_def::u2): Add a const_vector field.
19677 (CONST_VECTOR_NPATTERNS): New macro.
19678 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
19679 (CONST_VECTOR_DUPLICATE_P): Likewise.
19680 (CONST_VECTOR_STEPPED_P): Likewise.
19681 (CONST_VECTOR_ENCODED_ELT): Likewise.
19682 (const_vec_duplicate_p): Check for a duplicated vector encoding.
19683 (unwrap_const_vec_duplicate): Likewise.
19684 (const_vec_series_p): Check for a non-duplicated vector encoding.
19685 Say that the function only returns true for integer vectors.
19686 * emit-rtl.c: Include rtx-vector-builder.h.
19687 (gen_const_vec_duplicate_1): Delete.
19688 (gen_const_vector): Call gen_const_vec_duplicate instead of
19689 gen_const_vec_duplicate_1.
19690 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
19691 (gen_const_vec_duplicate): Use rtx_vector_builder.
19692 (gen_const_vec_series): Likewise.
19693 (gen_rtx_CONST_VECTOR): Likewise.
19694 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
19695 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
19696 Build a new vector rather than modifying a CONST_VECTOR in-place.
19697 (handle_special_swappables): Update call accordingly.
19698 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
19699 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
19700 Build a new vector rather than modifying a CONST_VECTOR in-place.
19701 (handle_special_swappables): Update call accordingly.
19703 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19705 * simplify-rtx.c (simplify_const_binary_operation): Use
19706 CONST_VECTOR_ELT instead of XVECEXP.
19708 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19710 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
19711 the selector elements to be different from the data elements
19712 if the selector is a VECTOR_CST.
19713 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
19714 ssizetype for the selector.
19716 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19718 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
19719 before testing each element individually.
19720 * tree-vect-generic.c (lower_vec_perm): Likewise.
19722 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19724 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
19725 * selftest-run-tests.c (selftest::run_tests): Call it.
19726 * vector-builder.h (vector_builder::operator ==): New function.
19727 (vector_builder::operator !=): Likewise.
19728 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
19729 (vec_perm_indices::all_from_input_p): New function.
19730 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
19731 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
19732 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
19733 instead of reading the VECTOR_CST directly. Detect whether both
19734 vector inputs are the same before constructing the vec_perm_indices,
19735 and update the number of inputs argument accordingly. Use the
19736 utility functions added above. Only construct sel2 if we need to.
19738 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19740 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
19741 the broadcast of the low byte.
19742 (expand_mult_highpart): Use an explicit encoding for the permutes.
19743 * optabs-query.c (can_mult_highpart_p): Likewise.
19744 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
19745 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
19746 (vectorizable_bswap): Likewise.
19747 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
19748 explicit encoding for the power-of-2 permutes.
19749 (vect_permute_store_chain): Likewise.
19750 (vect_grouped_load_supported): Likewise.
19751 (vect_permute_load_chain): Likewise.
19753 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19755 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
19756 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
19757 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
19758 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
19759 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
19760 (vect_gen_perm_mask_any): Likewise.
19762 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19764 * int-vector-builder.h: New file.
19765 * vec-perm-indices.h: Include int-vector-builder.h.
19766 (vec_perm_indices): Redefine as an int_vector_builder.
19767 (auto_vec_perm_indices): Delete.
19768 (vec_perm_builder): Redefine as a stand-alone class.
19769 (vec_perm_indices::vec_perm_indices): New function.
19770 (vec_perm_indices::clamp): Likewise.
19771 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
19772 (vec_perm_indices::new_vector): New function.
19773 (vec_perm_indices::new_expanded_vector): Update for new
19774 vec_perm_indices class.
19775 (vec_perm_indices::rotate_inputs): New function.
19776 (vec_perm_indices::all_in_range_p): Operate directly on the
19777 encoded form, without computing elided elements.
19778 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
19779 encoding. Update for new vec_perm_indices class.
19780 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
19781 the given vec_perm_builder.
19782 (expand_vec_perm_var): Update vec_perm_builder constructor.
19783 (expand_mult_highpart): Use vec_perm_builder instead of
19784 auto_vec_perm_indices.
19785 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
19786 vec_perm_indices instead of auto_vec_perm_indices. Use a single
19787 or double series encoding as appropriate.
19788 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
19789 vec_perm_indices instead of auto_vec_perm_indices.
19790 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
19791 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
19792 (vect_permute_store_chain): Likewise.
19793 (vect_grouped_load_supported): Likewise.
19794 (vect_permute_load_chain): Likewise.
19795 (vect_shift_permute_load_chain): Likewise.
19796 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
19797 (vect_transform_slp_perm_load): Likewise.
19798 (vect_schedule_slp_instance): Likewise.
19799 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
19800 (vectorizable_mask_load_store): Likewise.
19801 (vectorizable_bswap): Likewise.
19802 (vectorizable_store): Likewise.
19803 (vectorizable_load): Likewise.
19804 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
19805 vec_perm_indices instead of auto_vec_perm_indices. Use
19806 tree_to_vec_perm_builder to read the vector from a tree.
19807 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
19808 vec_perm_builder instead of a vec_perm_indices.
19809 (have_whole_vector_shift): Use vec_perm_builder and
19810 vec_perm_indices instead of auto_vec_perm_indices. Leave the
19811 truncation to calc_vec_perm_mask_for_shift.
19812 (vect_create_epilog_for_reduction): Likewise.
19813 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
19814 from auto_vec_perm_indices to vec_perm_indices.
19815 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
19816 instead of changing individual elements.
19817 (aarch64_vectorize_vec_perm_const): Use new_vector to install
19818 the vector in d.perm.
19819 * config/arm/arm.c (expand_vec_perm_d::perm): Change
19820 from auto_vec_perm_indices to vec_perm_indices.
19821 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
19822 instead of changing individual elements.
19823 (arm_vectorize_vec_perm_const): Use new_vector to install
19824 the vector in d.perm.
19825 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
19826 Update vec_perm_builder constructor.
19827 (rs6000_expand_interleave): Likewise.
19828 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
19829 (rs6000_expand_interleave): Likewise.
19831 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19833 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
19834 to qimode could truncate the indices.
19835 * optabs.c (expand_vec_perm_var): Likewise.
19837 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19839 * Makefile.in (OBJS): Add vec-perm-indices.o.
19840 * vec-perm-indices.h: New file.
19841 * vec-perm-indices.c: Likewise.
19842 * target.h (vec_perm_indices): Replace with a forward class
19844 (auto_vec_perm_indices): Move to vec-perm-indices.h.
19845 * optabs.h: Include vec-perm-indices.h.
19846 (expand_vec_perm): Delete.
19847 (selector_fits_mode_p, expand_vec_perm_var): Declare.
19848 (expand_vec_perm_const): Declare.
19849 * target.def (vec_perm_const_ok): Replace with...
19850 (vec_perm_const): ...this new hook.
19851 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
19852 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
19853 * doc/tm.texi: Regenerate.
19854 * optabs.def (vec_perm_const): Delete.
19855 * doc/md.texi (vec_perm_const): Likewise.
19856 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
19857 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
19858 expand_vec_perm for constant permutation vectors. Assert that
19859 the mode of variable permutation vectors is the integer equivalent
19860 of the mode that is being permuted.
19861 * optabs-query.h (selector_fits_mode_p): Declare.
19862 * optabs-query.c: Include vec-perm-indices.h.
19863 (selector_fits_mode_p): New function.
19864 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
19865 is defined, instead of checking whether the vec_perm_const_optab
19866 exists. Use targetm.vectorize.vec_perm_const instead of
19867 targetm.vectorize.vec_perm_const_ok. Check whether the indices
19868 fit in the vector mode before using a variable permute.
19869 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
19870 vec_perm_indices instead of an rtx.
19871 (expand_vec_perm): Replace with...
19872 (expand_vec_perm_const): ...this new function. Take the selector
19873 as a vec_perm_indices rather than an rtx. Also take the mode of
19874 the selector. Update call to shift_amt_for_vec_perm_mask.
19875 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
19876 Use vec_perm_indices::new_expanded_vector to expand the original
19877 selector into bytes. Check whether the indices fit in the vector
19878 mode before using a variable permute.
19879 (expand_vec_perm_var): Make global.
19880 (expand_mult_highpart): Use expand_vec_perm_const.
19881 * fold-const.c: Includes vec-perm-indices.h.
19882 * tree-ssa-forwprop.c: Likewise.
19883 * tree-vect-data-refs.c: Likewise.
19884 * tree-vect-generic.c: Likewise.
19885 * tree-vect-loop.c: Likewise.
19886 * tree-vect-slp.c: Likewise.
19887 * tree-vect-stmts.c: Likewise.
19888 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
19890 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
19891 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
19892 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
19893 (aarch64_vectorize_vec_perm_const): ...this new function.
19894 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
19895 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
19896 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
19897 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
19898 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
19899 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
19900 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
19902 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
19903 check for NEON modes.
19904 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
19905 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
19906 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
19907 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
19909 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
19910 the old VEC_PERM_CONST conditions.
19911 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
19912 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
19913 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
19914 (ia64_vectorize_vec_perm_const_ok): Merge into...
19915 (ia64_vectorize_vec_perm_const): ...this new function.
19916 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
19917 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
19918 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
19919 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
19920 * config/mips/mips.c (mips_expand_vec_perm_const)
19921 (mips_vectorize_vec_perm_const_ok): Merge into...
19922 (mips_vectorize_vec_perm_const): ...this new function.
19923 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
19924 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
19925 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
19926 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
19927 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
19928 (rs6000_expand_vec_perm_const): Delete.
19929 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
19931 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
19932 (altivec_expand_vec_perm_const_le): Take each operand individually.
19933 Operate on constant selectors rather than rtxes.
19934 (altivec_expand_vec_perm_const): Likewise. Update call to
19935 altivec_expand_vec_perm_const_le.
19936 (rs6000_expand_vec_perm_const): Delete.
19937 (rs6000_vectorize_vec_perm_const_ok): Delete.
19938 (rs6000_vectorize_vec_perm_const): New function.
19939 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
19940 an element count and rtx array.
19941 (rs6000_expand_extract_even): Update call accordingly.
19942 (rs6000_expand_interleave): Likewise.
19943 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
19944 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
19945 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
19946 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
19947 (rs6000_expand_vec_perm_const): Delete.
19948 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
19949 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
19950 (altivec_expand_vec_perm_const_le): Take each operand individually.
19951 Operate on constant selectors rather than rtxes.
19952 (altivec_expand_vec_perm_const): Likewise. Update call to
19953 altivec_expand_vec_perm_const_le.
19954 (rs6000_expand_vec_perm_const): Delete.
19955 (rs6000_vectorize_vec_perm_const_ok): Delete.
19956 (rs6000_vectorize_vec_perm_const): New function. Remove stray
19957 reference to the SPE evmerge intructions.
19958 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
19959 an element count and rtx array.
19960 (rs6000_expand_extract_even): Update call accordingly.
19961 (rs6000_expand_interleave): Likewise.
19962 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
19963 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
19965 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
19967 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19969 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
19970 vector mode and that that mode matches the mode of the data
19972 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
19973 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
19974 directly using expand_vec_perm_1 when forcing selectors into
19976 (expand_vec_perm_var): New function, split out from expand_vec_perm.
19978 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19980 * optabs-query.h (can_vec_perm_p): Delete.
19981 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
19982 * optabs-query.c (can_vec_perm_p): Split into...
19983 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
19984 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
19985 particular selector is valid.
19986 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
19987 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
19988 (vect_grouped_load_supported): Likewise.
19989 (vect_shift_permute_load_chain): Likewise.
19990 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
19991 (vect_transform_slp_perm_load): Likewise.
19992 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
19993 (vectorizable_bswap): Likewise.
19994 (vect_gen_perm_mask_checked): Likewise.
19995 * fold-const.c (fold_ternary_loc): Likewise. Don't take
19996 implementations of variable permutation vectors into account
19997 when deciding which selector to use.
19998 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
19999 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
20000 with a false third argument.
20001 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
20002 to test whether the constant selector is valid and can_vec_perm_var_p
20003 to test whether a variable selector is valid.
20005 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
20007 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
20008 * optabs-query.c (can_vec_perm_p): Likewise.
20009 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
20010 instead of vec_perm_indices.
20011 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
20012 (vect_gen_perm_mask_checked): Likewise,
20013 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
20014 (vect_gen_perm_mask_checked): Likewise,
20016 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
20018 * optabs-query.h (qimode_for_vec_perm): Declare.
20019 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
20020 (qimode_for_vec_perm): ...this new function.
20021 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
20023 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20025 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
20026 does not have a conditional at the top.
20028 2018-01-02 Richard Biener <rguenther@suse.de>
20030 * ipa-inline.c (big_speedup_p): Fix expression.
20032 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
20035 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
20038 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
20042 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
20043 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
20044 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
20045 cond_taken_branch_cost 3->4.
20047 2018-01-01 Jakub Jelinek <jakub@redhat.com>
20049 PR tree-optimization/83581
20050 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
20051 TODO_cleanup_cfg if any changes have been made.
20053 PR middle-end/83608
20054 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
20055 convert_modes if target mode has the right side, but different mode
20058 PR middle-end/83609
20059 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
20060 last argument when extracting from CONCAT. If either from_real or
20061 from_imag is NULL, use expansion through memory. If result is not
20062 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
20063 the parts directly to inner mode, if even that fails, use expansion
20066 PR middle-end/83623
20067 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
20068 check for bswap in mode rather than HImode and use that in expand_unop
20071 Copyright (C) 2018 Free Software Foundation, Inc.
20073 Copying and distribution of this file, with or without modification,
20074 are permitted in any medium without royalty provided the copyright
20075 notice and this notice are preserved.