1 2018-07-02 Jeff Law <law@redhat.com>
3 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
4 the H8/300, H8/300H and H8/S variants into a single pattern.
5 (movhi_h8300, movqi_h8300hs): Similarly.
6 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
7 (QHI mode iterator): New.
9 * config/h8300/h8300.md: Remove trailing whitespace.
11 2018-07-02 Jim Wilson <jimw@sifive.com>
13 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
14 instead of emit_insn for interrupt returns.
15 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
16 (riscv_sret, riscv_uret): Likewise.
18 2018-07-02 David Malcolm <dmalcolm@redhat.com>
20 * pretty-print.c (selftest::test_pp_format): Move save and restore
21 of quotes to class auto_fix_quotes, and add an instance.
22 * selftest.c: Include "intl.h".
23 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
24 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
25 * selftest.h (selftest::auto_fix_quotes): New class.
27 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
29 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
30 (aarch64_sve_prepare_conditional_op): Remove.
31 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
32 Allow aarch64_simd_reg_or_zero as select operand; remove
33 the aarch64_sve_prepare_conditional_op call.
34 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
35 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
36 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
37 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
38 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
39 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
40 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
41 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
42 and a splitters to match all of the *_any patterns.
43 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
45 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
46 (SVE_COND_FP_BINARY_REV): Remove.
47 (sve_int_op_rev, sve_fp_op_rev): New.
48 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
49 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
50 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
51 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
52 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
53 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
54 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
55 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
56 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
58 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
59 Remove match_dup 1 from the inner unspec.
60 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
62 * config/aarch64/aarch64.md (movprfx): New attr.
63 (length): Default movprfx to 8.
64 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
65 (*madd<SVE_I>, *msub<SVE_I): Likewise.
66 (*<su>mul<SVE_I>3_highpart): Likewise.
67 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
68 (*v<ASHIFT><SVE_I>3): Likewise.
69 (*<su><MAXMIN><SVE_I>3): Likewise.
70 (*<su><MAXMIN><SVE_F>3): Likewise.
71 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
72 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
73 (*div<SVE_F>4): Likewise.
75 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
77 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
80 2018-07-02 Richard Biener <rguenther@suse.de>
82 PR tree-optimization/86363
83 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
84 memset argument refers to a non-variable address.
86 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
88 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
89 {PLUS,MINUS}_EXPR code to...
90 (adjust_symbolic_bound): ...here,
91 (combine_bound): ...here,
92 (set_value_range_with_overflow): ...and here.
94 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
96 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
98 (extract_range_from_abs_expr): ...here.
100 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
102 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
103 -fno-omit-frame-pointer when not optimizing.
105 2018-07-02 Martin Liska <mliska@suse.cz>
108 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
109 (propagate_nothrow): Likewise.
111 2018-07-02 Martin Liska <mliska@suse.cz>
114 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
116 2018-07-02 David Malcolm <dmalcolm@redhat.com>
118 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
119 function in r262149, changing "loc" param from source_location to
120 const dump_location_t &.
121 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
122 declaration, as above.
124 2018-07-01 Paul Koning <ni1d@arrl.net>
126 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
127 -munit-asm, -mgnu-asm, -mdec-asm.
128 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
129 (pdp11_output_labelref): New.
130 (pdp11_output_def): New.
131 (pdp11_output_addr_vec_elt): New.
132 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
133 %# and %@ format codes.
134 (pdp11_option_override): New.
135 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
136 (pdp11_output_ident): New.
137 (pdp11_asm_named_section): New.
138 (pdp11_asm_init_sections): New.
139 (pdp11_file_start): New.
140 (pdp11_file_end): New.
141 (output_ascii): Use .ascii/.asciz for -mdec-asm.
142 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
143 %o, like %c but octal.
144 (pdp11_option_override): New.
145 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
147 (DATA_SECTION_ASM_OP): Ditto.
148 (READONLY_DATA_SECTION_ASM_OP): New.
149 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
150 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
151 (ASM_OUTPUT_LABELREF): Ditto.
152 (ASM_OUTPUT_DEF): Ditto.
153 (ASM_OUTPUT_EXTERNAL): New.
154 (ASM_OUTPUT_SOURCE_FILENAME): New.
155 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
156 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
157 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
158 %# and %@ format codes.
159 * config/pdp11/pdp11.opt (mgnu-asm): New.
160 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
161 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
162 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
164 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
166 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
167 dereferencing path[] beyond its length.
169 (debug_all_paths): New.
170 (rewire_first_differing_edge): New.
171 (adjust_paths_after_duplication): New.
172 (duplicate_thread_path): Call adjust_paths_after_duplication.
174 (thread_through_all_blocks): Add new argument to
175 duplicate_thread_path.
177 2018-06-30 Jim Wilson <jimw@sifive.com>
179 * config/riscv/predicates.md (p2m1_shift_operand): New.
180 (high_mask_shift_operand): New.
181 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
182 pattern using p2m1_shift_operand.
183 (lshsi3_zero_extend_3+2): New combiner pattern using
184 high_mask_shift_operand.
186 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
188 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
190 (vect_recog_rotate_pattern): ...here.
191 (vect_convert_input): Try to insert casts of invariants in the
193 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
194 preheader to be empty.
196 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
198 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
199 vector type. If given, install it in the new statement's
201 (vect_element_precision): New function.
202 (vect_unpromoted_value): New struct.
203 (vect_unpromoted_value::vect_unpromoted_value): New function.
204 (vect_unpromoted_value::set_op): Likewise.
205 (vect_look_through_possible_promotion): Likewise.
206 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
207 (vect_widened_op_tree, vect_convert_input): Likewise.
208 (vect_convert_inputs, vect_convert_output): Likewise.
209 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
210 to handle the optional cast of the multiplication result and
211 vect_widened_op_tree to detect the widened multiplication itself.
212 Do not require the input and output of promotion casts to have
213 the same sign, but base the signedness of the operation on the
214 input rather than the result. If the pattern includes two
215 promotions, check that those promotions have the same sign.
216 Do not restrict the MULT_EXPR handling to a double-width result;
217 handle quadruple-width results and wider. Use vect_convert_inputs
218 to convert the inputs to the common type.
219 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
220 to handle the optional cast of the ABS result. Also allow a sign
221 change or a sign extension between the ABS and MINUS.
222 Use vect_widened_op_tree to detect the widened subtraction and use
223 vect_convert_inputs to convert the inputs to the common type.
224 (vect_handle_widen_op_by_const): Delete.
225 (vect_recog_widen_op_pattern): New function.
226 (vect_recog_widen_mult_pattern): Use it.
227 (vect_recog_widen_shift_pattern): Likewise.
228 (vect_recog_widen_sum_pattern): Use
229 vect_look_through_possible_promotion to handle the promoted
232 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
234 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
235 the containing gimple_seq *.
236 * gimple-iterator.h (gsi_for_stmt): Declare it.
237 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
238 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
239 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
240 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
241 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
243 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
245 (vect_mark_pattern_stmts): ...here. Handle cases in which the
246 statement being replaced is part of an existing pattern
247 definition sequence, inserting the new pattern statements before
249 (vect_pattern_recog_1): Don't return a bool. If the statement
250 is already part of a pattern, instead apply pattern matching
251 to the pattern definition statements. Don't clear the
252 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
253 (vect_pattern_recog): Don't break after the first match;
254 continue processing the pattern definition statements instead.
255 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
257 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
259 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
260 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
261 (vect_recog_widen_sum_pattern): Use it.
263 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
265 * tree-vect-loop.c (vectorizable_reduction): Assert that the
266 phi is not a pattern statement and has not been replaced by
268 * tree-vect-patterns.c (type_conversion_p): Don't check
269 STMT_VINFO_IN_PATTERN_P.
270 (vect_recog_vector_vector_shift_pattern): Likewise.
271 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
272 the pattern statement rather than the original statement; check
273 directly for a WIDEN_MULT_EXPR here.
274 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
275 vect_is_simple_use to return the pattern statement rather
276 than the original statement; use is_pattern_stmt_p to check
277 for such a pattern statement.
278 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
279 to return the pattern statement rather than the original statement;
280 don't do the same transformation here.
281 (vect_is_simple_use): If the defining statement has been replaced
282 by a pattern statement, return the pattern statement instead.
283 Remove the corresponding (local) transformation from the vectype
286 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
288 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
289 end and default to null.
290 * tree-vect-loop.c (vect_create_epilog_for_reduction)
291 (vectorizable_reduction): Update calls accordingly, dropping the
292 gimple ** argument if the passed-back statement isn't needed.
293 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
294 (vect_recog_rotate_pattern): Likewise.
295 (vect_recog_mask_conversion_pattern): Likewise.
296 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
297 (vect_mask_constant_operand_p): Likewise.
298 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
299 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
300 (get_group_load_store_type, get_load_store_type): Likewise.
301 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
302 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
303 (vectorizable_conversion, vectorizable_assignment): Likewise.
304 (vectorizable_shift, vectorizable_operation): Likewise.
305 (vectorizable_store, vect_is_simple_cond): Likewise.
306 (vectorizable_condition, vectorizable_comparison): Likewise.
307 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
308 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
309 and move it to the end. Cope with null def_stmt_outs.
311 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
313 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
315 2018-06-29 Jeff Law <law@redhat.com>
317 * config/v850/v850.c (v850_legitimate_address_p): Handle large
318 displacements for TARGET_V850E2V3 and newer.
319 (TARGET_LRA_P): Remove. Defaults to LRA now.
320 * config/v850/v850.md (sign23byte_load): Remove.
321 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
322 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
324 2018-06-29 Martin Liska <mliska@suse.cz>
327 * coverage.c (coverage_init): Mangle full path name.
328 * doc/invoke.texi: Document the change.
329 * gcov-io.c (mangle_path): New.
330 * gcov-io.h (mangle_path): Likewise.
331 * gcov.c (mangle_name): Use mangle_path for path mangling.
333 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
335 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
336 if starting source register is not even.
338 2018-06-29 Martin Liska <mliska@suse.cz>
340 PR tree-optimization/86263
341 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
342 Make edge redirection.
344 2018-06-29 David Malcolm <dmalcolm@redhat.com>
346 * dumpfile.c (dump_loc): Add indentation based on scope depth.
347 (dump_scope_depth): New variable.
348 (get_dump_scope_depth): New function.
349 (dump_begin_scope): New function.
350 (dump_end_scope): New function.
351 * dumpfile.h (get_dump_scope_depth): New declaration.
352 (dump_begin_scope): New declaration.
353 (dump_end_scope): New declaration.
354 (class auto_dump_scope): New class.
355 (AUTO_DUMP_SCOPE): New macro.
356 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
359 2018-06-29 Richard Biener <rguenther@suse.de>
361 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
362 compute_all_dependences succeeds.
363 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
364 exceed --param loop-max-datarefs-for-datadeps.
366 2018-06-29 Jakub Jelinek <jakub@redhat.com>
368 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
370 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
373 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
374 alternative 0 in preferred_for_speed attribute.
376 2018-06-28 Paul Koning <ni1d@arrl.net>
378 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
379 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
380 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
381 * config/pdp11/pdp11.md: Correct "length" attribute calculation
382 for shift insn patterns.
384 2018-06-28 David Malcolm <dmalcolm@redhat.com>
386 * cgraph.c (cgraph_node::get_body): Replace assignments to
387 "dump_file" with calls to set_dump_file.
388 * dumpfile.c (alt_dump_file): Make static, and group with...
389 (alt_flags): ...this definition.
390 (dumps_are_enabled): New variable.
391 (refresh_dumps_are_enabled): New function.
392 (set_dump_file): New function.
393 (set_alt_dump_file): New function.
394 (gcc::dump_manager::dump_start): Replace assignments to
395 "dump_file" and "alt_dump_file" with calls to set_dump_file and
397 (gcc::dump_manager::dump_finish): Likewise.
398 * dumpfile.h (alt_dump_file): Delete decl.
399 (dumps_are_enabled): New variable decl.
400 (set_dump_file): New function decl.
401 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
403 * tree-nested.c (lower_nested_functions): Replace assignments to
404 "dump_file" with calls to set_dump_file.
406 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
408 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
409 goto_locus of each outgoing edge of each basic block.
411 2018-06-28 Richard Biener <rguenther@suse.de>
413 * dwarf2out.c (decl_scope_table): Remove.
414 (push_decl_scope): Likewise.
415 (pop_decl_scope): Likewise.
416 (gen_type_die_for_member): Do not call push/pop_decl_scope.
417 (gen_struct_or_union_type_die): Likewise.
418 (gen_tagged_type_die): Likewise.
419 (dwarf2out_init): Do not initialize decl_scope_table.
420 (dwarf2out_c_finalize): Do not free it.
422 2018-06-28 Richard Biener <rguenther@suse.de>
424 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
425 deciding whether to not re-use a DIE.
427 2018-06-28 Richard Biener <rguenther@suse.de>
429 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
430 DW_AT_abstract_origin attribute.
432 2018-06-28 Martin Liska <mliska@suse.cz>
434 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
435 Use newly introduced constants.
436 * tree-switch-conversion.h (struct jump_table_cluster):
437 Define max_ratio_for_size and max_ratio_for_speed.
439 2018-06-28 Martin Liska <mliska@suse.cz>
441 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
442 Add new checking assert to catch invalid state.
443 (jump_table_cluster::can_be_handled): Handle single case
445 (jump_table_cluster::is_beneficial): Bail out for such case.
446 (bit_test_cluster::find_bit_tests):
447 Add new checking assert to catch invalid state.
448 (bit_test_cluster::can_be_handled): Handle single case
450 (bit_test_cluster::is_beneficial): Bail out for such case.
451 (switch_decision_tree::analyze_switch_statement):
454 2018-06-28 Martin Liska <mliska@suse.cz>
456 * common.opt: Introduce -completion option.
457 * gcc.c (driver_handle_option): Handle it.
458 (driver::main): Print completions if completion
460 * opt-suggestions.c (option_proposer::get_completions):
462 (option_proposer::suggest_completion): Likewise.
463 (option_proposer::find_param_completions): Likewise.
464 (verify_autocompletions): Likewise.
465 (test_completion_valid_options): Likewise.
466 (test_completion_valid_params): Likewise.
467 (in_completion_p): Likewise.
468 (empty_completion_p): Likewise.
469 (test_completion_partial_match): Likewise.
470 (test_completion_garbage): Likewise.
471 (opt_proposer_c_tests): Likewise.
472 * opt-suggestions.h: Declare new functions.
473 * opts.c (common_handle_option): Handle OPT__completion_.
474 * selftest-run-tests.c (selftest::run_tests): Add
475 opt_proposer_c_tests.
476 * selftest.c (assert_str_startswith): New.
477 * selftest.h (assert_str_startswith): Likewise.
478 (opt_proposer_c_tests): New.
479 (ASSERT_STR_STARTSWITH): Likewise.
481 2018-06-28 Martin Liska <mliska@suse.cz>
483 * Makefile.in: Add opt-suggestions.o.
484 * gcc-main.c: Include opt-suggestions.h.
485 * gcc.c (driver::driver): Likewise.
486 (driver::~driver): Remove m_option_suggestions.
487 (driver::build_option_suggestions): Moved to option_proposer.
488 (driver::suggest_option): Likewise.
489 (driver::handle_unrecognized_options): Use option_proposer.
490 * gcc.h (class driver): Add new memver m_option_proposer.
491 * opt-suggestions.c: New file.
492 * opt-suggestions.h: New file.
494 2018-06-28 Martin Liska <mliska@suse.cz>
496 * vec.h (class auto_string_vec): New (moved from auto_argvec).
497 (auto_string_vec::~auto_string_vec): Likewise.
499 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
501 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
502 prevent_decl_creation_for_types fields up and add reset_location field.
503 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
504 statement if id->reset_location is true.
505 (copy_edges_for_bb): Do not set goto_locus on the new edges if
506 id->reset_location is true.
507 (copy_phis_for_bb): Force input_location on the arguments if
508 id->reset_location is true.
509 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
510 is set on the function to be inlined.
512 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
514 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
516 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
518 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
520 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
521 hard registers for the clobbered pseudo.
523 2018-06-27 Paul Koning <ni1d@arrl.net>
525 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
526 mutually exclusive options.
527 * config/pdp11/constraints.md (h): New constraint.
528 (O): Update definition to match shift code generation.
530 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
532 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
534 (output_jump): Change arguments.
535 (pdp11_fixed_cc_regs): New function.
536 (pdp11_cc_mode): Ditto.
537 (pdp11_expand_shift): Ditto.
538 (pdp11_assemble_shift): Ditto.
539 (pdp11_small_shift): Ditto.
540 (pdp11_branch_cost): Remove.
541 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
543 (pdp11_register_move_cost): Update for CC registers.
544 (pdp11_rtx_costs): Add case for LSHIFTRT.
545 (pdp11_output_jump): Add CCNZ mode conditional branches.
546 (notice_update_cc_on_set): Remove.
547 (pdp11_cc_mode): New function.
548 (simple_memory_operand): Correct pre/post decrement case.
549 (no_side_effect_operand): New function.
550 (pdp11_regno_reg_class): Add CC_REGS class.
551 (pdp11_fixed_cc_regs): New function.
552 (pdp11_small_shift): New function.
553 (pdp11_expand_shift): New function to expand shift insns.
554 (pdp11_assemble_shift): New function to output shifts.
555 (pdp11_branch_cost): Remove.
556 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
557 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
559 (PTRDIFF_TYPE): Ditto.
560 (ADJUST_INSN_LENGTH): New macro.
561 (FIXED_REGISTERS): Add CC registers.
562 (CALL_USED_REGISTERS): Ditto.
564 (REG_CLASS_NAMES): Ditto.
565 (REG_CLASS_CONTENTS): Ditto.
566 (SELECT_CC_MODE): Use new function.
567 (TARGET_FLAGS_REGNUM): New macro.
568 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
569 (cc0_reg_rtx): Remove.
570 (CC_STATUS_MDEP): Remove.
571 (CC_STATUS_MDEFP_INIT): Remove.
573 (NOTICE_UPDATE_CC): Remove.
574 (REGISTER_NAMES): Add CC registers.
575 (BRANCH_COST): Change to constant 1.
576 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
578 * config/pdp11/pdp11.opt (mbcopy): Remove.
579 (mbcopy-builtin): Remove.
580 (mbranch-cheap): Remove.
581 (mbranch-expensive): Remove.
582 * config/pdp11/predicates.md (expand_shift_operand): Update to
583 match shift code generation.
584 (ccnz_operator): New predicate.
585 * doc/invoke.texi (PDP-11 Options): Remove deleted options
586 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
587 Remove non-existent option -mabshi, -mno-abshi. Document mutually
589 * doc/md.texi (PDP-11): Document new D and h constraints. Update
590 description of O constraint.
592 2018-06-27 Jeff Law <law@redhat.com>
593 Austin Law <austinklaw@gmail.com>
595 * config/v850/v850.md (addsi3_set_flags): New pattern.
596 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
597 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
598 (zero_extendhisi2_v850_set_flags): Likewise.
599 (zero_extendqisi2_v850_set_flags): Likewise.
600 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
601 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
602 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
604 * config/v850/v850-protos.h (notice_update_cc): Remove.
605 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
606 (v850_print_operand): Handle 'D' and "d".
607 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
608 Add handling of arithmetic/logical operations compared against zero.
609 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
610 Do not look at v850_compare_op, instead get mode from last argument.
611 (v850_gen_compare): Remove
612 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
613 after reload for prologue insns.
614 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
616 (construct_save_jarl): Likewise.
617 (TARGET_FLAGS_REGNUM): Define.
618 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
619 (NOTICE_UPDATE_CC): Remove.
620 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
621 than cc0. Conditionalize on reload_completed.
622 (cmpsi_insn, setfcc_insn): Likewise.
623 (tst1 splitter): Turn into define_and_split which sets the flags
625 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
626 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
627 (cstoresf4, cstoredf4): Clobber the flags.
628 (cmpsi, cmpsf, cmpdf): Remove expanders.
629 (setf_insn): Remove pattern.
630 (addsi3): Turn into define_and_split which clobbers the flags after
631 reload and a suitable pattern (addsi3_clobber_flags) for use after
633 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
634 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
635 (ashrsi3, ashrsi3_v850e2): Likewise.
636 (bins): Clobber the flags.
637 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
638 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
639 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
640 (call_value_internal_short, call_value_internal_long): Likewise.
641 (callt_save_interrupt, callt_return_interrupt): Likewise.
642 (save_interrupt, return_interrupt): Likewise.
643 (callt_save_all_interrupt, save_all_interrupt): Likewise.
644 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
645 (restore_all_interrupt, _restore_all_interrupt): Likewise.
646 (All FP comparisons): Only allow after reload has completed.
648 (divh, divhu): Tweak output template.
649 (branch_z_normal, branch_z_invert): Remove
650 (branch_nz_normal, branch_nz_invert): Likewise.
651 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
653 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
654 * config/v850/v850.c (notice_update_cc): Remove.
655 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
656 (CC_NO_CARRY): Likewise.
657 (NOTICE_UPDATE_CC): Define to nothing.
658 * config/v850/v850.md: Remove block comment on cc0 handling
659 Remove "cc" attribute from all patterns. Remove cc_status handling
660 from all patterns. Minor formatting fixes.
662 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
664 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
665 (cortex-a76.cortex-a55): Likewise.
666 * config/aarch64/aarch64-tune.md: Regenerate.
667 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
668 cortex-a76.cortex-a55.
670 2018-06-27 Jeff Law <law@redhat.com>
672 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
673 (MULTILIB_DIRNAMES): Similarly.
675 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
677 * gimple.h (gimple_return_retbnd): Delete.
678 (gimple_return_set_retbnd): Likewise.
679 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
680 gimple_return_set_retbnd.
681 * gimple-pretty-print.c (dump_gimple_return): Remove call to
682 gimple_return_retbnd and adjust.
683 * tree-inline.h (struct copy_body_data): Remove retbnd field.
684 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
685 Explicitly return NULL in a couple more cases. Move assertion
686 on debug statements and remove unreachable code.
687 (reset_debug_binding): Do not test id->retbnd.
688 (expand_call_inline): Do not set it.
690 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
692 * configure.ac: Add --disable-gcov option.
693 * configure: Regenerate.
694 * Makefile.in: Honour @enable_gcov@.
695 * doc/install.texi: Document --disable-gcov.
697 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
699 * config/arm/arm-cpus.in (cortex-a76): New entry.
700 (cortex-a76.cortex-a55): Likewise.
701 * config/arm/arm-tables.opt: Regenerate.
702 * config/arm/arm-tune.md: Likewise.
703 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
704 * doc/invoke.texi (ARM Options): Document cortex-a76 and
705 cortex-a76.cortex-a55.
707 2018-06-27 Tamar Christina <tamar.christina@arm.com>
710 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
712 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
714 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
716 (EPILOGUE_USES): Likewise.
718 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
720 * tree-inline.c (remap_location): New function extracted from...
721 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
722 (copy_phis_for_bb): ...here. Call remap_location.
723 (copy_cfg_body): Adjust call to copy_edges_for_bb.
725 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
727 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
728 unaligned vsx for 16B memset.
730 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
733 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
734 ieee128_float_type_node to long_double_type_node unless
735 TARGET_LONG_DOUBLE_128 is set.
737 2018-06-26 David Malcolm <dmalcolm@redhat.com>
739 * cfgloop.c (get_loop_location): Convert return type from
740 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
741 by implicit construction from rtx_insn *, and using
742 dump_user_location_t::from_function_decl for the fallback case.
743 * cfgloop.h (get_loop_location): Convert return type from
744 location_t to dump_user_location_t.
745 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
746 dump_printf_loc to pass in a dump_location_t rather than a
747 location_t, via the gimple stmt.
748 * coverage.c (get_coverage_counts): Update calls to
749 dump_printf_loc to pass in dump_location_t rather than a
751 * doc/optinfo.texi (Dump types): Convert example of
752 dump_printf_loc from taking "locus" to taking "insn". Update
753 description of the "_loc" calls to cover dump_location_t.
754 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
756 (dump_user_location_t::dump_user_location_t): New constructors,
757 from gimple *stmt and rtx_insn *.
758 (dump_user_location_t::from_function_decl): New function.
759 (dump_loc): Make static.
760 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
761 const dump_location_t &.
762 (dump_generic_expr_loc): Delete.
763 (dump_printf_loc): Convert param "loc" from location_t to
764 const dump_location_t &.
765 (selftest::test_impl_location): New function.
766 (selftest::dumpfile_c_tests): New function.
767 * dumpfile.h: Include "profile-count.h".
768 (class dump_user_location_t): New class.
769 (struct dump_impl_location_t): New struct.
770 (class dump_location_t): New class.
771 (dump_printf_loc): Convert 2nd param from source_location to
772 const dump_location_t &.
773 (dump_generic_expr_loc): Delete.
774 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
775 const dump_location_t &.
776 * gimple-fold.c (fold_gimple_assign): Update call to
777 dump_printf_loc to pass in a dump_location_t rather than a
778 location_t, via the gimple stmt.
779 (gimple_fold_call): Likewise.
780 * gimple-loop-interchange.cc
781 (loop_cand::analyze_iloop_reduction_var): Update for change to
782 check_reduction_path.
783 (tree_loop_interchange::interchange): Update for change to
785 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
786 change in return-type of find_loop_location.
787 (graphite_regenerate_ast_isl): Likewise.
788 * graphite-optimize-isl.c (optimize_isl): Likewise.
789 * graphite.c (graphite_transform_loops): Likewise.
790 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
791 pass in a dump_location_t rather than a location_t, via the
793 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
794 * ipa.c (walk_polymorphic_call_targets): Likewise.
795 * loop-unroll.c (report_unroll): Convert "locus" param from
796 location_t to dump_location_t.
797 (decide_unrolling): Update for change to get_loop_location's
799 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
800 location_t to dump_user_location_t.
801 (grid_find_single_omp_among_assignments_1): Updates calls to
802 dump_printf_loc to pass in a dump_location_t rather than a
803 location_t, via the gimple stmt.
804 (grid_parallel_clauses_gridifiable): Convert "tloc" from
805 location_t to dump_location_t. Updates calls to dump_printf_loc
806 to pass in a dump_location_t rather than a location_t, via the
808 (grid_inner_loop_gridifiable_p): Likewise.
809 (grid_dist_follows_simple_pattern): Likewise.
810 (grid_gfor_follows_tiling_pattern): Likewise.
811 (grid_target_follows_gridifiable_pattern): Likewise.
812 (grid_attempt_target_gridification): Convert initialization
813 of local "grid" from memset to zero-initialization; FIXME: does
814 this require C++11? Update call to dump_printf_loc to pass in a
815 optinfo_location rather than a location_t, via the gimple stmt.
816 * profile.c (read_profile_edge_counts): Updates call to
817 dump_printf_loc to pass in a dump_location_t rather than a
819 (compute_branch_probabilities): Likewise.
820 * selftest-run-tests.c (selftest::run_tests): Call
822 * selftest.h (dumpfile_c_tests): New decl.
823 * tree-loop-distribution.c (pass_loop_distribution::execute):
824 Update for change in return type of find_loop_location.
825 * tree-parloops.c (parallelize_loops): Likewise.
826 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
827 "locus" from location_t to dump_user_location_t.
828 (canonicalize_loop_induction_variables): Likewise.
829 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
830 for change in return type of find_loop_location.
831 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
832 to dump_printf_loc to pass in a dump_location_t rather than a
833 location_t, via the stmt.
834 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
836 * tree-vect-loop-manip.c (find_loop_location): Convert return
837 type from source_location to dump_user_location_t.
838 (vect_do_peeling): Update for above change.
839 (vect_loop_versioning): Update for change in type of
841 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
842 from location_t to dump_user_location_t.
843 (vect_estimate_min_profitable_iters): Update for change in type
845 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
846 location_t to dump_location_t.
847 (vect_slp_bb): Update for change in type of vect_location.
848 * tree-vectorizer.c (vect_location): Convert from source_location
849 to dump_user_location_t.
850 (try_vectorize_loop_1): Update for change in vect_location's type.
851 (vectorize_loops): Likewise.
852 (increase_alignment): Likewise.
853 * tree-vectorizer.h (vect_location): Convert from source_location
854 to dump_user_location_t.
855 (find_loop_location): Convert return type from source_location to
856 dump_user_location_t.
857 (check_reduction_path): Convert 1st param from location_t to
858 dump_user_location_t.
859 * value-prof.c (check_counter): Update call to dump_printf_loc to
860 pass in a dump_user_location_t rather than a location_t; update
861 call to error_at for change in type of "locus".
862 (check_ic_target): Update call to dump_printf_loc to
863 pass in a dump_user_location_t rather than a location_t, via the
866 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
868 * config/s390/s390.h (enum processor_flags): Do not use
869 default tune parameter when -march was specified.
871 2018-06-26 Jakub Jelinek <jakub@redhat.com>
874 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
875 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
878 2018-06-26 Richard Biener <rguenther@suse.de>
880 PR tree-optimization/86287
882 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
883 (vect_analyze_loop): Initialize n_stmts.
885 2018-06-26 Richard Biener <rguenther@suse.de>
888 * fold-const.c (fold_convertible_p): Pointer extension
891 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
894 * dwarf2out.c (loc_list_has_views): Adjust comments.
895 (dw_loc_list): Split single cross-partition range with
898 2018-06-25 Jeff Law <law@redhat.com>
900 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
901 on -mbig-switch by default.
903 * config/v850/predicates.md (const_float_1_operand): Fix match_code
905 (const_float_0_operand): Remove unused predicate.
906 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
907 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
908 (recipsf2): New expander. Original pattern now called
910 (recipdf2, recipdf2_insn): Similarly.
911 (rsqrtsf2, rsqrtsf2_insn): Similarly
912 (rsqrtdf2, rsqrtdf2_insn): Similarly
914 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
916 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
917 Simplify logic for FreeBSD (twice).
919 2018-06-25 Martin Sebor <msebor@redhat.com>
921 PR tree-optimization/86204
922 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
923 a strnlen result if it's less than the length of the string.
925 2018-06-25 Martin Sebor <msebor@redhat.com>
927 PR tree-optimization/85700
928 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
929 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
930 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
932 2018-06-25 Martin Sebor <msebor@redhat.com>
934 * doc/extend.texi (Zero-length arrays): Update and clarify.
936 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
938 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
939 added IEEE/IBM long double multilib support on PowerPC little
940 endian Linux systems.
941 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
942 (MULTILIB_DEFAULTS): Likewise.
943 * config/rs6000/rs6000.c (rs6000_option_override_internal):
945 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
946 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
947 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
949 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
952 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
953 (REORDER_45): Likewise.
955 2018-06-25 Jeff Law <law@redhat.com>
957 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
958 dividend to 32 bits. Adjust length.
959 (udivmodhi4): Cleanup output template. Fix length.
961 2018-06-25 Carl Love <cel@us.ibm.com>
963 * config/rs6000/vsx.md: Change word selector to prefered location.
965 2018-06-25 Richard Biener <rguenther@suse.de>
967 PR tree-optimization/86304
968 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
969 epilogue-if-converted loops as well.
971 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
973 * lto-section-out.c (lto_begin_section): Do not print section
974 name for noaddr and unnumbered dumps.
976 2018-06-25 Richard Biener <rguenther@suse.de>
978 * tree-vectorizer.h (struct vec_info_shared): New structure
979 with parts split out from struct vec_info and loop_nest from
980 struct _loop_vec_info.
981 (struct vec_info): Adjust accordingly.
982 (struct _loop_vec_info): Likewise.
983 (LOOP_VINFO_LOOP_NEST): Adjust.
984 (LOOP_VINFO_DATAREFS): Likewise.
985 (LOOP_VINFO_DDRS): Likewise.
986 (struct _bb_vec_info): Likewise.
987 (BB_VINFO_DATAREFS): Likewise.
988 (BB_VINFO_DDRS): Likewise.
989 (struct _stmt_vec_info): Add dr_aux member.
990 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
991 (DR_MISALIGNMENT_UNINITIALIZED): New.
992 (set_dr_misalignment): Adjust.
993 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
994 (vect_analyze_loop): Adjust prototype.
995 (vect_analyze_loop_form): Likewise.
996 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
997 Compute dependences lazily.
998 (vect_record_base_alignments): Use shared datarefs/ddrs.
999 (vect_verify_datarefs_alignment): Likewise.
1000 (vect_analyze_data_refs_alignment): Likewise.
1001 (vect_analyze_data_ref_accesses): Likewise.
1002 (vect_analyze_data_refs): Likewise.
1003 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
1004 constructor parameter for shared part.
1005 (vect_analyze_loop_form): Pass in shared part and adjust.
1006 (vect_analyze_loop_2): Pass in storage for the number of
1007 stmts. Move loop nest finding to the caller. Compute
1009 (vect_analyze_loop): Pass in shared part.
1010 (vect_transform_loop): Verify shared datarefs are unchanged.
1011 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
1012 constructor parameter for shared part.
1013 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
1014 (vect_slp_bb): Verify shared datarefs are unchanged before
1016 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
1018 (new_stmt_vec_info): Initialize DR_AUX misalignment to
1019 DR_MISALIGNMENT_UNINITIALIZED.
1020 * tree-vectorizer.c (vec_info::vec_info): Add constructor
1021 parameter for shared part.
1022 (vec_info::~vec_info): Adjust.
1023 (vec_info_shared::vec_info_shared): New.
1024 (vec_info_shared::~vec_info_shared): Likewise.
1025 (vec_info_shared::save_datarefs): Likewise.
1026 (vec_info_shared::check_datarefs): Likewise.
1027 (try_vectorize_loop_1): Construct shared part live for analyses
1028 of a single loop for multiple vector sizes.
1029 * tree-parloops.c (gather_scalar_reductions): Adjust.
1031 2018-06-25 Richard Biener <rguenther@suse.de>
1033 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
1034 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
1035 (vect_analyze_data_refs): Remove similar code from here and
1036 simplify accordingly.
1038 2018-06-25 Richard Biener <rguenther@suse.de>
1040 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
1041 for reverse storage order accesses rather than asserting
1042 they cannot happen here.
1044 2018-06-25 Tom de Vries <tdevries@suse.de>
1047 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
1048 Use data16 instead of .byte for insn prefix.
1050 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
1053 * parser.c (make_char_string_pack): Pass this literal chars
1054 through cpp_interpret_string.
1055 (cp_parser_userdef_numeric_literal): Check the result of
1056 make_char_string_pack.
1058 2018-06-24 Maya Rashish <coypu@sdf.org>
1060 * ginclude/stddef.h: Simplify conditions around avoiding
1061 re-definition of __size_t.
1063 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
1065 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
1066 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
1068 2018-06-22 Maya Rashish <coypu@sdf.org>
1070 * doc/invoke.texi (mno-fancy-math-387): Update for changes
1071 made to OpenBSD and NetBSD through the years.
1073 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
1075 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
1076 behavior of vec_pack (vector double, vector double) to match
1077 behavior of vec_float2 (vector double, vector double).
1079 2018-06-22 Olivier Hainque <hainque@adacore.com>
1081 * gimplify.c (gimplify_function_tree): Prevent creation
1082 of a trampoline for the address of the current function
1083 passed to entry/exit instrumentation hooks.
1085 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
1088 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
1091 2018-06-22 Martin Liska <mliska@suse.cz>
1093 PR tree-optimization/86263
1094 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
1095 Bail out if is_enabled is false.
1096 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
1098 (jump_table_cluster::is_enabled): New function.
1100 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
1102 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
1103 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
1104 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
1105 (lto_input_ts_binfo_tree_pointers): Likewise.
1106 * tree-streamer-out.c (streamer_write_tree_bitfields,
1107 write_ts_binfo_tree_pointers): Likewise.
1108 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
1110 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
1112 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
1114 2018-06-22 Martin Liska <mliska@suse.cz>
1116 * symbol-summary.h (get): Make it pure and inline move
1117 functionality from ::get function.
1118 (get): Remove and inline into ::get and ::get_create.
1119 (get_create): Move code from ::get function.
1121 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1124 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
1125 -x assembler-with-cpp.
1127 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1129 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
1130 _FILE_OFFSET_BITS=64 for C++.
1132 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
1134 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
1135 conversion insn that shows up when pr85657-3.c is compiled using
1136 IEEE 128-bit long double.
1137 (neg<mode>2_internal): Use the correct mode to check whether the
1138 mode is IBM extended.
1139 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
1140 multiply and divide external functions from being created more
1143 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
1145 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
1147 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
1148 the edge can be forwarded.
1149 (cfg_layout_merge_blocks): Likewise.
1151 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
1153 * except.c (finish_eh_generation): Commit edge insertions only after
1154 the EH edges have been redirected from post-landing to landing pads.
1156 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
1158 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
1159 create_tmp_var_for to create the FRAME decl.
1160 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
1162 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
1164 * tree-inline.c (copy_edges_for_bb): Minor tweak.
1165 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
1166 debug statement when resetting its value.
1167 (expand_call_inline): Copy the locus of the call onto the assignment
1168 of the return value, if any. Use local variable in more cases.
1170 2018-06-21 Martin Liska <mliska@suse.cz>
1172 * ipa-pure-const.c (propagate_nothrow): Use
1173 funct_state_summaries->get.
1174 (dump_malloc_lattice): Likewise.
1175 (propagate_malloc): Likewise.
1177 2018-06-21 Richard Biener <rguenther@suse.de>
1179 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
1180 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
1181 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
1183 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
1184 BLOCK_ABSTRACT_ORIGIN unconditionally.
1186 2018-06-21 David Malcolm <dmalcolm@redhat.com>
1188 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
1190 * ipa-reference.c (ipa_reference_c_finalize): Delete
1191 ipa_ref_opt_sum_summaries and set it to NULL.
1193 2018-06-21 Tom de Vries <tdevries@suse.de>
1195 PR tree-optimization/85859
1196 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
1197 test with comment from bb_no_side_effects_p.
1199 2018-06-21 Richard Biener <rguenther@suse.de>
1201 PR tree-optimization/86232
1202 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
1203 max for constant niter.
1205 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
1207 * config/aarch64/aarch64-simd.md
1208 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
1210 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
1212 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
1213 Make opernads of the unspec commutative.
1215 2018-06-21 Richard Biener <rguenther@suse.de>
1217 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
1218 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
1219 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
1220 (vect_analyze_data_ref_dependence): Re-order checks to deal with
1222 (vect_record_base_alignments): Do not record base alignment
1223 for gathers or scatters.
1224 (vect_compute_data_ref_alignment): Drop return value that is always
1225 true. Bail out early for gathers or scatters.
1226 (vect_enhance_data_refs_alignment): Bail out early for gathers
1228 (vect_find_same_alignment_drs): Likewise.
1229 (vect_analyze_data_refs_alignment): Remove dead code.
1230 (vect_slp_analyze_and_verify_node_alignment): Likewise.
1231 (vect_analyze_data_refs): For possible gathers or scatters do
1232 not create an alternate DR, just check their possible validity
1233 and mark them. Adjust DECL_NONALIASED handling to not rely
1235 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
1236 update inits of gathers or scatters.
1237 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
1238 Also copy gather/scatter flag to pattern vinfo.
1240 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
1242 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
1243 behavior of vec_packsu (vector unsigned long long, vector unsigned
1244 long long) to match behavior of vec_packs with same signature.
1246 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
1247 Thomas Schwinge <thomas@codesourcery.com>
1248 Cesar Philippidis <cesar@codesourcery.com>
1250 * gimplify.c (gimplify_scan_omp_clauses): Add support for
1251 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
1252 (gimplify_adjust_omp_clauses): Likewise.
1253 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
1254 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
1255 (gimplify_omp_target_update): Update handling of acc update and
1257 * omp-low.c (install_var_field): Remove unused parameter
1258 base_pointers_restrict.
1259 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
1260 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
1262 (omp_target_base_pointers_restrict_p): Delete.
1263 (scan_omp_target): Update call to scan_sharing_clauses.
1264 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
1266 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
1267 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
1268 (convert_local_omp_clauses): Likewise.
1269 * tree-pretty-print.c (dump_omp_clause): Likewise.
1270 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
1272 (omp_clause_code_name): Likewise.
1274 2018-06-20 Jakub Jelinek <jakub@redhat.com>
1277 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
1280 PR tree-optimization/86231
1281 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
1282 anti-range don't overwrite *vr0min before using it to compute *vr0max.
1284 2018-06-20 Tom de Vries <tdevries@suse.de>
1286 PR tree-optimization/86097
1287 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
1288 iv type if signedness of iv type is not the same as that of *nit.
1290 2018-06-20 Jakub Jelinek <jakub@redhat.com>
1292 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
1293 EDGE_EH edges, verify they are all EDGE_EH.
1295 2018-06-20 Maya Rashish <coypu@sdf.org>
1297 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
1299 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1301 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
1302 * config/aarch64/aarch64.c (xgene1_tunings): Add
1303 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
1304 (aarch64_mode_valid_for_sched_fusion_p):
1305 Allow 16-byte modes.
1306 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
1307 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
1309 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
1311 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
1312 * config/aarch64/iterators.md (VQ2): New mode iterator.
1314 2018-06-20 Martin Liska <mliska@suse.cz>
1316 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
1317 Change default ratio from 10 to 8.
1319 2018-06-20 Martin Liska <mliska@suse.cz>
1321 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
1323 (bit_test_cluster::find_bit_tests): Likewise.
1324 (switch_decision_tree::analyze_switch_statement): Find clusters.
1325 * tree-switch-conversion.h (struct jump_table_cluster): Document
1328 2018-06-20 Martin Liska <mliska@suse.cz>
1330 * tree-switch-conversion.c (switch_conversion::collect):
1331 Record m_uniq property.
1332 (switch_conversion::expand): Bail out for special conditions.
1333 (group_cluster::~group_cluster): New.
1334 (group_cluster::group_cluster): Likewise.
1335 (group_cluster::dump): Likewise.
1336 (jump_table_cluster::emit): New.
1337 (switch_decision_tree::fix_phi_operands_for_edges): New.
1338 (struct case_node): Remove struct.
1339 (jump_table_cluster::can_be_handled): New.
1340 (case_values_threshold): Moved to header.
1341 (reset_out_edges_aux): Likewise.
1342 (jump_table_cluster::is_beneficial): New.
1343 (bit_test_cluster::can_be_handled): Likewise.
1344 (add_case_node): Remove.
1345 (bit_test_cluster::is_beneficial): New.
1346 (case_bit_test::cmp): New.
1347 (bit_test_cluster::emit): New.
1348 (expand_switch_as_decision_tree_p): Remove.
1349 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
1350 (fix_phi_operands_for_edge): Likewise.
1351 (switch_decision_tree::analyze_switch_statement): New.
1352 (compute_cases_per_edge): Move ...
1353 (switch_decision_tree::compute_cases_per_edge): ... here.
1354 (try_switch_expansion): Likewise.
1355 (switch_decision_tree::try_switch_expansion): Likewise.
1356 (record_phi_operand_mapping): Likewise.
1357 (switch_decision_tree::record_phi_operand_mapping): Likewise.
1358 (emit_case_decision_tree): Likewise.
1359 (switch_decision_tree::emit): Likewise.
1360 (balance_case_nodes): Likewise.
1361 (switch_decision_tree::balance_case_nodes): Likewise.
1362 (dump_case_nodes): Likewise.
1363 (switch_decision_tree::dump_case_nodes): Likewise.
1364 (emit_jump): Likewise.
1365 (switch_decision_tree::emit_jump): Likewise.
1366 (emit_cmp_and_jump_insns): Likewise.
1367 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
1368 (emit_case_nodes): Likewise.
1369 (switch_decision_tree::emit_case_nodes): Likewise.
1370 (conditional_probability): Remove.
1371 * tree-switch-conversion.h (enum cluster_type): New.
1373 (struct cluster): Likewise.
1374 (cluster::cluster): Likewise.
1375 (struct simple_cluster): Likewise.
1376 (simple_cluster::simple_cluster): Likewise.
1377 (struct group_cluster): Likewise.
1378 (struct jump_table_cluster): Likewise.
1379 (struct bit_test_cluster): Likewise.
1380 (struct min_cluster_item): Likewise.
1381 (struct case_tree_node): Likewise.
1382 (case_tree_node::case_tree_node): Likewise.
1383 (jump_table_cluster::case_values_threshold): Likewise.
1384 (struct case_bit_test): Likewise.
1385 (struct switch_decision_tree): Likewise.
1386 (struct switch_conversion): Likewise.
1387 (switch_decision_tree::reset_out_edges_aux): Likewise.
1389 2018-06-20 Martin Liska <mliska@suse.cz>
1391 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
1392 (hoist_edge_and_branch_if_true): Likewise.
1393 (expand_switch_using_bit_tests_p): Likewise.
1394 (struct case_bit_test): Likewise.
1395 (case_bit_test_cmp): Likewise.
1396 (emit_case_bit_tests): Likewise.
1397 (switch_conversion::switch_conversion): New class.
1398 (struct switch_conv_info): Remove old struct.
1399 (collect_switch_conv_info): More to ...
1400 (switch_conversion::collect): ... this.
1401 (check_range): Likewise.
1402 (switch_conversion::check_range): Likewise.
1403 (check_all_empty_except_final): Likewise.
1404 (switch_conversion::check_all_empty_except_final): Likewise.
1405 (check_final_bb): Likewise.
1406 (switch_conversion::check_final_bb): Likewise.
1407 (create_temp_arrays): Likewise.
1408 (switch_conversion::create_temp_arrays): Likewise.
1409 (free_temp_arrays): Likewise.
1410 (gather_default_values): Likewise.
1411 (switch_conversion::gather_default_values): Likewise.
1412 (build_constructors): Likewise.
1413 (switch_conversion::build_constructors): Likewise.
1414 (constructor_contains_same_values_p): Likewise.
1415 (switch_conversion::contains_same_values_p): Likewise.
1416 (array_value_type): Likewise.
1417 (switch_conversion::array_value_type): Likewise.
1418 (build_one_array): Likewise.
1419 (switch_conversion::build_one_array): Likewise.
1420 (build_arrays): Likewise.
1421 (switch_conversion::build_arrays): Likewise.
1422 (gen_def_assigns): Likewise.
1423 (switch_conversion::gen_def_assigns): Likewise.
1424 (prune_bbs): Likewise.
1425 (switch_conversion::prune_bbs): Likewise.
1426 (fix_phi_nodes): Likewise.
1427 (switch_conversion::fix_phi_nodes): Likewise.
1428 (gen_inbound_check): Likewise.
1429 (switch_conversion::gen_inbound_check): Likewise.
1430 (process_switch): Use the newly created class.
1431 (switch_conversion::expand): New.
1432 (switch_conversion::~switch_conversion): New.
1433 * tree-switch-conversion.h: New file.
1435 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
1437 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
1438 tree-vect-patterns.c.
1439 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
1440 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
1441 (vect_recog_sad_pattern): Likewise.
1442 (vect_recog_widen_sum_pattern): Likewise.
1443 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
1444 (vect_recog_widen_shift_pattern): Remove the type_in argument.
1445 (vect_recog_rotate_pattern): Likewise.
1446 (vect_recog_mult_pattern): Likewise.
1447 (vect_recog_vector_vector_shift_pattern): Likewise.
1448 (vect_recog_divmod_pattern): Likewise.
1449 (vect_recog_mixed_size_cond_pattern): Likewise.
1450 (vect_recog_bool_pattern): Likewise.
1451 (vect_recog_mask_conversion_pattern): Likewise.
1452 (vect_try_gather_scatter_pattern): Likewise.
1453 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
1454 (vect_recog_over_widening_pattern): Likewise.
1455 (vect_recog_gather_scatter_pattern): Likewise.
1456 (vect_recog_func_ptr): Move from tree-vectorizer.h
1457 (vect_vect_recog_func_ptrs): Move further down the file.
1458 (vect_recog_func): Likewise. Remove the third argument.
1459 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
1460 (vect_pattern_recog_1): Expect the pattern function to do any
1461 necessary target tests. Also expect it to provide a vector type.
1462 Remove the type_in handling.
1464 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
1466 * tree-vect-patterns.c (vect_pattern_detected): New function.
1467 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
1468 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
1469 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
1470 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
1471 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
1472 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
1473 (vect_recog_mask_conversion_pattern)
1474 (vect_try_gather_scatter_pattern): Likewise.
1476 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
1478 * tree-vect-patterns.c (vect_get_internal_def): New function.
1479 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
1480 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
1481 (search_type_for_mask_1): Use it.
1483 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
1485 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
1486 redundant WIDEN_SUM_EXPR handling.
1487 (vect_recog_sad_pattern): Likewise.
1489 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
1491 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
1492 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
1493 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
1494 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
1495 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
1497 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
1499 * tree-vect-stmts.c (vectorizable_call): Make sure that we
1500 use the stmt_vec_info of the original bb statement for the
1501 new zero assignment, even if the call is part of a pattern.
1503 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
1505 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
1506 that the sequence is attached to the original statement rather
1507 than the pattern statement.
1508 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
1509 PATTERN_DEF_SEQ from the original statement rather than
1510 the main pattern statement.
1511 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
1512 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
1513 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
1515 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
1517 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
1518 definition statements before the early exit for statements that aren't
1520 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
1522 (vect_transform_loop): ...here. Process pattern definition
1523 statements without first checking whether the main pattern
1524 statement is live or relevant.
1526 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
1528 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
1529 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
1531 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
1533 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
1534 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
1535 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
1536 (expand_block_compare): Change select_block_compare_mode call.
1537 (expand_strncmp_align_check): Use new functions, fix comment.
1538 (emit_final_str_compare_gpr): New function.
1539 (expand_strn_compare): Refactor and clean up code.
1540 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
1542 2018-06-19 Tony Reix <tony.reix@atos.com>
1543 Damien Bergamini <damien.bergamini@atos.com>
1544 David Edelsohn <dje.gcc@gmail.com>
1546 * collect2.c (static_obj): New variable.
1547 (static_libs): New variable.
1548 (is_in_list): Uncomment declaration.
1549 (main): Track AIX libraries linked statically.
1550 (is_in_list): Uncomment definition.
1551 (scan_prog_file): Don't add AIX shared libraries initializer
1552 to constructor list if linking statically.
1554 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
1556 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
1558 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
1560 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
1562 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
1565 2018-06-19 Martin Liska <mliska@suse.cz>
1567 * config/i386/i386.c (ix86_can_inline_p): Do not use
1568 ipa_fn_summaries::get_create.
1569 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
1571 (devirtualization_time_bonus): Likewise.
1572 (ipcp_propagate_stage): Likewise.
1573 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
1574 (edge_set_predicate): Likewise.
1575 (evaluate_conditions_for_known_args): Likewise.
1576 (evaluate_properties_for_edge): Likewise.
1577 (ipa_call_summary::reset): Tranform to ...
1578 (ipa_call_summary::~ipa_call_summary): ... this.
1579 (ipa_fn_summary::reset): Transform to ...
1580 (ipa_fn_summary::~ipa_fn_summary): ... this.
1581 (ipa_fn_summary_t::remove): Rename to ...
1582 (ipa_fn_summary_t::remove_callees): ... this.
1583 (ipa_fn_summary_t::duplicate): Use placement new
1584 instead of memory copy.
1585 (ipa_call_summary_t::duplicate): Likewise.
1586 (ipa_call_summary_t::remove): Remove.
1587 (dump_ipa_call_summary): Change get_create to get.
1588 (ipa_dump_fn_summary): Dump only when summary exists.
1589 (analyze_function_body): Use symbol_summary::get instead
1591 (compute_fn_summary): Likewise.
1592 (estimate_edge_devirt_benefit): Likewise.
1593 (estimate_edge_size_and_time): Likewise.
1594 (inline_update_callee_summaries): Likewise.
1595 (remap_edge_change_prob): Likewise.
1596 (remap_edge_summaries): Likewise.
1597 (ipa_merge_fn_summary_after_inlining): Likewise.
1598 (write_ipa_call_summary): Likewise.
1599 (ipa_fn_summary_write): Likewise.
1600 (ipa_free_fn_summary): Likewise.
1601 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
1602 (struct ipa_call_summary): Likewise.
1603 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
1605 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
1606 (estimate_size_after_inlining): Likewise.
1607 (estimate_growth): Likewise.
1608 (growth_likely_positive): Likewise.
1609 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
1610 (inline_call): Likewise.
1611 * ipa-inline.c (caller_growth_limits): Likewise.
1612 (can_inline_edge_p): Likewise.
1613 (can_inline_edge_by_limits_p): Likewise.
1614 (compute_uninlined_call_time): Likewise.
1615 (compute_inlined_call_time): Likewise.
1616 (want_inline_small_function_p): Likewise.
1617 (edge_badness): Likewise.
1618 (update_caller_keys): Likewise.
1619 (update_callee_keys): Likewise.
1620 (inline_small_functions): Likewise.
1621 (inline_to_all_callers_1): Likewise.
1622 (dump_overall_stats): Likewise.
1623 (early_inline_small_functions): Likewise.
1624 (early_inliner): Likewise.
1625 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
1626 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
1627 * ipa-pure-const.c (malloc_candidate_p): Likewise.
1628 * ipa-split.c (execute_split_functions): Likewise.
1629 * symbol-summary.h: Likewise.
1630 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
1632 2018-06-19 Richard Biener <rguenther@suse.de>
1634 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
1635 (vectorize_loops): ... here. Fix dbgcnt handling.
1636 (try_vectorize_loop): Wrap try_vectorize_loop_1.
1638 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
1641 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
1642 ieee128 argument takes up only one (vector) register, not two (floating
1645 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
1647 * gimplify.c (gimplify_init_constructor): Really never clear for an
1648 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
1650 2018-06-19 Richard Biener <rguenther@suse.de>
1652 PR tree-optimization/86179
1653 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
1654 after failed recognition.
1656 2018-06-18 Martin Sebor <msebor@redhat.com>
1659 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
1660 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
1661 Handle integer subtraction.
1662 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
1663 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
1665 2018-06-18 David Malcolm <dmalcolm@redhat.com>
1667 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
1668 param from rtx to rtx_insn *.
1669 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
1671 (frv_ifcvt_modify_insn): Likwise.
1672 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
1673 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
1674 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
1675 as_a <rtx_insn *> cast to local "unprotected_region" once
1676 it's been established that it's not NULL or pc_rtx.
1677 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
1678 param "sethi" from rtx to rtx_insn *.
1679 (nds32_group_float_insns): Likewise for param "insn".
1680 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
1682 (vax_output_int_subtract): Likewise.
1683 * config/vax/vax.c (vax_output_int_add): Likewise for param
1685 (vax_output_int_subtract): Likewise.
1686 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
1687 (emit_pattern_after): Likewise for param "after".
1688 (emit_insn_after): Likewise.
1689 (emit_jump_insn_after): Likewise.
1690 (emit_call_insn_after): Likewise.
1691 (emit_debug_insn_after): Likewise.
1692 (emit_pattern_before): Likewise for param "before".
1693 (emit_insn_before): Likewise.
1694 (emit_jump_insn_before): Likewise.
1695 * final.c (get_insn_template): Likewise for param "insn", removing
1697 * output.h (get_insn_template): Likewise for 2nd param.
1698 * rtl.h (emit_insn_before): Likewise.
1699 (emit_jump_insn_before): Likewise.
1700 (emit_debug_insn_before_noloc): Likewise.
1701 (emit_insn_after): Likewise.
1702 (emit_jump_insn_after): Likewise.
1703 (emit_call_insn_after): Likewise.
1704 (emit_debug_insn_after): Likewise.
1705 (set_insn_deleted): Likewise for param.
1707 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
1710 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
1711 floating point modes, so that IFmode is numerically greater than
1712 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
1713 to declare the ordering. This prevents IFmode from being
1714 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
1715 machine. Include rs6000-modes.h to share the fractional values
1716 between genmodes* and the rest of the compiler.
1720 * config/rs6000/rs6000-modes.h: New file.
1721 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
1722 meaning of rs6000_long_double_size so that 126..128 selects an
1723 appropriate 128-bit floating point type.
1724 (rs6000_option_override_internal): Likewise.
1725 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
1726 (TARGET_LONG_DOUBLE_128): Change the meaning of
1727 rs6000_long_double_size so that 126..128 selects an appropriate
1728 128-bit floating point type.
1729 (LONG_DOUBLE_TYPE_SIZE): Update comment.
1730 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
1731 source and destination to match the standard usage.
1732 (truncifkf2): Likewise.
1733 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
1734 ISA 2.07 to use an explicit clobber, instead of passing in a
1736 (copysign<mode>3_soft): Likewise.
1738 2018-06-18 David Malcolm <dmalcolm@redhat.com>
1740 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
1741 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
1742 (vect_slp_analyze_instance_dependence): Likewise.
1743 (vect_enhance_data_refs_alignment): Likewise.
1744 (vect_analyze_data_refs_alignment): Likewise.
1745 (vect_slp_analyze_and_verify_instance_alignment
1746 (vect_analyze_data_ref_accesses): Likewise.
1747 (vect_prune_runtime_alias_test_list): Likewise.
1748 (vect_analyze_data_refs): Likewise.
1749 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
1750 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
1751 (vect_analyze_scalar_cycles_1): Likewise.
1752 (vect_get_loop_niters): Likewise.
1753 (vect_analyze_loop_form_1): Likewise.
1754 (vect_update_vf_for_slp): Likewise.
1755 (vect_analyze_loop_operations): Likewise.
1756 (vect_analyze_loop): Likewise.
1757 (vectorizable_induction): Likewise.
1758 (vect_transform_loop): Likewise.
1759 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
1760 * tree-vect-slp.c (vect_analyze_slp): Likewise.
1761 (vect_make_slp_decision): Likewise.
1762 (vect_detect_hybrid_slp): Likewise.
1763 (vect_slp_analyze_operations): Likewise.
1764 (vect_slp_bb): Likewise.
1765 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
1766 (vectorizable_bswap): Likewise.
1767 (vectorizable_call): Likewise.
1768 (vectorizable_simd_clone_call): Likewise.
1769 (vectorizable_conversion): Likewise.
1770 (vectorizable_assignment): Likewise.
1771 (vectorizable_shift): Likewise.
1772 (vectorizable_operation): Likewise.
1773 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
1775 2018-06-18 Martin Sebor <msebor@redhat.com>
1777 PR tree-optimization/81384
1778 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
1779 * builtins.c (expand_builtin_strnlen): New function.
1780 (expand_builtin): Call it.
1781 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
1782 * builtins.def (BUILT_IN_STRNLEN): New.
1783 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
1784 Warn for bounds in excess of maximum object size.
1785 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
1786 single-value ranges. Handle strnlen.
1787 (handle_builtin_strlen): Handle strnlen.
1788 (strlen_check_and_optimize_stmt): Same.
1789 * doc/extend.texi (Other Builtins): Document strnlen.
1791 2018-06-18 Maya Rashish <coypu@sdf.org>
1793 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
1794 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
1795 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
1797 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
1799 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
1801 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1803 * tree.c (escaped_string::escape): Replace cast to char * by
1804 const_cast<char *> (unescaped).
1806 2018-06-18 Nick Clifton <nickc@redhat.com>
1809 * tree.c (escaped_string): New class. Converts an unescaped
1810 string into its escaped equivalent.
1811 (warn_deprecated_use): Use the new class to convert the
1812 deprecation message, if present.
1813 (test_escaped_strings): New self test.
1814 (test_c_tests): Add test_escaped_strings.
1815 * doc/extend.texi (deprecated): Add a note that the
1816 deprecation message is affected by the -fmessage-length
1817 option, and that control characters will be escaped.
1818 (#pragma GCC error): Document this pragma.
1819 (#pragma GCC warning): Likewise.
1820 * doc/invoke.texi (-fmessage-length): Document this option's
1821 effect on the #warning and #error preprocessor directives and
1822 the deprecated attribute.
1824 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
1826 * tree.c (decl_value_expr_lookup): Revert latest change.
1827 (decl_value_expr_insert): Likewise.
1829 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
1831 * gimplify.c (nonlocal_vlas): Delete.
1832 (nonlocal_vla_vars): Likewise.
1833 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
1835 (gimplify_body): Do not create and destroy nonlocal_vlas.
1836 * tree-nested.c: Include diagnostic.h.
1837 (use_pointer_in_frame): Tweak.
1838 (lookup_field_for_decl): Add assertion and declare the transformation.
1839 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
1840 internal error when the reference is in a wrong context. Do not
1841 create a debug decl by default.
1842 (note_nonlocal_block_vlas): Delete.
1843 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
1844 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
1845 create a debug decl by default.
1846 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
1847 call is in a wrong context.
1848 (fixup_vla_decls): New function.
1849 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
1850 debug variables were created.
1851 * tree.c (decl_value_expr_lookup): Add checking assertion.
1852 (decl_value_expr_insert): Likewise.
1854 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
1857 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
1858 * tree-scalar-evolution.c (interpret_expr): Likewise.
1859 (expression_expensive_p): Likewise.
1860 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
1861 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
1862 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
1863 (ssa_defined_by_minus_one_stmt_p): New.
1865 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
1868 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
1869 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
1870 * dojump.c (do_jump): Likewise.
1871 * expr.c (expand_expr_real_2): Check operand type's sign.
1872 * fold-const.c (const_unop): Handle ABSU_EXPR.
1873 (fold_abs_const): Likewise.
1874 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
1875 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
1876 (strip_sign_op_1): Likesise.
1877 * match.pd: Add new pattern to generate ABSU_EXPR.
1878 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
1879 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
1880 * tree-eh.c (operation_could_trap_helper_p): Likewise.
1881 * tree-inline.c (estimate_operator_cost): Likewise.
1882 * tree-pretty-print.c (dump_generic_node): Likewise.
1883 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
1884 * tree.def (ABSU_EXPR): New.
1886 2018-06-16 Jakub Jelinek <jakub@redhat.com>
1889 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
1890 documented as preserved for backward compatibility only.
1891 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
1893 PR rtl-optimization/86108
1894 * bb-reorder.c (create_forwarder_block): Renamed to ...
1895 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
1896 jump from new landing pad to the second part.
1897 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
1900 2018-06-15 Jakub Jelinek <jakub@redhat.com>
1903 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
1904 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
1905 Only call store_expr for halves if the mode is the same.
1908 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
1909 Fix up comment formatting.
1911 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
1913 * typed-splay-tree.h (typed_splay_tree::remove): New function.
1914 (typed_splay_tree::closure,
1915 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
1916 (typed_splay_tree::typed_splay_tree,
1917 typed_splay_tree::operator =): Declared private.
1918 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
1919 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
1920 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
1921 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
1922 typed_splay_tree::splay_tree_splay,
1923 typed_splay_tree::splay_tree_foreach_helper,
1924 typed_splay_tree::splay_tree_insert,
1925 typed_splay_tree::splay_tree_remove,
1926 typed_splay_tree::splay_tree_lookup,
1927 typed_splay_tree::splay_tree_predecessor,
1928 typed_splay_tree::splay_tree_successor,
1929 typed_splay_tree::splay_tree_min,
1930 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
1931 (typed_splay_tree::root, typed_splay_tree::comp,
1932 typed_splay_tree::delete_key,
1933 typed_splay_tree::delete_value): New data members.
1934 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
1935 typed_splay_tree::remove.
1937 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
1939 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
1940 -mginv and -mno-ginv to the assembler.
1941 * config/mips/mips.opt (-mcrc): New option.
1943 * doc/invoke.text (-mcrc): Document.
1946 2018-06-15 Nick Clifton <nickc@redhat.com>
1949 * tree.c (escaped_string): New class. Converts an unescaped
1950 string into its escaped equivalent.
1951 (warn_deprecated_use): Use the new class to convert the
1952 deprecation message, if present.
1953 (test_escaped_strings): New self test.
1954 (test_c_tests): Add test_escaped_strings.
1955 * doc/extend.texi (deprecated): Add a note that the
1956 deprecation message is affected by the -fmessage-length
1957 option, and that control characters will be escaped.
1958 (#pragma GCC error): Document this pragma.
1959 (#pragma GCC warning): Likewise.
1960 * doc/invoke.texi (-fmessage-length): Document this option's
1961 effect on the #warning and #error preprocessor directives and
1962 the deprecated attribute.
1964 2018-06-15 Richard Biener <rguenther@suse.de>
1966 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
1967 here, also noting vector size used.
1968 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
1969 size used in MSG_OPTIMIZED_LOCATIONS dump.
1970 (pass_slp_vectorize::execute): Adjust.
1972 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
1975 * config/arc/arc.c (arc_return_address_register): Fix
1978 2018-06-15 Richard Biener <rguenther@suse.de>
1981 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
1982 leave useless conversion stripping to force_gimple_operand_gsi.
1983 (gimplify_build2): Likewise.
1984 (gimplify_build1): Likewise.
1986 2018-06-15 Richard Biener <rguenther@suse.de>
1989 * tree-cfg.c (move_stmt_op): unshare invariant addresses
1990 before adjusting their block.
1992 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
1994 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
1995 multilibs for *-*-rtems*.
1996 * config/riscv/t-rtems: New file.
1998 2018-06-14 Jakub Jelinek <jakub@redhat.com>
2001 * match.pd ((A +- CST1) +- CST2): Punt if last resort
2002 unsigned_type_for returns NULL.
2005 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
2006 subregs of multi-word pseudos unless the float mode has word size.
2008 2018-06-14 Richard Biener <rguenther@suse.de>
2011 * tree-vect-generic.c (build_word_mode_vector_type): Remove
2012 duplicate and harmful type_hash_canon.
2013 * tree.c (type_hash_canon): Assert we didn't find ourselves.
2015 2018-06-14 Richard Biener <rguenther@suse.de>
2018 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
2021 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
2023 * config/rtems.h (STDINT_LONG32): Define.
2025 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
2026 Prachi Godbole <prachi.godbole@imgtec.com>
2028 * config/mips/mips-cpus.def: Define P6600.
2029 * config/mips/mips-tables.opt: Regenerate.
2030 * config/mips/mips.c (mips_ucbranch_type): New enum.
2031 (mips_rtx_cost_data): Add support for P6600.
2032 (mips_issue_rate): Likewise.
2033 (mips_multipass_dfa_lookahead): Likewise.
2034 (mips_avoid_hazard): Likewise.
2035 (mips_reorg_process_insns): Likewise.
2036 (mips_classify_branch_p6600): New function.
2037 * config/mips/mips.h (TUNE_P6600): New define.
2038 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
2039 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
2040 * config/mips/mips.md: Include p6600.md.
2041 (processor): Add p6600.
2042 * config/mips/p6600.md: New file.
2043 * doc/invoke.texi: Add p6600 to supported architectures.
2045 2018-06-13 Martin Sebor <msebor@redhat.com>
2047 PR tree-optimization/86114
2048 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
2050 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
2052 2018-06-13 Richard Biener <rguenther@suse.de>
2054 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
2055 Properly set vector type of the intermediate stmt.
2056 * tree-vect-stmts.c (vectorizable_operation): The destination
2057 var always has vectype_out type.
2059 2018-06-13 Jeff Law <law@redhat.com>
2061 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
2062 integer 0 for argument to print_rtl_with_bb.
2063 (rl78_reorg): Likewise.
2065 2018-06-13 David Malcolm <dmalcolm@redhat.com>
2067 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
2068 from rtx to rtx_insn *.
2069 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
2071 (add_sched_insns_for_speculation): Likewise for local "target",
2072 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
2073 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
2074 from rtx_insn ** to rtx_code_label **.
2075 (reorg_emit_nops): Likewise.
2076 (c6x_reorg): Likewise for local "call_labels".
2077 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
2079 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
2080 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
2081 the loops over LABEL_REFS.
2082 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
2084 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
2085 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
2086 (split_branches): Strengthen local "olabel" from rtx to
2087 rtx_insn *, adding a safe_as_a cast.
2088 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
2090 (add_insn_after): Likewise for first two params.
2091 (add_insn_before): Likewise.
2092 (remove_insn): Likewise for param.
2093 (emit_pattern_before_noloc): Likewise for second and third params.
2094 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
2095 (emit_call_insn_before_noloc): Likewise.
2096 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
2098 (emit_barrier_before): Likewise.
2099 (emit_label_before): Strengthen "label" param from "rtx" to
2100 "rtx_code_label *". Strengthen "before" param from "rtx" to
2102 (emit_insn_after_1): Strengthen "after" param from "rtx" to
2104 (emit_pattern_after_noloc): Likewise.
2105 (emit_insn_after_noloc): Likewise.
2106 (emit_jump_insn_after_noloc): Likewise.
2107 (emit_call_insn_after_noloc): Likewise.
2108 (emit_debug_insn_after_noloc): Likewise.
2109 (emit_barrier_after): Likewise.
2110 (emit_label_after): Likewise for both params.
2111 (emit_pattern_after_setloc): Likewise for "after" param. Convert
2112 "loc" param from "int" to "location_t".
2113 (emit_insn_after_setloc): Likewise.
2114 (emit_jump_insn_after_setloc): Likewise.
2115 (emit_call_insn_after_setloc): Likewise.
2116 (emit_debug_insn_after_setloc): Likewise.
2117 (emit_pattern_before_setloc): Likewise for "before" param. Convert
2118 "loc" param from "int" to "location_t".
2119 (emit_pattern_before): Convert NULL_RTX to NULL.
2120 (emit_insn_before_setloc): Convert "loc" param from "int" to
2122 (emit_jump_insn_before_setloc): Likewise.
2123 (emit_call_insn_before_setloc): Likewise.
2124 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
2125 rtx_insn *. Convert "loc" param from "int" to "location_t".
2126 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
2127 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
2128 Convert 3rd param from "int" to "location_t".
2129 (emit_barrier_before, emit_barrier_after, next_real_insn):
2130 Strengthen param from rtx to rtx_insn *.
2131 (emit_label_before): Strengthen 1st param from "rtx" to
2132 "rtx_code_label *". Strengthen 2nd param from "rtx" to
2134 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
2135 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
2136 Strengthen 2nd param from "rtx" to "rtx_insn *".
2137 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
2138 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
2139 Likewise. Convert 3rd param from "int" to "location_t".
2140 (emit_label_after): Strengthen 1st param from "rtx" to
2142 (next_real_insn, remove_insn): Strengthen param from "rtx" to
2144 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
2145 from "rtx" to "rtx_insn *".
2147 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
2149 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
2150 bodies streamed in with -Q.
2151 * dumpfile.c (dump_files): Add lto-stream-out dump file.
2152 * dumpfile.h (tree_dump_index): Add lto_stream_out.
2153 * gimple-streamer-out.c: Include gimple-pretty-print.h
2154 (output_bb): Dump stmts streamed.
2155 * lto-section-out.c: Include print-tree.h
2156 (lto_begin_section): Dump sections created.
2157 (lto_output_decl_index): Dump decl encoded.
2158 * lto-streamer-out.c: Include print-tree.h
2159 (create_output_block): Dump output block created.
2160 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
2161 (output_function): Dump function output.
2162 (output_constructor): Dump constructor streamed.
2163 (write_global_stream): Output indexes encoded.
2164 (produce_asm_for_decls): Dump streams encoded.
2165 * lto-streamer.c (streamer_dump_file): New global var.
2166 * lto-streamer.h (streamer_dump_file): Declare.
2167 * passes.c (ipa_write_summaries): Initialize streamer dump.
2168 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
2171 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
2174 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
2175 offsets for register save directives. Emit a second batch of save
2176 directives, if need be, when the function accesses prior frames.
2178 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
2180 * config/arc/fpu.md (fmasf4): Force operand to register.
2181 (fnmasf4): Likewise.
2183 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
2185 * config/arc/arc-protos.h (arc_pad_return): Remove.
2186 * config/arc/arc.c (machine_function): Remove force_short_suffix
2188 (arc_print_operand): Adjust printing of '&'.
2189 (arc_verify_short): Remove conditional printing of short suffix.
2190 (arc_final_prescan_insn): Remove reference to size_reason.
2191 (pad_return): New function.
2192 (arc_reorg): Call pad_return.
2193 (arc_pad_return): Remove.
2194 (arc_init_machine_status): Remove reference to force_short_suffix.
2195 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
2196 (attr length): When attribute iscompact is true force to 2
2197 regardless; in the case of maybe check if we want to force the
2198 instruction to have 4 bytes length.
2199 (nopv): Change it to generate 4 byte long nop as well.
2200 (blockage): New pattern.
2201 (simple_return): Remove call to arc_pad_return.
2202 (p_return_i): Likewise.
2204 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
2206 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
2208 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
2210 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
2213 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
2215 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
2216 for ARC700 and ARCv2.
2218 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
2221 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
2222 operands[2] instead of operands[1].
2225 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
2227 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
2228 case, check whether the outer register overlaps an unallocatable
2229 register, not just whether it fits the required class.
2231 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
2233 * poly-int.h (can_div_trunc_p): Add new overload in which all values
2235 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
2236 (memrefs_conflict_p): Likewise.
2237 (init_alias_analysis): Likewise.
2238 * cfgexpand.c (expand_debug_expr): Likewise.
2239 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
2240 * cse.c (fold_rtx): Likewise.
2241 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
2242 * expr.c (emit_block_move_hints): Likewise.
2243 (clear_storage_hints, push_block, emit_push_insn): Likewise.
2244 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
2245 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
2246 (emit_group_store): Likewise.
2247 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
2248 to read the PRE/POST_MODIFY increment.
2249 * calls.c (store_one_arg): Use strip_offset.
2250 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
2252 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
2254 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
2255 (simplify_binary_operation_1): Extend CONST_INT handling to
2257 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
2258 than a HOST_WIDE_INT.
2259 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
2261 (adjust_mems, add_stores): Update accodingly.
2262 (vt_canonicalize_addr): Track polynomial offsets.
2263 (emit_note_insn_var_location): Likewise.
2264 (vt_add_function_parameter): Likewise.
2265 (vt_initialize): Likewise.
2267 2018-06-12 Jeff Law <law@redhat.com>
2269 * config.gcc (alpha*-*-freebsd*): Remove.
2270 * config/alpha/freebsd.h: Remove.
2272 2018-06-12 David Malcolm <dmalcolm@redhat.com>
2275 * spellcheck-tree.c (levenshtein_distance): Rename to...
2276 (get_edit_distance): ...this, and update for underlying renaming.
2277 * spellcheck-tree.h (levenshtein_distance): Rename to...
2278 (get_edit_distance): ...this.
2279 * spellcheck.c (levenshtein_distance): Rename to...
2280 (get_edit_distance): ...this. Convert from Levenshtein distance
2281 to Damerau-Levenshtein distance by supporting transpositions of
2282 adjacent characters. Rename "v1" to "v_next" and "v0" to
2284 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
2285 (selftest::test_edit_distance_unit_test_oneway): ...this, and
2286 update for underlying renaming.
2287 (selftest::levenshtein_distance_unit_test): Rename to...
2288 (selftest::test_get_edit_distance_unit): ...this, and update for
2289 underlying renaming.
2290 (selftest::test_find_closest_string): Add example from PR 69968
2291 where transposition helps
2292 (selftest::test_metric_conditions): Update for renaming.
2293 (selftest::test_metric_conditions): Likewise.
2294 (selftest::spellcheck_c_tests): Likewise.
2295 * spellcheck.h (levenshtein_distance): Rename both overloads to...
2296 (get_edit_distance): ...this.
2297 (best_match::consider): Update for renaming.
2299 2018-06-12 Martin Sebor <msebor@redhat.com>
2301 PR tree-optimization/85259
2302 * builtins.c (compute_objsize): Handle constant offsets.
2303 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
2304 true iff a warning has been issued.
2305 * gimple.h (gimple_nonartificial_location): New function.
2306 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
2307 gimple_nonartificial_location and handle -Wno-system-headers.
2308 (handle_builtin_stxncpy): Same.
2310 2018-06-12 Martin Sebor <msebor@redhat.com>
2313 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
2315 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
2317 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
2318 BUILTIN_VEC_XST entries for pointer to double and long long.
2320 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
2323 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
2325 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
2328 2018-06-12 Martin Liska <mliska@suse.cz>
2330 * doc/options.texi: Document IntegerRange.
2332 2018-06-12 Martin Liska <mliska@suse.cz>
2334 * config/i386/i386.opt: Make MPX-related options as Deprecated.
2335 * opt-functions.awk: Handle Deprecated flag.
2336 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
2338 (read_cmdline_option): Report warning for a deprecated option.
2339 * opts.h (struct cl_option): Add new field cl_deprecated.
2340 (CL_ERR_DEPRECATED): New.
2342 2018-06-12 Martin Liska <mliska@suse.cz>
2344 * doc/options.texi: Document Deprecated option flag.
2346 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
2348 * config/arc/arc-arch.h (arc_extras): New enum.
2349 (arc_cpu_t):Add field extra.
2350 (arc_cpu_types): Consider the extras.
2351 * config/arc/arc-cpus.def: Add extras info.
2352 * config/arc/arc-opts.h (processor_type): Consider extra field.
2353 * config/arc/arc.c (arc_override_options): Handle extra field.
2355 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
2357 * config/arc/arc-arch.h: Update ARC_OPTX macro.
2358 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
2360 * config/arc/arc.c (arc_init): Update pic warning.
2361 (irq_range): Update irq range parsing warnings.
2362 (arc_override_options): Update various warning messages.
2363 (arc_handle_aux_attribute): Likewise.
2365 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
2367 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
2369 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
2371 * doc/sourcebuild.texi: Document usage of line number 0 in verify
2372 compiler messages directives.
2374 2018-06-12 Matthew Fortune <mfortune@gmail.com>
2376 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
2377 * config/mips/mips-tables.opt: Regenerate.
2378 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
2380 * doc/invoke.texi: Document -march=i6500.
2382 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
2384 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
2385 (i6400_gpmul): Add cpu_unit.
2386 (i6400_gpdiv): Likewise.
2387 (i6400_msa_add_d): Update reservations.
2388 (i6400_msa_int_add) Likewise.
2389 (i6400_msa_short_logic3) Likewise.
2390 (i6400_msa_short_logic2) Likewise.
2391 (i6400_msa_short_logic) Likewise.
2392 (i6400_msa_move) Likewise.
2393 (i6400_msa_cmp) Likewise.
2394 (i6400_msa_short_float2) Likewise.
2395 (i6400_msa_div_d) Likewise.
2396 (i6400_msa_long_logic1) Likewise.
2397 (i6400_msa_long_logic2) Likewise.
2398 (i6400_msa_mult) Likewise.
2399 (i6400_msa_long_float2) Likewise.
2400 (i6400_msa_long_float4) Likewise.
2401 (i6400_msa_long_float5) Likewise.
2402 (i6400_msa_long_float8) Likewise.
2403 (i6400_fpu_fadd): Include frint type.
2404 (i6400_fpu_store): New define_insn_reservation.
2405 (i6400_fpu_load): Likewise.
2406 (i6400_fpu_move): Likewise.
2407 (i6400_fpu_fcmp): Likewise.
2408 (i6400_fpu_fmadd): Likewise.
2409 (i6400_int_mult): Include imul3nc type and update reservation.
2410 (i6400_int_div): Include idiv3 type and update reservation.
2411 (i6400_int_load): Update to check type not move_type.
2412 (i6400_int_store): Likewise.
2413 (i6400_int_prefetch): Set zero latency.
2415 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
2417 * gcc.c: Document new %@{...} sequence.
2418 (LINK_COMMAND_SPEC): Use it for the -L switches.
2419 (cpp_unique_options): Use it for the -I switches.
2420 (at_file_argbuf): New global variable.
2421 (in_at_file): Likewise.
2422 (alloc_args): Create at_file_argbuf.
2423 (clear_args): Truncate at_file_argbuf.
2424 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
2425 (open_at_file): New function.
2426 (close_at_file): Likewise.
2427 (create_at_file): Delete.
2428 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
2431 (validate_switches_from_spec): Deal with %@{...} sequence.
2432 (validate_switches): Likewise.
2433 (driver::finalize): Call clear_args.
2435 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
2437 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
2439 2018-06-11 Martin Sebor <msebor@redhat.com>
2441 * doc/invoke.texi (-Wall): List -Wc++17-compat.
2442 (Wno-class-memaccess): Add @opindex.
2443 (Wno-templates, Wno-multiple-inheritance): Same.
2444 (Wno-virtual-inheritance, Wno-namespaces): Same.
2445 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
2446 (Wno-format-overflow, Wno-format-truncation): Same.
2447 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
2448 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
2449 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
2450 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
2451 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
2452 (Wno-misspelled-isr): Same.
2454 2018-06-11 Martin Sebor <msebor@redhat.com>
2456 * PR tree-optimization/86083
2457 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
2459 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
2461 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
2463 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
2466 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
2467 on the correct operand.
2468 (*movdi_internal64): Ditto.
2470 2018-06-11 Martin Liska <mliska@suse.cz>
2472 PR tree-optimization/86089
2473 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
2475 2018-06-11 Julia Koval <julia.koval@intel.com>
2477 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
2478 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
2479 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
2481 2018-06-11 Olivier Hainque <hainque@adacore.com>
2483 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
2484 for Ada with strict dwarf2.
2486 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
2489 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
2492 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
2494 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
2496 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
2498 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
2501 2018-06-08 Martin Liska <mliska@suse.cz>
2503 * tree-cfg.h (debug_function): Fix argument type to match
2506 2018-06-08 Martin Liska <mliska@suse.cz>
2508 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
2509 Remove usage of MPX-related (and removed) fields.
2510 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
2512 2018-06-08 David Malcolm <dmalcolm@redhat.com>
2514 * cfg.c (debug): Use TDF_NONE rather than 0.
2515 * cfghooks.c (debug): Likewise.
2516 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
2517 (struct dump_option_value_info): Convert to...
2518 (struct kv_pair): ...this template type.
2519 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
2521 (optinfo_verbosity_options): Likewise.
2522 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
2524 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
2525 than int for "optgroup_flags" param.
2526 (dump_generic_expr_loc): Use dump_flags_t rather than int for
2528 (dump_dec): Likewise.
2529 (dump_finish): Use TDF_NONE rather than 0.
2530 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
2531 rather than int for "optgroup_flags" param. Use TDF_NONE rather
2532 than 0. Update for change to option_ptr.
2533 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
2534 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
2535 0. Update for changes to optinfo_verbosity_options and
2537 (opt_info_switch_p): Convert optgroup_flags from int to
2539 (dump_basic_block): Use dump_flags_t rather than int
2540 for "dump_kind" param.
2541 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
2542 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
2543 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
2544 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
2545 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
2546 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
2547 TDF_NONE): Convert from macros to...
2548 (enum dump_flag): ...this new enum.
2549 (dump_flags_t): Update to use enum.
2550 (operator|, operator&, operator~, operator|=, operator&=):
2551 Implement for dump_flags_t.
2552 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
2553 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
2554 Convert from macros to...
2555 (enum optgroup_flag): ...this new enum.
2556 (optgroup_flags_t): New typedef.
2557 (operator|, operator|=): Implement for optgroup_flags_t.
2558 (struct dump_file_info): Convert field "alt_flags" to
2559 dump_flags_t. Convert field "optgroup_flags" to
2561 (dump_basic_block): Use dump_flags_t rather than int for param.
2562 (dump_generic_expr_loc): Likewise.
2563 (dump_dec): Likewise.
2564 (dump_register): Convert param "optgroup_flags" to
2566 (opt_info_enable_passes): Likewise.
2567 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
2569 * gimple-pretty-print.c (debug): Likewise.
2570 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
2571 (merged_store_group::apply_stores): Likewise.
2572 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
2573 * gimple.c (verify_gimple_pp): Likewise.
2574 * graphite-poly.c (print_pbb_body): Likewise.
2575 * passes.c (pass_manager::register_one_dump_file): Convert
2576 local "optgroup_flags" to optgroup_flags_t.
2577 * print-tree.c (print_node): Use TDF_NONE rather than 0.
2579 (debug_body): Likewise.
2580 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
2581 to optgroup_flags_t.
2582 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
2584 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
2585 (convert_mult_to_fma): Likewise.
2586 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
2587 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
2588 * tree-vect-data-refs.c (dump_lower_bound): Convert param
2589 "dump_kind" to dump_flags_t.
2591 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
2593 * config/rs6000/rs6000.c (min, max): Delete.
2595 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
2597 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
2598 -mabi=spe and -mabi=no-spe.
2600 2018-06-08 Martin Liska <mliska@suse.cz>
2602 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
2603 where we expect an existing summary.
2605 2018-06-08 Martin Liska <mliska@suse.cz>
2607 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
2608 * ipa-inline.h (estimate_edge_growth): Likewise.
2610 2018-06-08 Martin Liska <mliska@suse.cz>
2612 * cgraph.c (function_version_hasher::hash): Use
2613 cgraph_node::get_uid ().
2614 (function_version_hasher::equal):
2615 * cgraph.h (cgraph_node::get_uid): New method.
2616 * ipa-inline.c (update_caller_keys): Use
2617 cgraph_node::get_uid ().
2618 (update_callee_keys): Likewise.
2619 * ipa-utils.c (searchc): Likewise.
2620 (ipa_reduced_postorder): Likewise.
2621 * lto-cgraph.c (input_node): Likewise.
2622 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
2623 * symbol-summary.h (symtab_insertion): Likewise.
2624 (symtab_removal): Likewise.
2625 (symtab_duplication): Likewise.
2626 * tree-pretty-print.c (dump_function_header): Likewise.
2627 * tree-sra.c (convert_callers_for_node): Likewise.
2629 2018-06-08 Martin Liska <mliska@suse.cz>
2631 * cgraph.c (symbol_table::create_edge): Always assign a new
2633 (symbol_table::free_edge): Do not recycle numbers.
2634 * cgraph.h (cgraph_edge::get): New method.
2635 * symbol-summary.h (symtab_removal): Use it.
2636 (symtab_duplication): Likewise.
2637 (call_summary::hashable_uid): Remove.
2639 2018-06-08 Martin Liska <mliska@suse.cz>
2641 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
2642 (initialize_growth_caches): Remove.
2643 (free_growth_caches): Likewise.
2644 (do_estimate_edge_time): Use edge_growth_cache.
2645 (do_estimate_edge_size): Likewise.
2646 (do_estimate_edge_hints): Likewise.
2647 * ipa-inline.c (reset_edge_caches): Likewise.
2648 (recursive_inlining): Likewise.
2649 (inline_small_functions): Likewise.
2650 * ipa-inline.h (initialize_growth_caches): Remove.
2651 (estimate_edge_size): Likewise.
2652 (estimate_edge_time): Likewise.
2653 (estimate_edge_hints): Likewise.
2654 (reset_edge_growth_cache): Likewise.
2655 * symbol-summary.h (call_summary::remove): New method.
2657 2018-06-08 Martin Liska <mliska@suse.cz>
2659 * ipa-cp.c (class edge_clone_summary): New summary.
2660 (grow_edge_clone_vectors): Remove.
2661 (ipcp_edge_duplication_hook): Remove.
2662 (class edge_clone_summary_t): New call_summary class.
2663 (ipcp_edge_removal_hook): Remove.
2664 (edge_clone_summary_t::duplicate): New function.
2665 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
2666 (create_specialized_node): Likewise.
2667 (ipcp_driver): Initialize edge_clone_summaries and do not
2670 2018-06-08 Martin Liska <mliska@suse.cz>
2672 * symbol-summary.h (get): New function.
2673 (call_summary::m_initialize_when_cloning): New class member.
2675 2018-06-08 Martin Liska <mliska@suse.cz>
2677 * cgraph.c (cgraph_node::remove): Do not recycle uid.
2678 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
2679 (symbol_table::allocate_cgraph_symbol): Do not set uid.
2680 * passes.c (uid_hash_t): Record removed_nodes by their uids.
2681 (remove_cgraph_node_from_order): Use the removed_nodes set.
2682 (do_per_function_toporder): Likwise.
2683 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
2684 instead of summary_uid.
2685 (symtab_removal): Likewise.
2686 (symtab_duplication): Likewise.
2688 2018-06-08 Martin Liska <mliska@suse.cz>
2690 * ipa-cp.c (ipcp_store_bits_results): Use
2691 ipcp_transformation_sum.
2692 (ipcp_store_vr_results): Likewise.
2693 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
2695 (ipcp_transformation_initialize): ... this.
2696 (ipa_set_node_agg_value_chain):
2697 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
2698 (write_ipcp_transformation_info): Likewise.
2699 (read_ipcp_transformation_info): Likewise.
2700 (ipcp_update_bits): Likewise.
2701 (ipcp_update_vr): Likewise.
2702 (ipcp_transform_function): Likewise.
2703 * ipa-prop.h: Rename ipcp_transformation_summary to
2704 ipcp_transformation.
2705 (class ipcp_transformation_t): New function summary.
2706 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
2707 (ipa_get_agg_replacements_for_node): Likewise.
2709 2018-06-08 Martin Liska <mliska@suse.cz>
2711 * ipa-pure-const.c (struct funct_state_d): Do it class instead
2713 (class funct_state_summary_t): New function_summary class.
2714 (has_function_state): Remove.
2715 (get_function_state): Likewise.
2716 (set_function_state): Likewise.
2717 (add_new_function): Likewise.
2718 (funct_state_summary_t::insert): New function.
2719 (duplicate_node_data): Remove.
2720 (remove_node_data): Remove.
2721 (funct_state_summary_t::duplicate): New function.
2722 (register_hooks): Create new funct_state_summaries.
2723 (pure_const_generate_summary): Use it.
2724 (pure_const_write_summary): Likewise.
2725 (pure_const_read_summary): Likewise.
2726 (propagate_pure_const): Likewise.
2727 (propagate_nothrow): Likewise.
2728 (dump_malloc_lattice): Likewise.
2729 (propagate_malloc): Likewise.
2730 (execute): Do not register hooks, just remove summary
2732 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
2735 2018-06-08 Martin Liska <mliska@suse.cz>
2737 * ipa-reference.c (remove_node_data): Remove.
2738 (duplicate_node_data): Likewise.
2739 (class ipa_ref_var_info_summary_t): New class.
2740 (class ipa_ref_opt_summary_t): Likewise.
2741 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
2742 (get_reference_optimization_summary): Use
2743 ipa_ref_opt_sum_summaries.
2744 (set_reference_vars_info): Remove.
2745 (set_reference_optimization_summary): Likewise.
2746 (ipa_init): Create summaries.
2747 (init_function_info): Use function summary.
2748 (ipa_ref_opt_summary_t::duplicate): New function.
2749 (ipa_ref_opt_summary_t::remove): New function.
2750 (get_read_write_all_from_node): Fix GNU coding style.
2751 (propagate): Use function summary.
2752 (write_node_summary_p): Fix GNU coding style.
2753 (stream_out_bitmap): Likewise.
2754 (ipa_reference_read_optimization_summary): Use function summary.
2755 (ipa_reference_c_finalize): Do not release hooks.
2757 2018-06-08 Martin Liska <mliska@suse.cz>
2759 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
2760 (analyze_function_body): Extract multiple calls of get_create.
2761 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
2762 * ipa-inline.c (recursive_inlining): Use ::get method.
2763 * ipa-inline.h (estimate_edge_growth): Likewise.
2765 2018-06-08 Martin Liska <mliska@suse.cz>
2767 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
2769 (hsa_function_summary::hsa_function_summary): Use the new enum
2771 (hsa_gpu_implementation_p): Use hsa_summaries::get.
2772 * hsa-gen.c (hsa_get_host_function): Likewise.
2773 (get_brig_function_name): Likewise.
2774 * ipa-hsa.c (process_hsa_functions): Likewise.
2775 (ipa_hsa_write_summary): Likewise.
2776 * symbol-summary.h (symtab_duplication): Use ::get function/
2777 (get): New function.
2779 2018-06-08 Martin Liska <mliska@suse.cz>
2781 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
2783 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
2784 (hsa_register_kernel): Likewise.
2785 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
2786 * hsa-gen.c (hsa_get_host_function): Likewise.
2787 (get_brig_function_name): Likewise.
2788 (generate_hsa): Likewise.
2789 (pass_gen_hsail::execute): Likewise.
2790 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
2791 (devirtualization_time_bonus): Likewise.
2792 (ipcp_propagate_stage): Likewise.
2793 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
2794 (edge_set_predicate): Likewise.
2795 (evaluate_conditions_for_known_args): Likewise.
2796 (evaluate_properties_for_edge): Likewise.
2797 (ipa_fn_summary::reset): Likewise.
2798 (ipa_fn_summary_t::duplicate): Likewise.
2799 (dump_ipa_call_summary): Likewise.
2800 (ipa_dump_fn_summary): Likewise.
2801 (analyze_function_body): Likewise.
2802 (compute_fn_summary): Likewise.
2803 (estimate_edge_devirt_benefit): Likewise.
2804 (estimate_edge_size_and_time): Likewise.
2805 (estimate_calls_size_and_time): Likewise.
2806 (estimate_node_size_and_time): Likewise.
2807 (inline_update_callee_summaries): Likewise.
2808 (remap_edge_change_prob): Likewise.
2809 (remap_edge_summaries): Likewise.
2810 (ipa_merge_fn_summary_after_inlining): Likewise.
2811 (ipa_update_overall_fn_summary): Likewise.
2812 (read_ipa_call_summary): Likewise.
2813 (inline_read_section): Likewise.
2814 (write_ipa_call_summary): Likewise.
2815 (ipa_fn_summary_write): Likewise.
2816 (ipa_free_fn_summary): Likewise.
2817 * ipa-hsa.c (process_hsa_functions): Likewise.
2818 (ipa_hsa_write_summary): Likewise.
2819 (ipa_hsa_read_section): Likewise.
2820 * ipa-icf.c (sem_function::merge): Likewise.
2821 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
2822 (do_estimate_edge_time): Likewise.
2823 (estimate_size_after_inlining): Likewise.
2824 (estimate_growth): Likewise.
2825 (growth_likely_positive): Likewise.
2826 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
2827 (inline_call): Likewise.
2828 * ipa-inline.c (caller_growth_limits): Likewise.
2829 (can_inline_edge_p): Likewise.
2830 (can_inline_edge_by_limits_p): Likewise.
2831 (compute_uninlined_call_time): Likewise.
2832 (compute_inlined_call_time): Likewise.
2833 (want_inline_small_function_p): Likewise.
2834 (edge_badness): Likewise.
2835 (update_caller_keys): Likewise.
2836 (update_callee_keys): Likewise.
2837 (recursive_inlining): Likewise.
2838 (inline_small_functions): Likewise.
2839 (inline_to_all_callers_1): Likewise.
2840 (dump_overall_stats): Likewise.
2841 (early_inline_small_functions): Likewise.
2842 (early_inliner): Likewise.
2843 * ipa-inline.h (estimate_edge_growth): Likewise.
2844 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
2845 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
2846 * ipa-prop.h (IPA_NODE_REF): Likewise.
2847 (IPA_EDGE_REF): Likewise.
2848 * ipa-pure-const.c (malloc_candidate_p): Likewise.
2849 (propagate_malloc): Likewise.
2850 * ipa-split.c (execute_split_functions): Likewise.
2851 * symbol-summary.h: Rename get to get_create.
2853 (get_create): Likewise.
2854 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
2856 2018-06-08 Martin Liska <mliska@suse.cz>
2858 * symbol-summary.h (release): Move definition out of class
2860 (symtab_removal): Likewise.
2861 (symtab_duplication): Likewise.
2863 2018-06-08 Martin Liska <mliska@suse.cz>
2865 * symbol-summary.h (function_summary): Move constructor
2866 implementation out of class declaration.
2867 (release): Likewise.
2868 (symtab_insertion): Likewise.
2869 (symtab_removal): Likewise.
2870 (symtab_duplication): Likewise.
2873 2018-06-08 Martin Liska <mliska@suse.cz>
2875 * Makefile.in: Remove support for MPX (macros, related functions,
2876 fields in cgraph_node, ...).
2877 * builtin-types.def (BT_BND): Likewise.
2878 (BT_FN_BND_CONST_PTR): Likewise.
2879 (BT_FN_CONST_PTR_BND): Likewise.
2880 (BT_FN_VOID_PTR_BND): Likewise.
2881 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
2882 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
2883 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
2884 (expand_builtin_mempcpy_with_bounds): Likewise.
2885 (expand_builtin_memset_with_bounds): Likewise.
2886 (expand_builtin_memset_args): Likewise.
2887 (std_expand_builtin_va_start): Likewise.
2888 (expand_builtin): Likewise.
2889 (expand_builtin_with_bounds): Likewise.
2890 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
2891 (DEF_LIB_BUILTIN_CHKP): Likewise.
2892 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
2893 (DEF_CHKP_BUILTIN): Likewise.
2894 (BUILT_IN_MEMCPY): Likewise.
2895 (BUILT_IN_MEMMOVE): Likewise.
2896 (BUILT_IN_MEMPCPY): Likewise.
2897 (BUILT_IN_MEMSET): Likewise.
2898 (BUILT_IN_STPCPY): Likewise.
2899 (BUILT_IN_STRCAT): Likewise.
2900 (BUILT_IN_STRCHR): Likewise.
2901 (BUILT_IN_STRCPY): Likewise.
2902 (BUILT_IN_STRLEN): Likewise.
2903 (BUILT_IN_MEMCPY_CHK): Likewise.
2904 (BUILT_IN_MEMMOVE_CHK): Likewise.
2905 (BUILT_IN_MEMPCPY_CHK): Likewise.
2906 (BUILT_IN_MEMSET_CHK): Likewise.
2907 (BUILT_IN_STPCPY_CHK): Likewise.
2908 (BUILT_IN_STRCAT_CHK): Likewise.
2909 (BUILT_IN_STRCPY_CHK): Likewise.
2910 * calls.c (store_bounds): Likewise.
2911 (emit_call_1): Likewise.
2912 (special_function_p): Likewise.
2913 (maybe_warn_nonstring_arg): Likewise.
2914 (initialize_argument_information): Likewise.
2915 (finalize_must_preallocate): Likewise.
2916 (compute_argument_addresses): Likewise.
2917 (expand_call): Likewise.
2918 * cfgexpand.c (expand_call_stmt): Likewise.
2919 (expand_return): Likewise.
2920 (expand_gimple_stmt_1): Likewise.
2921 (pass_expand::execute): Likewise.
2922 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
2923 (cgraph_node::remove): Likewise.
2924 (cgraph_node::dump): Likewise.
2925 (cgraph_node::verify_node): Likewise.
2926 * cgraph.h (chkp_function_instrumented_p): Likewise.
2927 (symtab_node::get_alias_target): Likewise.
2928 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
2929 (cgraph_local_p): Likewise.
2930 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
2931 (cgraph_edge::rebuild_references): Likewise.
2932 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
2933 (walk_polymorphic_call_targets): Likewise.
2934 (cgraph_node::expand_thunk): Likewise.
2935 (symbol_table::output_weakrefs): Likewise.
2936 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
2937 (ix86_handle_option): Likewise.
2938 * config/i386/constraints.md: Likewise.
2939 * config/i386/i386-builtin-types.def (BND): Likewise.
2943 * config/i386/i386-builtin.def (BDESC_END): Likewise.
2944 (BDESC_FIRST): Likewise.
2946 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
2947 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
2948 * config/i386/i386.c (enum reg_class): Likewise.
2949 (ix86_target_string): Likewise.
2950 (ix86_option_override_internal): Likewise.
2951 (ix86_conditional_register_usage): Likewise.
2952 (ix86_valid_target_attribute_inner_p): Likewise.
2953 (ix86_set_indirect_branch_type): Likewise.
2954 (ix86_set_current_function): Likewise.
2955 (ix86_function_arg_regno_p): Likewise.
2956 (init_cumulative_args): Likewise.
2957 (ix86_function_arg_advance): Likewise.
2958 (ix86_function_arg): Likewise.
2959 (ix86_pass_by_reference): Likewise.
2960 (ix86_function_value_regno_p): Likewise.
2961 (ix86_function_value_1): Likewise.
2962 (ix86_function_value_bounds): Likewise.
2963 (ix86_return_in_memory): Likewise.
2964 (ix86_setup_incoming_vararg_bounds): Likewise.
2965 (ix86_va_start): Likewise.
2966 (indirect_thunk_need_prefix): Likewise.
2967 (print_reg): Likewise.
2968 (ix86_print_operand): Likewise.
2969 (ix86_expand_call): Likewise.
2970 (ix86_output_function_return): Likewise.
2971 (reg_encoded_number): Likewise.
2972 (BDESC_VERIFYS): Likewise.
2973 (ix86_init_mpx_builtins): Likewise.
2974 (ix86_init_builtins): Likewise.
2975 (ix86_emit_cmove): Likewise.
2976 (ix86_emit_move_max): Likewise.
2977 (ix86_expand_builtin): Likewise.
2978 (ix86_builtin_mpx_function): Likewise.
2979 (ix86_get_arg_address_for_bt): Likewise.
2980 (ix86_load_bounds): Likewise.
2981 (ix86_store_bounds): Likewise.
2982 (ix86_load_returned_bounds): Likewise.
2983 (ix86_store_returned_bounds): Likewise.
2984 (ix86_class_likely_spilled_p): Likewise.
2985 (ix86_hard_regno_mode_ok): Likewise.
2986 (x86_order_regs_for_local_alloc): Likewise.
2987 (ix86_mitigate_rop): Likewise.
2988 (ix86_bnd_prefixed_insn_p): Likewise.
2989 (ix86_mpx_bound_mode): Likewise.
2990 (ix86_make_bounds_constant): Likewise.
2991 (ix86_initialize_bounds): Likewise.
2992 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
2993 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
2994 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
2995 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
2996 (TARGET_CHKP_BOUND_MODE): Likewise.
2997 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
2998 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
2999 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
3000 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
3001 * config/i386/i386.h (TARGET_MPX): Likewise.
3002 (TARGET_MPX_P): Likewise.
3003 (VALID_BND_REG_MODE): Likewise.
3004 (FIRST_BND_REG): Likewise.
3005 (LAST_BND_REG): Likewise.
3006 (enum reg_class): Likewise.
3007 (BND_REG_P): Likewise.
3008 (BND_REGNO_P): Likewise.
3009 (BNDmode): Likewise.
3010 (ADJUST_INSN_LENGTH): Likewise.
3011 * config/i386/i386.md: Likewise.
3012 * config/i386/i386.opt: Likewise.
3013 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
3014 (defined): Likewise.
3015 (LINK_MPX): Likewise.
3016 (MPX_SPEC): Likewise.
3017 (LIBMPX_SPEC): Likewise.
3018 (LIBMPXWRAPPERS_SPEC): Likewise.
3019 (CHKP_SPEC): Likewise.
3020 * config/i386/predicates.md: Likewise.
3021 * dbxout.c (dbxout_type): Likewise.
3022 * doc/extend.texi: Likewise.
3023 * doc/invoke.texi: Likewise.
3024 * doc/md.texi: Likewise.
3025 * doc/tm.texi: Likewise.
3026 * doc/tm.texi.in: Likewise.
3027 * dwarf2out.c (is_base_type): Likewise.
3028 (gen_formal_types_die): Likewise.
3029 (gen_subprogram_die): Likewise.
3030 (gen_type_die_with_usage): Likewise.
3031 (gen_decl_die): Likewise.
3032 (dwarf2out_late_global_decl): Likewise.
3033 * expr.c (expand_assignment): Likewise.
3034 (emit_storent_insn): Likewise.
3035 (store_expr_with_bounds): Likewise.
3036 (store_expr): Likewise.
3037 (expand_expr_real_1): Likewise.
3038 * expr.h (store_expr_with_bounds): Likewise.
3039 * function.c (use_register_for_decl): Likewise.
3040 (struct bounds_parm_data): Likewise.
3041 (assign_parms_augmented_arg_list): Likewise.
3042 (assign_parm_find_entry_rtl): Likewise.
3043 (assign_parm_is_stack_parm): Likewise.
3044 (assign_parm_load_bounds): Likewise.
3045 (assign_bounds): Likewise.
3046 (assign_parms): Likewise.
3047 (expand_function_start): Likewise.
3048 * gcc.c (CHKP_SPEC): Likewise.
3049 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
3050 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
3051 (wrestrict_dom_walker::check_call): Likewise.
3052 * gimple.c (gimple_build_call_from_tree): Likewise.
3053 * gimple.h (enum gf_mask): Likewise.
3054 (gimple_call_with_bounds_p): Likewise.
3055 (gimple_call_set_with_bounds): Likewise.
3056 * gimplify.c (gimplify_init_constructor): Likewise.
3057 * ipa-cp.c (initialize_node_lattices): Likewise.
3058 (propagate_constants_across_call): Likewise.
3059 (find_more_scalar_values_for_callers_subset): Likewise.
3060 * ipa-hsa.c (process_hsa_functions): Likewise.
3061 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
3062 * ipa-icf.c (sem_function::merge): Likewise.
3063 * ipa-inline.c (early_inliner): Likewise.
3064 * ipa-pure-const.c (warn_function_noreturn): Likewise.
3065 (warn_function_cold): Likewise.
3066 (propagate_pure_const): Likewise.
3067 * ipa-ref.h (enum GTY): Likewise.
3068 * ipa-split.c (find_retbnd): Likewise.
3069 (consider_split): Likewise.
3070 (split_function): Likewise.
3071 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
3072 * ipa.c (walk_polymorphic_call_targets): Likewise.
3073 (symbol_table::remove_unreachable_nodes): Likewise.
3074 (process_references): Likewise.
3075 (cgraph_build_static_cdtor_1): Likewise.
3076 * lto-cgraph.c (lto_output_node): Likewise.
3077 (output_refs): Likewise.
3078 (compute_ltrans_boundary): Likewise.
3079 (input_overwrite_node): Likewise.
3080 (input_node): Likewise.
3081 (input_cgraph_1): Likewise.
3082 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
3083 * passes.c (pass_manager::execute_early_local_passes): Likewise.
3084 (class pass_chkp_instrumentation_passes): Likewise.
3085 (make_pass_chkp_instrumentation_passes): Likewise.
3086 * passes.def: Likewise.
3087 * rtl.h (struct GTY): Likewise.
3088 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
3089 * stor-layout.c (layout_type): Likewise.
3090 * symtab.c: Likewise.
3091 * target.def: Likewise.
3092 * targhooks.c (default_chkp_bound_type): Likewise.
3093 (default_chkp_bound_mode): Likewise.
3094 (default_builtin_chkp_function): Likewise.
3095 (default_chkp_function_value_bounds): Likewise.
3096 (default_chkp_make_bounds_constant): Likewise.
3097 (default_chkp_initialize_bounds): Likewise.
3098 * targhooks.h (default_chkp_bound_type): Likewise.
3099 (default_chkp_bound_mode): Likewise.
3100 (default_builtin_chkp_function): Likewise.
3101 (default_chkp_function_value_bounds): Likewise.
3102 (default_chkp_make_bounds_constant): Likewise.
3103 (default_chkp_initialize_bounds): Likewise.
3104 * toplev.c (compile_file): Likewise.
3105 (process_options): Likewise.
3106 * tree-core.h (DEF_BUILTIN): Likewise.
3107 (DEF_BUILTIN_CHKP): Likewise.
3108 * tree-inline.c (declare_return_variable): Likewise.
3109 (remap_gimple_stmt): Likewise.
3110 (copy_bb): Likewise.
3111 (initialize_inlined_parameters): Likewise.
3112 (expand_call_inline): Likewise.
3113 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
3114 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
3115 (make_pass_ipa_chkp_produce_thunks): Likewise.
3116 (make_pass_chkp): Likewise.
3117 (make_pass_chkp_opt): Likewise.
3118 (make_pass_chkp_instrumentation_passes): Likewise.
3119 * tree-pretty-print.c (dump_generic_node): Likewise.
3120 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
3121 * tree-ssa-dce.c (propagate_necessity): Likewise.
3122 (eliminate_unnecessary_stmts): Likewise.
3123 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
3124 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
3125 * tree-ssa-sccvn.h: Likewise.
3126 * tree-ssa-strlen.c (get_string_length): Likewise.
3127 (valid_builtin_call): Likewise.
3128 (adjust_last_stmt): Likewise.
3129 (handle_builtin_strchr): Likewise.
3130 (handle_builtin_strcpy): Likewise.
3131 (handle_builtin_stxncpy): Likewise.
3132 (handle_builtin_memcpy): Likewise.
3133 (handle_builtin_strcat): Likewise.
3134 (strlen_check_and_optimize_stmt): Likewise.
3135 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
3136 * tree-streamer-in.c: Likewise.
3137 * tree-streamer.c (record_common_node): Likewise.
3138 * tree.c (tree_code_size): Likewise.
3139 (wide_int_to_tree_1): Likewise.
3140 (type_contains_placeholder_1): Likewise.
3141 (build_common_tree_nodes): Likewise.
3142 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
3143 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
3144 (POINTER_BOUNDS_P): Likewise.
3145 (BOUNDED_TYPE_P): Likewise.
3146 (BOUNDED_P): Likewise.
3147 (CALL_WITH_BOUNDS_P): Likewise.
3148 (pointer_bounds_type_node): Likewise.
3149 * value-prof.c (gimple_ic): Likewise.
3150 * var-tracking.c (vt_add_function_parameters): Likewise.
3151 * varasm.c (make_decl_rtl): Likewise.
3152 (assemble_start_function): Likewise.
3153 (output_constant): Likewise.
3154 (maybe_assemble_visibility): Likewise.
3155 * varpool.c (ctor_for_folding): Likewise.
3156 * chkp-builtins.def: Remove.
3157 * ipa-chkp.c: Remove.
3158 * ipa-chkp.h: Remove.
3159 * rtl-chkp.c: Remove.
3160 * rtl-chkp.h: Remove.
3161 * tree-chkp-opt.c: Remove.
3162 * tree-chkp.c: Remove.
3163 * tree-chkp.h: Remove.
3165 2018-06-07 Carl Love <cel@us.ibm.com>
3167 * gcc/config/rs6000/vsx.md (vextract_fp_from_shorth,
3168 vextract_fp_from_shortl): Add BE support.
3170 2018-06-07 Paul Koning <ni1d@arrl.net>
3172 * compare-elim.c (try_merge_compare): Don't merge compare if
3173 address contains a side effect.
3174 (try_eliminate_compare): Likewise.
3176 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
3178 * config.gcc: Support "tremont".
3179 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
3180 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3182 * config/i386/i386.c (m_TREMONT): Define.
3183 (processor_target_table): Add "tremont".
3184 (PTA_TREMONT): Define.
3185 (ix86_lea_outperforms): Add TARGET_TREMONT.
3186 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
3187 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
3188 and M_INTEL_GOLDMONT_PLUS.
3189 (fold_builtin_cpu): Add "tremont".
3190 (ix86_add_stmt_cost): Add TARGET_TREMONT.
3191 (ix86_option_override_internal): Add "tremont".
3192 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
3193 (processor_type): Add PROCESSOR_TREMONT.
3194 * config/i386/x86-tune.def: Add m_TREMONT.
3195 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
3197 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
3199 * gcc/config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
3200 symbol defined for msp430i* devices to be lower case.
3202 gcc/testsuite/gcc.target/msp430/
3203 * msp430i-device-symbol.c: New test.
3204 * msp430f-device-symbol.c: New test.
3205 * msp430.h: New test header file.
3207 2018-06-07 Richard Biener <rguenther@suse.de>
3209 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
3210 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
3211 Properly wrap signed arithmetic if overflow wraps.
3213 2018-06-07 Jakub Jelinek <jakub@redhat.com>
3215 PR tree-optimization/69615
3216 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
3217 of a cast from a same precision integral SSA_NAME in a bb dominated
3218 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
3219 cast to utype if rhs2 has already a compatible type.
3221 2018-06-07 Richard Biener <rguenther@suse.de>
3223 PR tree-optimization/85935
3224 * graphite-scop-detection.c (find_params_in_bb): Analyze
3225 condition operands with respect to the correct loop. Assert
3226 the analysis doesn't fail.
3228 2018-06-04 Carl Love <cel@us.ibm.com>
3230 * gcc/config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
3231 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
3232 as it is slightly cheaper.
3233 (first_match_or_eos_index_<mode>):
3234 Calculate index using natural element order.
3235 (first_match_index_<mode>):
3236 Calculate index using natural element order.
3237 (first_match_or_eos_index_<mode>):
3238 Calculate index using natural order.
3239 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
3240 for BE and LE modes.
3241 * gcc/config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
3242 P9V_BUILTIN_VCLZLSBB_V16QI.
3243 * gcc/config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
3246 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
3248 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
3249 indentation and line wrap for many prototypes. Add missing
3250 @smallexample directives around block of prototypes for vec_xl and
3253 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
3255 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
3256 track if we pass or return IEEE 128-bit floating point.
3257 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
3258 C++ mangling that is compatible with GCC 8.1.
3259 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
3260 (init_cumulative_args): Note if we pass or return IEEE 128-bit
3261 floating point types.
3262 (rs6000_function_arg_advance_1): Likewise.
3263 (rs6000_mangle_type): Optionally generate mangled names that match
3264 what GCC 8.1 generated for IEEE 128-bit floating point types.
3265 (rs6000_globalize_decl_name): If we have an external function that
3266 passes or returns IEEE 128-bit types, generate a weak reference
3267 from the mangled name used in GCC 8.1 to the current mangled
3269 (rs6000_init_builtins): Make __ibm128 use the long double type if
3270 long double is IBM extended double. Make __float128 use the long
3271 double type if long double is IEEE 128-bit.
3274 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
3275 macro for __ibm128 built-in functions.
3276 (PACK_IF): Add __ibm128 pack/unpack functions.
3277 (UNPACK_IF): Likewise.
3278 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
3279 enable long double built-in functions if long double is IEEE
3280 128-bit floating point.
3281 (rs6000_invalid_builtin): Update long double built-in function
3283 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
3284 functions, adjust the built-in function to use the long double
3285 built-in function if __ibm128 and long double are the same type.
3286 * doc/extend.texi (PowerPC builtins): Update documention for
3287 __builtin_{,un}pack_longdouble. Add documentation for
3288 __builtin_{,un}pack_ibm128.
3290 2018-06-06 Jim Wilson <jimw@sifive.com>
3292 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
3293 (struct machine_function): New field interrupt_mode.
3294 (riscv_handle_type_attribute): New function. Add forward declaration.
3295 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
3296 (riscv_expand_epilogue): Check interrupt_mode field.
3297 (riscv_set_current_function): Check interrupt attribute args and
3298 set interrupt_mode field.
3299 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
3300 (riscv_sret, riscv_uret): New.
3301 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
3302 new arguments to interrupt attribute.
3304 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
3307 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
3308 Don't handle -mcpu=power8 if -mpower9-vector is also used.
3310 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
3312 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
3313 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
3314 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3315 several redundant entries.
3317 2018-06-06 David Malcolm <dmalcolm@redhat.com>
3319 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
3320 type from "rtx" to "rtx_insn *".
3321 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
3322 for local "call_insn", removing cast.
3323 (ix86_expand_call): Likewise, introducing a "call_insn" local.
3325 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
3327 PR tree-optimization/86066
3328 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
3329 for BIT_INSERT_EXPR stores.
3331 2018-06-06 Richard Biener <rguenther@suse.de>
3333 PR tree-optimization/86062
3334 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
3335 component refs ontop
3336 of to be offsetted base.
3338 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
3340 * gcc/config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
3341 to be static and remove check on interrupt attribute name.
3343 gcc/testsuite/gcc.target/msp430/
3344 * function-attributes-4.c: New test.
3345 * static-interrupts.c: New test.
3347 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
3349 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
3350 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
3352 2018-06-05 Steve Ellcey <sellcey@cavium.com>
3355 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
3357 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
3358 Remove second argument, change how error is called.
3359 (aarch64_layout_arg): Remove second argument from
3360 aarch64_err_no_fpadvsimd call.
3361 (aarch64_init_cumulative_args): Ditto.
3362 (aarch64_gimplify_va_arg_expr): Ditto.
3363 * config/aarch64/aarch64.md (mov<mode>): Ditto.
3365 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
3367 * config/i386/i386.md (simple_return_indirect_internal): New expander.
3368 (*simple_return_indirect_internal<mode>): Rename from
3369 simple_return_indirect_internal. Use W mode iterator.
3370 (rstorssp): New expander.
3371 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
3372 (clrssbsy): New expander.
3373 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
3375 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
3377 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
3379 (cmse_check_pointed_object): Likewise.
3381 2018-06-05 Martin Liska <mliska@suse.cz>
3383 PR gcov-profile/47618
3384 * doc/invoke.texi: Document how -fprofile-dir format
3387 2018-06-05 Richard Biener <rguenther@suse.de>
3389 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
3390 removal pretend DOM info isn't available so we do not update
3391 it and only remove edges, not dominated blocks. Actually free
3392 DOM info in case we removed something. Remove unreachable blocks.
3393 (mfb_keep_latches): Work with either DOM info or marked backedges.
3394 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
3395 first. Mark backedges if DOM info isn't available.
3396 (Re-)compute DOM info after cleanup_control_flow_pre.
3398 2018-06-05 Richard Biener <rguenther@suse.de>
3400 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
3401 (locus_discrim_hasher::hash): Adjust.
3402 (locus_discrim_hasher::equal): Likewise.
3403 (next_discriminator_for_locus): Work on line directly.
3404 (same_line_p): Pass in expanded locus1 as well.
3405 (assign_discriminators): Avoid redundant location expansions.
3407 2018-06-05 Richard Biener <rguenther@suse.de>
3409 PR tree-optimization/86046
3410 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
3411 if required after clearing TREE_ADDRESSABLE.
3413 2018-06-05 Richard Biener <rguenther@suse.de>
3415 PR tree-optimization/86047
3416 * tree-ssa-loop.c (for_each_index): Glob handling of all
3417 decls and constants and really handle all of them.
3419 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3422 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
3423 qualifier_void_pointer and qualifier_const_void_pointer.
3424 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
3425 (arm_init_builtins): Handle the above.
3426 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
3427 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
3428 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
3431 2018-06-05 Martin Liska <mliska@suse.cz>
3433 * auto-profile.c (read_autofdo_file): Do not use
3434 gcov_ctr_summary struct.
3435 (afdo_callsite_hot_enough_for_early_inline): Likewise.
3436 * coverage.c (struct counts_entry): Likewise.
3437 (read_counts_file): Read just single summary entry.
3438 (get_coverage_counts): Use gcov_summary struct.
3439 * coverage.h (get_coverage_counts): Likewise.
3440 * gcov-dump.c (dump_working_sets): Likewise.
3441 (tag_summary): Dump just single summary.
3442 * gcov-io.c (gcov_write_summary): Write just histogram
3444 (gcov_read_summary): Read just single summary.
3445 (compute_working_sets): Use gcov_summary struct.
3446 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
3447 of GCOV_COUNTERS_SUMMABLE.
3448 (GCOV_COUNTERS_SUMMABLE): Remove.
3449 (GCOV_FIRST_VALUE_COUNTER): Replace with
3450 GCOV_COUNTER_V_INTERVAL.
3451 (struct gcov_ctr_summary): Remove.
3452 (struct gcov_summary): Directly use fields of former
3454 (compute_working_sets): Use gcov_summary struct.
3455 * gcov.c (read_count_file): Do not use ctrs fields.
3456 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
3458 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
3460 * profile.c: Likewise.
3461 * profile.h: Likewise.
3463 2018-06-05 Martin Liska <mliska@suse.cz>
3465 PR gcov-profile/84846
3466 * gcov.c (output_lines): Print working directory only
3467 in intermediate format.
3469 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
3471 * config/s390/s390-builtin-types.def: Add void function type.
3472 * config/s390/s390-builtins.def: Use the function type for the
3475 2018-06-04 Jim Wilson <jimw@sifive.com>
3477 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
3479 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
3480 and maybe_eh_return. Change regno to unsigned int. Use new args to
3481 handle EH_RETURN_DATA_REGNO registers properly.
3482 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
3483 (riscv_expand_epilogue): Update comment. Change argument name and
3484 type. Update code to use new name and type. Pass new args to
3485 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
3487 * config/riscv/riscv.md (NORMAL_RETURN): New.
3488 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
3489 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
3490 (eh_return): Call gen_eh_return_internal and emit barrier.
3491 (eh_return_internal): Call riscv_expand_epilogue.
3493 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
3495 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
3496 bit_insertion field and declare can_be_merged_into method.
3497 (merged_store_group::can_be_merged_into): New method.
3498 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
3499 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
3500 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
3502 2018-06-04 Richard Biener <rguenther@suse.de>
3504 PR tree-optimization/85955
3505 * builtins.c (fold_builtin_sincos): Convert pointers to
3506 destination to appropriate type before dereferencing.
3508 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
3510 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
3512 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
3514 * expr.c (expand_expr_real_1): Force the operand into memory if
3515 its TYPE_MODE is BLKmode and if there is no integer mode for
3516 the number of bits being extracted.
3518 2018-06-04 Jakub Jelinek <jakub@redhat.com>
3522 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
3523 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
3525 2018-06-04 Richard Biener <rguenther@suse.de>
3527 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
3528 (cleanup_tree_cfg_noloop): ... single caller. Do
3529 start_recording_case_labels later.
3531 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
3533 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
3534 to _IMMINTRIN_H_INCLUDED.
3535 * config/i386/pconfigintrin.h: Ditto.
3536 * config/i386/waitpkgintrin.h: Ditto.
3537 * config/i386/immintrin.h: Add includes for sgxintrin.h,
3538 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
3539 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
3540 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
3541 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
3542 waitpkgintrin.h and cldemoteintrin.h.
3544 2018-06-04 Richard Biener <rguenther@suse.de>
3546 PR tree-optimization/86038
3547 * tracer.c (find_best_successor): Check probability for
3548 being initialized, bail out if not.
3550 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
3553 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
3554 of bits to ignore when comparing architectures.
3556 2018-06-04 Jakub Jelinek <jakub@redhat.com>
3558 PR tree-optimization/69615
3559 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
3560 maximum or minimum of the type, try to merge it also as if
3561 range1 is + [-, x - 1] or + [x + 1, -].
3564 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
3566 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
3568 PR tree-optimization/86034
3569 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
3570 the unsigned bitfield type in a bit insertion sequence if it does not
3571 have a larger precision than the bitfield size.
3572 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
3574 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
3576 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
3578 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
3580 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
3581 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
3582 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
3583 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
3585 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
3587 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
3588 Disable -fdelete-null-pointer-checks for ELF toolchain.
3590 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
3591 Kito Cheng <kito.cheng@gmail.com>
3593 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
3594 (nds32le-*-*, nds32be-*-*): Integrate checking process.
3595 (nds32*-*-*): Add glibc and uclibc conditions.
3596 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
3597 (TARGET_EXCEPT_UNWIND_INFO): Define.
3598 * config/nds32/elf.h: New file.
3599 * config/nds32/linux.h: New file.
3600 * config/nds32/nds32-elf.opt: New file.
3601 * config/nds32/nds32-linux.opt: New file.
3602 * config/nds32/nds32-fp-as-gp.c
3603 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
3604 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
3606 (nds32_asm_file_end): Ditto.
3607 (nds32_print_operand): Ditto.
3608 (nds32_insert_attributes): Ditto.
3609 (nds32_init_libfuncs): New function.
3610 (TARGET_HAVE_TLS): Define.
3611 (TARGET_INIT_LIBFUNCS): Define.
3612 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
3614 (TARGET_ELF): Apply different mcmodel setting.
3615 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
3616 been migrated into elf.h and linux.h files.
3617 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
3618 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
3619 (mcmodel): The content has been migrated into nds32-elf.opt and
3620 nds32-linux.opt files.
3621 * config/nds32/t-elf: New file.
3622 * config/nds32/t-linux: New file.
3624 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
3625 Shiva Chen <shiva0217@gmail.com>
3627 * config/nds32/constants.md (unspec_volatile_element): Add
3628 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
3629 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
3631 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
3632 (make_pass_nds32_fp_as_gp): Declare.
3633 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
3635 (nds32_asm_function_end_prologue): Remove unused asm output.
3636 (nds32_asm_function_begin_epilogue): Remove unused asm output.
3637 (nds32_asm_file_start): Output necessary fp_as_gp information.
3638 (nds32_option_override): Adjust register usage.
3639 (nds32_expand_prologue): Consider fp_as_gp situation.
3640 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
3641 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
3644 (simple_return): Ditto.
3645 (omit_fp_begin): Output special directive for fp_as_gp.
3646 (omit_fp_end): Output special directive for fp_as_gp.
3647 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
3648 mforbid-fp-as-gp): New options.
3650 2018-06-01 Mark Wielaard <mark@klomp.org>
3652 * dwarf2out.c (dwarf2out_finish): Remove generation of
3653 DW_AT_loclists_base.
3655 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
3657 * gimple-ssa-store-merging.c: Include gimple-fold.h.
3658 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
3659 (struct merged_store_group): Add bit_insertion field.
3660 (dump_char_array): Use standard hexadecimal format.
3661 (merged_store_group::merged_store_group): Set bit_insertion to false.
3662 (merged_store_group::apply_stores): Use optimal buffer size. Deal
3663 with BIT_INSERT_EXPR stores. Move up code updating the mask and
3664 also print the mask in the dump file.
3665 (pass_store_merging::gate): Minor tweak.
3666 (imm_store_chain_info::coalesce_immediate): Fix wrong association
3667 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
3668 stores with INTEGER_CST stores.
3669 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
3670 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
3671 and use it throughout. Generate bit insertion sequences if need be.
3672 (pass_store_merging::process_store): Remove redundant condition.
3673 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
3675 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
3677 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
3678 the 128-bit floating point types. Fix function comment.
3680 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3682 * config/aarch64/aarch64-simd.md
3683 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
3685 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
3688 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
3690 PR tree-optimization/85989
3691 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
3693 (backprop::intersect_uses): Check it when deciding whether this
3694 is a backedge reference.
3695 (backprop::process_block): Add each phi to m_visited_phis
3696 after visiting it, then clear it at the end.
3698 2018-06-01 Richard Biener <rguenther@suse.de>
3700 * tree-vectorizer.h (vect_dr_stmt): New function.
3701 (vect_get_load_cost): Adjust.
3702 (vect_get_store_cost): Likewise.
3703 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
3704 Use vect_dr_stmt instead of DR_SMTT.
3705 (vect_record_base_alignments): Likewise.
3706 (vect_calculate_target_alignment): Likewise.
3707 (vect_compute_data_ref_alignment): Likewise and make static.
3708 (vect_update_misalignment_for_peel): Likewise.
3709 (vect_verify_datarefs_alignment): Likewise.
3710 (vector_alignment_reachable_p): Likewise.
3711 (vect_get_data_access_cost): Likewise. Pass down
3712 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
3713 (vect_get_peeling_costs_all_drs): Likewise.
3714 (vect_peeling_hash_get_lowest_cost): Likewise.
3715 (vect_enhance_data_refs_alignment): Likewise.
3716 (vect_find_same_alignment_drs): Likewise.
3717 (vect_analyze_data_refs_alignment): Likewise.
3718 (vect_analyze_group_access_1): Likewise.
3719 (vect_analyze_group_access): Likewise.
3720 (vect_analyze_data_ref_access): Likewise.
3721 (vect_analyze_data_ref_accesses): Likewise.
3722 (vect_vfa_segment_size): Likewise.
3723 (vect_small_gap_p): Likewise.
3724 (vectorizable_with_step_bound_p): Likewise.
3725 (vect_prune_runtime_alias_test_list): Likewise.
3726 (vect_analyze_data_refs): Likewise.
3727 (vect_supportable_dr_alignment): Likewise.
3728 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
3729 (vect_gen_prolog_loop_niters): Likewise.
3730 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
3731 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
3733 (vect_recog_mask_conversion_pattern): Likewise.
3734 (vect_try_gather_scatter_pattern): Likewise.
3735 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
3736 to vect_get_store_cost.
3737 (vect_get_store_cost): Get stmt_info instead of DR.
3738 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
3739 (vect_get_load_cost): Get stmt_info instead of DR.
3741 2018-06-01 Richard Biener <rguenther@suse.de>
3744 * gimple-fold.c (var_decl_component_p): Also allow offsetted
3745 vars wrapped in MEM_REFs.
3747 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
3749 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
3750 Fix subreg tests so that we only return a choice between
3751 GENERAL_REGS and FP_REGS if the original classes included both.
3753 2018-06-01 Richard Biener <rguenther@suse.de>
3756 * tree-ssa-structalias.c (get_function_part_constraint):
3757 Handle NULL fi->decl.
3758 (find_func_aliases_for_call): Properly handle indirect
3759 fi from direct call.
3760 (find_func_clobbers): Likewise.
3761 (ipa_pta_execute): Likewise.
3762 (create_variable_info_for): For functions that are ifunc_resolver
3763 resolve to a varinfo that contains the result of the resolver call.
3764 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
3767 2018-05-31 Michael Collison <michael.collison@arm.com>
3769 * config/aarch64/aarch64.md:
3770 (*fix_to_zero_extenddfdi2): New pattern.
3771 * gcc.target/aarch64/fix_extend1.c: New testcase.
3773 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
3777 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
3778 and BUILT_IN_STRNCMP_EQ.
3779 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
3780 BUILT_IN_STRNCMP_EQ.
3781 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
3782 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
3783 (gimple_fold_builtin): Likewise.
3784 * tree-ssa-strlen.c (compute_string_length): New function.
3785 (determine_min_obsize): New function.
3786 (handle_builtin_string_cmp): New function to handle calls to
3787 string compare functions.
3788 (strlen_optimize_stmt): Add handling to builtin string compare
3790 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
3791 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
3792 * tree.c (build_common_builtin_nodes): Add new defines of
3793 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
3795 2018-05-31 Jakub Jelinek <jakub@redhat.com>
3798 * bb-reorder.c (pass_partition_blocks::gate): Return false for
3799 functions with naked attribute.
3801 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
3803 * config/i386/sse.md (avx_vec_concat<mode>):
3804 Substitute concat_tg_mode mode attribute with xtg_mode.
3805 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
3806 (concat_tg_mode): Remove mode attribute.
3808 2018-05-31 Martin Sebor <msebor@redhat.com>
3811 * calls.c (alloc_max_size): Correct a logic error/typo.
3812 Treat excessive arguments as infinite. Warn for invalid arguments.
3813 * doc/invoke.texi (-Walloc-size-larger-than): Update.
3815 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
3818 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
3819 and movx for Haswell.
3821 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
3822 Cesar Philippidis <cesar@codesourcery.com>
3825 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
3826 when emitting error on private/firstprivate reductions.
3827 * omp-low.c (lower_omp_target): Avoid reference-type processing
3828 on pointers for firstprivate clause.
3830 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
3832 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
3835 * config/aarch64/aarch64-simd.md
3836 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
3837 (aarch64_ld1_x3_<mode>): Likewise
3838 (aarch64_st1x2<VALLDIF:mode>): Likewise
3839 (aarch64_st1_x2_<mode>): Likewise
3840 (aarch64_st1x3<VALLDIF:mode>): Likewise
3841 (aarch64_st1_x3_<mode>): Likewise
3842 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
3843 (vld1_s8_x3): Likewise.
3844 (vld1_u16_x3): Likewise.
3845 (vld1_s16_x3): Likewise.
3846 (vld1_u32_x3): Likewise.
3847 (vld1_s32_x3): Likewise.
3848 (vld1_u64_x3): Likewise.
3849 (vld1_s64_x3): Likewise.
3850 (vld1_f16_x3): Likewise.
3851 (vld1_f32_x3): Likewise.
3852 (vld1_f64_x3): Likewise.
3853 (vld1_p8_x3): Likewise.
3854 (vld1_p16_x3): Likewise.
3855 (vld1_p64_x3): Likewise.
3856 (vld1q_u8_x3): Likewise.
3857 (vld1q_s8_x3): Likewise.
3858 (vld1q_u16_x3): Likewise.
3859 (vld1q_s16_x3): Likewise.
3860 (vld1q_u32_x3): Likewise.
3861 (vld1q_s32_x3): Likewise.
3862 (vld1q_u64_x3): Likewise.
3863 (vld1q_s64_x3): Likewise.
3864 (vld1q_f16_x3): Likewise.
3865 (vld1q_f32_x3): Likewise.
3866 (vld1q_f64_x3): Likewise.
3867 (vld1q_p8_x3): Likewise.
3868 (vld1q_p16_x3): Likewise.
3869 (vld1q_p64_x3): Likewise.
3870 (vst1_s64_x2): Likewise.
3871 (vst1_u64_x2): Likewise.
3872 (vst1_f64_x2): Likewise.
3873 (vst1_s8_x2): Likewise.
3874 (vst1_p8_x2): Likewise.
3875 (vst1_s16_x2): Likewise.
3876 (vst1_p16_x2): Likewise.
3877 (vst1_s32_x2): Likewise.
3878 (vst1_u8_x2): Likewise.
3879 (vst1_u16_x2): Likewise.
3880 (vst1_u32_x2): Likewise.
3881 (vst1_f16_x2): Likewise.
3882 (vst1_f32_x2): Likewise.
3883 (vst1_p64_x2): Likewise.
3884 (vst1q_s8_x2): Likewise.
3885 (vst1q_p8_x2): Likewise.
3886 (vst1q_s16_x2): Likewise.
3887 (vst1q_p16_x2): Likewise.
3888 (vst1q_s32_x2): Likewise.
3889 (vst1q_s64_x2): Likewise.
3890 (vst1q_u8_x2): Likewise.
3891 (vst1q_u16_x2): Likewise.
3892 (vst1q_u32_x2): Likewise.
3893 (vst1q_u64_x2): Likewise.
3894 (vst1q_f16_x2): Likewise.
3895 (vst1q_f32_x2): Likewise.
3896 (vst1q_f64_x2): Likewise.
3897 (vst1q_p64_x2): Likewise.
3898 (vst1_s64_x3): Likewise.
3899 (vst1_u64_x3): Likewise.
3900 (vst1_f64_x3): Likewise.
3901 (vst1_s8_x3): Likewise.
3902 (vst1_p8_x3): Likewise.
3903 (vst1_s16_x3): Likewise.
3904 (vst1_p16_x3): Likewise.
3905 (vst1_s32_x3): Likewise.
3906 (vst1_u8_x3): Likewise.
3907 (vst1_u16_x3): Likewise.
3908 (vst1_u32_x3): Likewise.
3909 (vst1_f16_x3): Likewise.
3910 (vst1_f32_x3): Likewise.
3911 (vst1_p64_x3): Likewise.
3912 (vst1q_s8_x3): Likewise.
3913 (vst1q_p8_x3): Likewise.
3914 (vst1q_s16_x3): Likewise.
3915 (vst1q_p16_x3): Likewise.
3916 (vst1q_s32_x3): Likewise.
3917 (vst1q_s64_x3): Likewise.
3918 (vst1q_u8_x3): Likewise.
3919 (vst1q_u16_x3): Likewise.
3920 (vst1q_u32_x3): Likewise.
3921 (vst1q_u64_x3): Likewise.
3922 (vst1q_f16_x3): Likewise.
3923 (vst1q_f32_x3): Likewise.
3924 (vst1q_f64_x3): Likewise.
3925 (vst1q_p64_x3): Likewise.
3927 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
3929 * config/msp430/msp430.c (msp430_output_labelref): Prepend
3930 user_label_prefix to name.
3932 * tree-core.h: Update comment about the format of NAME string
3933 passed to handler in attribute_spec.
3935 * config/msp430/msp430.md: Remove erroneous subreg expression from
3936 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
3937 zero_extend{q,h}isi2.
3939 2018-05-30 Borislav Petkov <bp@suse.de>
3941 * doc/extend.texi: Document some architecture specific
3942 constraints and sort entries.
3944 2018-05-30 Martin Sebor <msebor@redhat.com>
3947 * builtins.c (expand_builtin_stpcpy_1): New function.
3948 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
3949 only if the former succeeds.
3951 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
3953 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
3956 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
3958 * doc/invoke.texi (-flinker-output): Document
3960 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
3962 * passes.c (ipa_write_summaries): Only modify statements if body
3964 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
3965 incrementally linking.
3966 (ipa_passes): Likewise.
3967 * lto-cgraph.c (lto_output_node): When incrementally linking do not
3968 pass down resolution info.
3969 * common.opt (flag_incremental_link): Update info.
3970 * gcc.c (plugin specs): Turn flinker-output=* to
3971 -plugin-opt=-linker-output-known
3972 * toplev.c (compile_file): Also cut compilation when doing incremental
3974 * flag-types. (enum lto_partition_model): Add
3975 LTO_LINKER_OUTPUT_NOLTOREL.
3976 (invoke.texi): Add -flinker-output docs.
3977 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
3978 link same way as WPA; do not stream in dead initializers.
3980 * dwarf2out.c (dwarf2out_die_ref_for_decl,
3981 darf2out_register_external_decl): Support incremental link.
3983 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
3985 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
3987 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
3989 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
3990 it down to simple_object_copy_lto_debug_sections.
3991 (run_gcc): Determine incremental LTO link time and configure
3992 lto1 into non-wpa mode, disable renaming of debug sections.
3994 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
3996 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
3997 descriptions of various incorrectly documented functions.
3999 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
4002 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
4003 address check not strict.
4005 2018-05-30 Richard Biener <rguenther@suse.de>
4007 PR tree-optimization/85964
4008 * tracer.c (better_p): Drop initialized count check, we only
4009 call the function with initialized counts now.
4010 (find_best_successor): Do find a best edge if one
4011 has uninitialized count.
4012 (find_best_predecessor): Likewise. Do BB frequency check only
4013 if count is initialized.
4015 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
4017 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
4018 (aarch64_ldrstr_offset_compare): New.
4019 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
4020 load/store orderings.
4021 (aarch64_gen_adjusted_ldpstp): Likewise.
4023 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
4025 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
4026 Check for subset of GENERAL_REGS and FP_REGS.
4027 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
4030 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
4032 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
4033 and wi::to_poly_offset. Add the current offset and then check
4034 whether the sum fits, rather than using an unchecked addition of
4035 a checked term. Check for a shwi rather than a uhwi.
4036 * expr.c (get_bit_range): Use tree_to_poly_uint64.
4037 (store_constructor): Use poly_int_tree_p.
4038 (expand_expr_real_1): Likewise.
4039 * function.c (assign_temp): Likewise.
4040 * fold-const.c (const_binop): Use poly_int_tree_p and
4042 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
4044 * ipa-icf-gimple.c (func_checker::compare_operand): Use
4045 to_poly_offset for MEM offsets.
4046 * ipa-icf.c (sem_variable::equals): Likewise.
4047 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
4048 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
4049 wi::to_poly_offset for BIT_FIELD_REF offsets.
4050 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
4052 * var-tracking.c (emit_note_insn_var_location): Use
4053 tree_to_poly_uint64.
4055 2018-05-29 Jim Wilson <jimw@sifive.com>
4057 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
4059 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
4062 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
4063 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
4065 (sse4_1_round<mode>2): Use nonimmediate_operand
4066 for operand 1 predicate.
4068 2018-05-29 Martin Sebor <msebor@redhat.com>
4069 Richard Biener <rguenther@suse.de>
4072 * calls.c (get_size_range): Call determine_value_range instead
4073 of get_value_range..
4074 * tree-vrp.h (determine_value_range): Declared new function.
4075 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
4077 2018-05-29 Richard Biener <rguenther@suse.de>
4079 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
4080 sure to use non-pattern stmts for get_earlier_stmt arguments.
4081 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
4082 called on pattern stmts.
4083 (get_later_stmt): Likewise.
4085 2018-05-29 Martin Liska <mliska@suse.cz>
4087 PR gcov-profile/85759
4088 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
4091 2018-05-29 Jakub Jelinek <jakub@redhat.com>
4093 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
4095 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
4099 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
4100 VEC_PACK_FLOAT_EXPR): New tree codes.
4101 * tree-pretty-print.c (op_code_prio): Handle
4102 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
4103 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
4104 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
4105 * tree-inline.c (estimate_operator_cost): Likewise.
4106 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
4107 * fold-const.c (const_binop): Likewise.
4108 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
4109 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
4110 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
4111 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
4112 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
4113 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
4114 * expr.c (expand_expr_real_2): Likewise.
4115 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
4116 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
4117 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
4119 * optabs.c (expand_widen_pattern_expr): For
4120 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
4121 sign from result type rather than operand's type.
4122 (expand_binop_directly): For vec_packu_float_optab and
4123 vec_packs_float_optab allow result type to be different from operand's
4125 * optabs-tree.c (optab_for_tree_code): Handle
4126 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
4127 VEC_PACK_FLOAT_EXPR. Formatting fixes.
4128 * tree-vect-generic.c (expand_vector_operations_1): Handle
4129 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
4130 VEC_PACK_FLOAT_EXPR.
4131 * tree-vect-stmts.c (supportable_widening_operation): Handle
4133 (supportable_narrowing_operation): Handle FLOAT_EXPR.
4134 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
4135 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
4136 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
4137 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
4139 (vec_pack<floatprefix>_float_<mode>): New expander.
4140 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
4142 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
4143 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
4144 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
4145 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
4146 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
4148 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
4149 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
4150 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
4151 VEC_PACK_FLOAT_EXPR): Document.
4153 2018-05-29 Richard Biener <rguenther@suse.de>
4155 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
4157 (stmt_vec_info_vec): Make pointer.
4158 (init_stmt_vec_info_vec): Remove.
4159 (free_stmt_vec_info_vec): Likewise.
4160 (set_stmt_vec_info_vec): New function.
4161 (free_stmt_vec_infos): Likewise.
4162 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
4163 (set_vinfo_for_stmt): Likewise.
4164 (get_earlier_stmt): Likewise.
4165 (get_later_stmt): Likewise.
4166 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
4167 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
4168 (vec_info::~vec_info): Free stmt_vec_infos.
4169 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
4170 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
4171 (pass_slp_vectorize::execute): Likewise.
4172 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
4173 (free_stmt_vec_info_vec): Likewise.
4174 (set_stmt_vec_info_vec): New function.
4175 (free_stmt_vec_infos): Likewise.
4176 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
4177 the global stmt_vec_info_vec.
4178 * tree-parloops.c (gather_scalar_reductions): Use
4179 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
4182 2018-05-29 Richard Biener <rguenther@suse.de>
4184 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
4186 2018-05-29 Martin Liska <mliska@suse.cz>
4187 David Malcolm <dmalcolm@redhat.com>
4189 * vec.c (test_reverse): New.
4190 (vec_c_tests): Add new test.
4191 * vec.h (vl_ptr>::reverse): New function.
4193 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
4195 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
4197 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
4200 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
4202 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
4204 2018-05-28 Richard Biener <rguenther@suse.de>
4206 PR tree-optimization/85933
4207 * tree-vect-data-refs.c (vect_record_base_alignments): Only
4208 look at stmts marked as vectorizable.
4210 2018-05-28 Richard Biener <rguenther@suse.de>
4212 PR tree-optimization/85934
4213 * tree-vect-generic.c (expand_vector_operations_1): Hoist
4214 vector boolean check before scalar optimization.
4216 2018-05-28 Jakub Jelinek <jakub@redhat.com>
4218 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
4221 2018-05-28 Mark Wielaard <mark@klomp.org>
4223 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
4224 if it is an expression containing a minus sign.
4226 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
4228 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
4230 2018-05-27 Paul Koning <ni1d@arrl.net>
4232 * config/pdp11/pdp11.md (truncsihi2): Remove.
4234 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
4235 Chung-Ju Wu <jasonwucj@gmail.com>
4237 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
4239 (unaligned_store_dw): Ditto.
4240 * config/nds32/nds32-memory-manipulation.c
4241 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
4242 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
4243 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
4244 (emit_setmem_word_loop): Rename to ...
4245 (emit_setmem_doubleword_loop): ... this.
4246 (nds32_gen_dup_4_byte_to_word_value): New function.
4247 (nds32_gen_dup_8_byte_to_double_word_value): New function.
4248 (nds32_expand_setmem_loop): Refine implementation.
4249 (nds32_expand_setmem_loop_v3m): Ditto.
4250 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
4253 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
4255 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
4257 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
4259 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
4260 (nds32_init_machine_status): Initialize machine->attr_naked_p and
4261 machine->attr_no_prologue_p.
4262 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
4263 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
4264 (nds32_expand_epilogue): Consider attr_naked_p.
4265 (nds32_expand_epilogue_v3pop): Likewise.
4266 (nds32_can_use_return_insn): Likewise.
4267 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
4268 attr_no_prologue_p fields.
4269 * config/nds32/nds32.opt (mret-in-naked-func): New option.
4271 2018-05-27 Jakub Jelinek <jakub@redhat.com>
4274 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
4276 * config/i386/sse.md
4277 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
4279 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
4281 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
4283 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
4285 (*<floatsuffix>floatv2div2sf2): Rename to ...
4286 (*float<floatunssuffix>v2div2sf2): ... this.
4287 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
4288 (float<floatunssuffix>v2div2sf2_mask): ... this.
4289 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
4290 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
4291 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
4293 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
4295 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
4297 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
4299 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
4301 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
4303 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
4304 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
4305 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
4306 gen_ufix_truncv8dfv8si2.
4307 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
4308 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
4309 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
4310 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
4311 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
4312 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
4313 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
4314 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
4316 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
4320 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
4322 2018-05-25 Jim Wilson <jimw@sifive.com>
4324 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
4325 * config/riscv/riscv.c (struct machine_function): Add
4326 interrupt_handler_p and attribute_checked_p fields.
4327 (riscv_attribute_table): Add interrupt.
4328 (riscv_interrupt_type_p): New.
4329 (riscv_save_reg_p): Save extra regs for interrupt handler.
4330 (riscv_use_save_libcall): Return false for interrupt handler.
4331 (riscv_first_stack_step): Add forward declaration.
4332 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
4333 for interrupt handler with large frame. Use it for saved reg list.
4334 (riscv_expand_prologue): Move flag_stack_usage_info support to
4335 eliminate duplication.
4336 (riscv_expand_epilogue): Generate mret for interrupt handler.
4337 (riscv_epilogue_uses): New.
4338 (riscv_can_use_return_insn): Return false for interrupt handler.
4339 (riscv_function_ok_for_sibcall): Likewise.
4340 (riscv_set_current_function): Add interrupt handler support.
4341 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
4342 * config/riscv/riscv.md (UNSPECV_MRET): New.
4344 (riscv_frflags, riscv_fsflags): Use tab after opcode.
4346 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
4348 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
4350 PR tree-optimization/85712
4351 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
4352 this candidate has already been replaced in-situ by a copy.
4354 2018-05-25 Jason Merrill <jason@redhat.com>
4356 PR c++/80485 - inline function non-zero address.
4357 * symtab.c (nonzero_address): Check DECL_COMDAT.
4359 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
4362 * config/alpha/alpha.md (ashlsi3): New insn pattern.
4363 (*ashlsi_se): Rename from *ashldi_se. Define as sign
4364 extension of SImode operation. Use const123_operand predicate.
4365 (*saddsi_1): Remove.
4366 (*saddl_se_1): Ditto.
4368 (*ssubl_se_1): Ditto.
4369 * config/alpha/predicates.md (const123_operand): New predicate.
4370 * config/alpha/constraints.md (P): Use IN_RANGE.
4372 2018-05-25 Richard Biener <rguenther@suse.de>
4374 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
4376 (ref_maybe_used_by_stmt_p): Likewise.
4377 (stmt_may_clobber_ref_p): Likewise.
4378 (stmt_may_clobber_ref_p_1): Likewise.
4379 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
4381 (ref_maybe_used_by_stmt_p): Likewise.
4382 (stmt_may_clobber_ref_p): Likewise.
4383 (stmt_may_clobber_ref_p_1): Likewise.
4384 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
4385 the alias oracle to disambiguate DRs with stmts DR analysis
4387 (vect_analyze_data_refs): Do not give up on not analyzable
4388 DRs for BB vectorization. Remove code truncating the dataref
4391 2018-05-25 Jakub Jelinek <jakub@redhat.com>
4394 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
4395 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
4396 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
4398 2018-05-25 Richard Biener <rguenther@suse.de>
4400 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
4401 function, combining stmt data ref gathering and fatal analysis
4403 (vect_analyze_data_refs): Remove now redudnant code and simplify.
4404 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
4405 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
4406 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
4407 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
4409 2018-05-25 Bin Cheng <bin.cheng@arm.com>
4411 PR tree-optimization/85720
4412 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
4413 SCC if all partitions are builtins.
4414 (version_loop_by_alias_check): New parameter. Generate cancelable
4415 runtime alias check if all partitions are builtins.
4416 (distribute_loop): Update call to above function.
4418 2018-05-25 Bin Cheng <bin.cheng@arm.com>
4420 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
4421 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
4422 (parm_default_def_partition_arg): Ditto.
4423 (set_parm_default_def_partition): Ditto.
4424 (get_parm_default_def_partitions): Ditto and make it static.
4425 (get_undefined_value_partitions): Ditto and make it static.
4426 (remove_ssa_form): Refactor call to init_var_map here.
4427 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
4428 computation for loop region.
4429 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
4430 (register_default_def): Delete.
4431 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
4432 (parm_default_def_partition_arg): Ditto.
4433 (set_parm_default_def_partition): Ditto.
4434 (get_parm_default_def_partitions): Ditto and make it static.
4435 (get_undefined_value_partitions): Ditto and make it static.
4436 (coalesce_with_default, coalesce_with_default): Update comment.
4437 (create_coalesce_list_for_region): New func factored out from
4438 create_outofssa_var_map.
4439 (populate_coalesce_list_for_outofssa): New func factored out from
4440 create_outofssa_var_map and coalesce_ssa_name.
4441 (create_outofssa_var_map): Delete.
4442 (coalesce_ssa_name): Refactor to support live range computation.
4443 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
4444 (get_parm_default_def_partitions): Delete.
4445 (get_undefined_value_partitions): Ditto.
4446 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
4447 computation for loop region.
4448 (new_tree_live_info, loe_visit_block): Ditto.
4449 (live_worklist, set_var_live_on_entry): Ditto.
4450 (calculate_live_on_exit, verify_live_on_entry): Ditto.
4451 * tree-ssa-live.h (struct _var_map): New fields.
4452 (init_var_map): Change decl.
4453 (region_contains_p): New.
4455 2018-05-25 Bin Cheng <bin.cheng@arm.com>
4457 * tree-ssa-live.h (live_merge_and_clear): Delete.
4459 2018-05-25 Richard Biener <rguenther@suse.de>
4462 * tree-dump.c (dequeue_and_dump): Remove access to removed
4463 operand 2 of a SWITCH_EXPR.
4465 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
4467 * doc/sourcebuild.texi (vect_double_cond_arith): Include
4468 multiplication and division.
4469 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
4470 (cond_udiv@var{m}, cond_umod@var{m}): Document.
4471 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
4472 (cond_udiv_optab, cond_umod_optab): New optabs.
4473 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
4474 (IFN_COND_RDIV): New internal functions.
4475 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
4476 TRUNC_MOD_EXPR and RDIV_EXPR.
4477 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
4478 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
4480 (SVE_INT_BINARY): Include mult.
4481 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
4482 (optab, sve_int_op): Handle mult.
4483 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
4485 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
4486 for SVE_INT_BINARY_SD.
4488 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
4490 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
4491 (optab, sve_int_op): Handle div and udiv.
4492 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
4493 for SVE_INT_BINARY_SD.
4494 (*<optab><mode>3): New insn for the same.
4496 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
4498 * tree-vect-patterns.c: Include predict.h.
4499 (vect_recog_divmod_pattern): Restrict check for division support
4500 to when optimizing for size.
4502 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
4504 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
4505 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
4506 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
4507 (gimple_match_op::set_op): Likewise.
4508 (gimple_resimplify4): Declare.
4509 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
4510 (expr::gen_transform): Likewise.
4511 (decision_tree::gen): Generate a simplification routine for 4 operands.
4512 * gimple-match-head.c (gimple_simplify): Add an overload for
4513 4 operands. In the top-level function, handle up to 4 call
4514 arguments and call gimple_resimplify4.
4515 (gimple_resimplify4): New function.
4516 (build_call_internal): Pass a fourth operand.
4517 (maybe_push_to_seq): Likewise.
4518 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
4519 Fold VEC_COND_EXPRs of an operation and a default value into
4520 an IFN_COND_* function if possible.
4521 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
4523 (SVE_COND_FP_BINARY): Include them.
4524 (optab, sve_fp_op): Handle them.
4525 (SVE_INT_BINARY_REV): New code iterator.
4526 (SVE_COND_FP_BINARY_REV): New int iterator.
4527 (commutative): New int attribute.
4528 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
4530 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
4532 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
4533 (*cond_<optab><mode>): New patterns for reversed operands.
4535 2018-05-25 Richard Biener <rguenther@suse.de>
4537 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
4538 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
4539 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
4540 (STMT_VINFO_GROUPED_ACCESS): Adjust.
4541 * tree-vect-data-refs.c (everywhere): Adjust users.
4542 * tree-vect-loop.c (everywhere): Likewise.
4543 * tree-vect-slp.c (everywhere): Likewise.
4544 * tree-vect-stmts.c (everywhere): Likewise.
4545 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
4547 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4549 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
4551 (gcc_cv_as_section_exclude): ... this.
4552 Try Solaris as #exclude syntax.
4553 * configure: Regenerate.
4554 * config.in: Regenerate.
4555 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
4557 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
4558 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
4560 * varasm.c (default_elf_asm_named_section): Don't check if
4561 HAVE_GAS_SECTION_EXCLUDE is defined.
4563 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
4565 * doc/md.texi: Update the documentation of the cond_* optabs
4566 to mention the new final operand. Fix GET_MODE_NUNITS call.
4567 Describe the scalar case too.
4568 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
4569 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
4571 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
4572 (get_conditional_internal_fn): Update comment.
4573 * tree-vect-loop.c (vectorizable_reduction): Pass the original
4574 accumulator value as a final argument to conditional functions.
4575 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
4576 a define_expand and add an "else" operand. Assert for now that
4577 the else operand is equal to operand 2. Use SVE_INT_BINARY and
4578 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
4579 (*cond_<optab><mode>): New patterns.
4580 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
4581 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
4582 (UNSPEC_COND_EOR): Delete.
4583 (optab): Remove associated mappings.
4584 (SVE_INT_BINARY): New code iterator.
4585 (sve_int_op): Remove int attribute and add "minus" to the code
4587 (SVE_COND_INT_OP): Delete.
4588 (SVE_COND_FP_OP): Rename to...
4589 (SVE_COND_FP_BINARY): ...this.
4591 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
4593 * optabs.c (can_reuse_operands_p): New function.
4594 (maybe_legitimize_operands): Try to reuse the results for
4597 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
4599 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
4600 Add {q} suffix to insn mnemonic.
4602 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
4604 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
4605 (msp430_warn_func_return): New.
4607 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
4609 * fold-const.c (tree_nonzero_bits): New function.
4610 * fold-const.h (tree_nonzero_bits): Likewise.
4611 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
4612 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
4614 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
4618 * varasm.c (assemble_alias): Check ifunc_resolver only on
4621 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
4624 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
4625 when memory input operand is handled.
4627 2018-05-24 Luis Machado <luis.machado@linaro.org>
4629 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
4631 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
4633 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
4635 * match.pd: Delay FMA folds until after vectorization.
4637 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
4640 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
4641 address check not strict.
4643 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
4645 * gimple-match.h (gimple_match_op): New class.
4646 (mprts_hook): Replace parameters with a gimple_match_op *.
4647 (maybe_build_generic_op): Likewise.
4648 (gimple_simplified_result_is_gimple_val): Replace parameters with
4649 a const gimple_match_op *.
4650 (gimple_simplify): Replace code_helper * and tree * parameters with
4651 a gimple_match_op * parameter.
4652 (gimple_resimplify1): Replace code_helper *, tree and tree *
4653 parameters with a gimple_match_op * parameter.
4654 (gimple_resimplify2): Likewise.
4655 (gimple_resimplify3): Likewise.
4656 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
4657 parameters with a gimple_match_op * parameter.
4658 * gimple-match-head.c (gimple_simplify): Change prototypes of
4659 auto-generated functions to take a gimple_match_op * instead of
4660 separate code_helper * and tree * parameters. Make the same
4661 change in the top-level overload and update calls to the
4662 gimple_resimplify routines. Update calls to the auto-generated
4663 functions and to maybe_push_res_to_seq in the publicly-facing
4664 operation-specific gimple_simplify overloads.
4665 (gimple_match_op::MAX_NUM_OPS): Define.
4666 (gimple_resimplify1): Replace rcode and ops with a single res_op
4667 parameter. Update call to gimple_simplify.
4668 (gimple_resimplify2): Likewise.
4669 (gimple_resimplify3): Likewise.
4670 (mprts_hook): Replace parameters with a gimple_match_op *.
4671 (maybe_build_generic_op): Likewise.
4672 (build_call_internal): Replace type, nargs and ops with
4673 a gimple_match_op *.
4674 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
4675 with a single gimple_match_op *. Update calls to mprts_hook,
4676 build_call_internal and gimple_simplified_result_is_gimple_val.
4677 Factor out code that is common to the tree_code and combined_fn cases.
4678 * genmatch.c (expr::gen_transform): Replace tem_code and
4679 tem_ops with a gimple_match_op called tem_op. Update calls
4680 to the gimple_resimplify functions and maybe_push_res_to_seq.
4681 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
4682 res_ops. Update call to the gimple_resimplify functions.
4683 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
4684 (decision_tree::gen): Make the functions take a gimple_match_op *
4685 called res_op instead of separate res_code and res_ops parameters.
4686 Update call accordingly.
4687 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
4688 and ops with a single res_op parameter. Update calls to
4689 maybe_build_generic_op and maybe_push_res_to_seq.
4690 (fold_stmt_1): Update calls to gimple_simplify and
4691 replace_stmt_with_simplification.
4692 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
4693 and gimple_simplified_result_is_gimple_val.
4694 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
4696 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
4697 with a gimple_match_op *.
4698 (vn_nary_build_or_lookup): Likewise. Update call to
4699 vn_nary_build_or_lookup_1.
4700 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
4701 gimple_match_op *. Update calls to the gimple_resimplify routines
4702 and to gimple_simplified_result_is_gimple_val.
4703 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
4704 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
4705 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
4706 (visit_nary_op): Likewise.
4707 (visit_reference_op_load): Likewise.
4709 2018-05-23 Luis Machado <luis.machado@linaro.org>
4711 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
4712 modifier for printing the step amount.
4714 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
4717 * gcc/tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
4720 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
4722 * doc/sourcebuild.texi (Endianness): New subsubsection.
4724 2018-05-23 Luis Machado <luis.machado@linaro.org>
4726 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
4727 <prefetch_dynamic_strides>: New const bool field.
4728 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
4729 prefetch_dynamic_strides.
4730 (exynosm1_prefetch_tune): Likewise.
4731 (thunderxt88_prefetch_tune): Likewise.
4732 (thunderx_prefetch_tune): Likewise.
4733 (thunderx2t99_prefetch_tune): Likewise.
4734 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
4736 (aarch64_override_options_internal): Update to set
4737 PARAM_PREFETCH_DYNAMIC_STRIDES.
4738 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
4739 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
4740 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
4741 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
4742 prefetch-dynamic-strides setting.
4744 2018-05-23 Luis Machado <luis.machado@linaro.org>
4746 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
4747 <minimum_stride>: New const int field.
4748 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
4749 minimum_stride field defaulting to -1.
4750 (exynosm1_prefetch_tune): Likewise.
4751 (thunderxt88_prefetch_tune): Likewise.
4752 (thunderx_prefetch_tune): Likewise.
4753 (thunderx2t99_prefetch_tune): Likewise.
4754 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
4755 <default_opt_level>: Set to 3.
4756 (aarch64_override_options_internal): Update to set
4757 PARAM_PREFETCH_MINIMUM_STRIDE.
4758 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
4759 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
4760 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
4761 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
4762 stride is constant and is below the minimum stride threshold.
4764 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4766 * config/arm/arm-cpus.in (mode26): Delete.
4767 (armv4): Delete mode26 reference.
4768 * config/arm/arm.c (arm_configure_build_target): Delete use of
4771 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
4773 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
4775 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
4776 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
4778 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
4779 (floatunsdidf2): Ditto.
4781 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
4783 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
4784 (fixuns_trunc<mode>si2_avx512f): Ditto.
4785 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
4786 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
4787 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
4789 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
4791 PR rtl-optimization/79985
4792 * df-scan.c (df_insn_refs_collect): Remove special case for
4793 global registers and asm statements.
4795 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
4797 * extend.texi (Global Register Variables): Rewrite the bullet list.
4798 Note that the register is available for allocation. Note that access
4799 via inline asm must use constraints. Add note about async-signal
4800 handlers. Remove paragraph about automagic register selection.
4802 2018-05-23 Richard Biener <rguenther@suse.de>
4804 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
4805 of fixed offset from memset VN.
4807 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
4809 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
4811 (alloc_cand_and_find_basis): Initialize first_interp field.
4812 (slsr_process_mul): Modify first_interp field.
4813 (slsr_process_add): Likewise.
4814 (slsr_process_cast): Modify first_interp field for each new
4816 (slsr_process_copy): Likewise.
4817 (dump_candidate): Dump first_interp field.
4818 (replace_mult_candidate): Process all interpretations, not just
4820 (replace_rhs_if_not_dup): Likewise.
4821 (replace_one_candidate): Likewise.
4823 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
4825 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
4827 (aarch64_needs_frame_chain): New function.
4828 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
4830 2018-05-23 Sudakshina Das <sudi.das@arm.com>
4833 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
4834 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
4835 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
4836 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
4837 as true for strict-align.
4838 (aarch64_can_inline_p): Perform checks even when callee has no
4839 attributes to check for strict alignment.
4840 * doc/extend.texi (AArch64 Function Attributes): Document
4842 * doc/invoke.texi: (AArch64 Options): Likewise.
4844 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
4846 PR tree-optimization/85853
4847 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
4848 the handling of the root of the node to...
4849 (vect_slp_analyze_node_operations_1): ...this new function,
4850 and run the whole thing with the child nodes' def types
4851 set according to their SLP node's def type.
4853 2018-05-23 Richard Biener <rguenther@suse.de>
4856 * tree-data-ref.c (create_runtime_alias_checks): Defer
4857 and ignore overflow warnings.
4859 2018-05-23 Yury Gribov <tetra2005@gmail.com>
4861 PR tree-optimization/85822
4862 * tree-vrp.c (is_masked_range_test): Fix handling of negative
4865 2018-05-23 Richard Biener <rguenther@suse.de>
4867 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
4868 memset constants via native_interpret_expr.
4870 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
4873 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
4875 (cgraph_node::create_alias): Likewise.
4876 (cgraph_node::get_availability): Check ifunc_resolver instead
4877 of looking up ifunc attribute.
4878 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
4879 * varasm.c (do_assemble_alias): Likewise.
4880 (assemble_alias): Likewise.
4881 (default_binds_local_p_3): Likewise.
4882 * cgraph.h (cgraph_node): Add ifunc_resolver.
4883 (cgraph_node::only_called_directly_or_aliased_p): Return false
4885 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
4887 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
4888 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
4889 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
4890 instead of looking up ifunc attribute.
4892 2018-05-22 Luis Machado <luis.machado@linaro.org>
4894 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
4896 2018-05-22 Martin Sebor <msebor@redhat.com>
4899 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
4900 only when expasion succeeds.
4901 (expand_builtin_strcmp): Same.
4902 (expand_builtin_strncmp): Same.
4904 2018-05-22 Martin Sebor <msebor@redhat.com>
4906 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
4908 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
4909 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4911 * config/aarch64/aarch64-ldpstp.md: Replace uses of
4912 aarch64_mem_pair_operand with memory_operand and delete operand swapping
4914 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
4915 Add check for legitimate_address.
4916 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
4917 (aarch64_swap_ldrstr_operands): New.
4918 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
4921 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
4922 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4924 * config/aarch64/aarch64.md: New patterns to generate stp
4926 (store_pair_sw, store_pair_dw): New patterns to generate stp for
4927 single words and double words.
4928 (load_pair_sw, load_pair_dw): Likewise.
4929 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
4931 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
4933 * config/aarch64/aarch64-ldpstp.md: Modify peephole
4934 for different mode ldpstp and add peephole for merged zero stores.
4936 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
4938 (aarch64_gen_store_pair): Rename calls to match new patterns.
4939 (aarch64_gen_load_pair): Rename calls to match new patterns.
4940 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
4941 (load_pair<DREG:mode><DREG2:mode>): ... This.
4942 (store_pair<mode>): Rename to...
4943 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
4944 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
4946 (V_INT_EQUIV): Handle SImode.
4947 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
4950 2018-05-22 Martin Sebor <msebor@redhat.com>
4953 * calls.c (maybe_warn_nonstring_arg): Use string length to set
4954 or ajust the presumed bound on an operation to avoid unnecessary
4957 2018-05-22 Martin Sebor <msebor@redhat.com>
4959 PR tree-optimization/85826
4960 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
4961 assuming that a DECL necesarily has a constant size.
4963 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
4966 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
4968 2018-05-22 Richard Biener <rguenther@suse.de>
4970 PR tree-optimization/85834
4971 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
4972 non-constant and non-zero memset arguments.
4974 2018-05-22 Martin Liska <mliska@suse.cz>
4977 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
4979 2018-05-22 Richard Biener <rguenther@suse.de>
4981 PR tree-optimization/85863
4982 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
4983 comparisons when vectype is specified.
4984 (vectorizable_condition): Do not specify vectype for
4985 vect_is_simple_cond when SLP vectorizing.
4987 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
4990 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
4991 define __ibm128 as long double.
4992 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
4993 as a distinct type when IEEE 128-bit support is enabled.
4994 (init_float128_ieee): Fix up conversions between IFmode and IEEE
4995 128-bit types to use the correct functions.
4996 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
4997 convert between 128-bit floating point types that have different
4998 modes but the same representation, instead of using gen_lowpart to
5000 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
5002 (IFKF_reg): New attributes to give the register constraints for
5004 (extend<mode>tf2_internal): New insns to mark an explicit
5005 conversion between 128-bit floating point types that have a
5006 different mode but share the same representation.
5008 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
5010 PR tree-optimization/85814
5011 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
5012 a null return from get_strinfo when unsharing the next
5013 strinfo in the chain.
5015 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
5018 * varasm.c (weak_finish): Clean up weak_decls.
5020 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5022 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
5023 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
5024 UNSPEC_UADALP values.
5025 * config/aarch64/iterators.md (ABAL): New int iterator.
5028 (sur): Add mappings for the above.
5029 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
5031 (aarch64_<sur>abal<mode>_4): Likewise.
5032 (aarch64_<sur>adalp<mode>_3): Likewise.
5033 (<sur>sadv16qi): New define_expand.
5035 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
5037 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
5038 (*movdf_internal): Ditto.
5039 (*rcpsf2_sse): Ditto.
5040 (*rsqrtsf2_sse): Ditto.
5041 (*sqrt<mode>2_sse): Ditto.
5043 2018-05-21 Tamar Christina <tamar.christina@arm.com>
5045 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
5047 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
5048 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
5049 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
5050 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
5052 * config/aarch64/arm_neon.h: Likewise.
5053 * config/aarch64/iterators.md (VQ_I): New.
5055 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
5057 * config.gcc: Add arc/t-multilib-linux to tmake_file for
5059 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
5062 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
5064 * config/nds32/constraints.md (S): New constraint.
5065 * config/nds32/nds32.md (call_internal): Use constraint S.
5066 (call_value_internal): Likewise.
5067 (sibcall_internal): Likewise.
5068 (sibcall_value_internal): Likewise.
5070 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
5071 Chung-Ju Wu <jasonwucj@gmail.com>
5073 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
5076 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
5077 Chung-Ju Wu <jasonwucj@gmail.com>
5079 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
5080 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
5081 (nds32_rtx_costs_impl): Simplify.
5082 (nds32_address_cost_impl): Simplify.
5083 (nds32_init_rtx_costs): New function.
5084 (nds32_rtx_costs_speed_prefer): Likewise.
5085 (nds32_rtx_costs_size_prefer): Likewise.
5086 (nds32_address_cost_speed_prefer): Likewise.
5087 (nds32_address_cost_speed_fwprop): Likewise.
5088 (nds32_address_cost_size_prefer): Likewise.
5089 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
5090 * config/nds32/nds32.c (nds32_option_override): Use
5091 nds32_init_rtx_costs function.
5093 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
5095 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
5096 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
5097 (TARGET_PIPELINE_N8): Likewise.
5098 (TARGET_PIPELINE_N10): Likewise.
5099 (TARGET_PIPELINE_N13): Likewise.
5100 (TARGET_PIPELINE_GRAYWOLF): Likewise.
5102 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
5104 * config/nds32/nds32-fpu.md: Update copyright year.
5106 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
5108 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
5110 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
5112 * config/nds32/nds32.c
5113 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
5114 * config/nds32/nds32.opt (minline-asm-r15): New option.
5116 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
5118 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
5120 * config/nds32/nds32.md (abssi2): New pattern.
5122 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
5124 * config/i386/i386.md (rex64namesuffix): New mode attribute.
5125 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
5126 Merge insn pattern from sse_cvtsi2ss<round_name> and
5127 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
5128 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
5129 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
5130 using SWI48 mode iterator.
5131 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
5132 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
5133 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
5134 pattern from sse_cvttss2si<round_saeonly_name>
5135 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
5136 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
5137 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
5138 using SWI48 mode iterator.
5139 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
5140 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
5141 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
5142 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
5143 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
5144 using SWI48 mode iterator.
5145 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
5146 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
5147 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
5148 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
5149 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
5150 SWI48 mode iterator.
5151 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
5152 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
5153 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
5154 pattern from sse_cvttsd2si<round_saeonly_name>
5155 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
5157 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
5159 * config/nds32/nds32-md-auxiliary.c
5160 (nds32_valid_smw_lwm_base_p): Refine.
5161 (nds32_output_smw_single_word): Refine.
5162 (nds32_output_smw_double_word): New.
5163 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
5165 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
5167 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
5168 (nds32_output_stack_pop): Refine.
5169 (nds32_expand_unaligned_load): Refine.
5170 (nds32_expand_unaligned_store): Refine.
5172 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
5173 Chung-Ju Wu <jasonwucj@gmail.com>
5175 * config/nds32/constants.md: Add TP_REGNUM constant.
5176 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
5177 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
5179 * config/nds32/nds32-doubleword.md: Consider flag_pic.
5180 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
5181 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
5182 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
5183 and PIC code generation.
5184 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
5186 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
5188 * config/nds32/nds32.md: Support TLS and PIC.
5189 * config/nds32/nds32.c: Support TLS and PIC.
5190 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
5191 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
5194 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
5196 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
5197 mode with E_ prefix.
5199 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
5200 Chung-Ju Wu <jasonwucj@gmail.com>
5202 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
5203 * config/nds32/nds32-md-auxiliary.c
5204 (symbolic_reference_mentioned_p): New.
5205 (nds32_legitimize_ict_address): New.
5206 (nds32_expand_ict_move): New.
5207 (nds32_indirect_call_referenced_p): New.
5208 (nds32_symbol_binds_local_p): Delete.
5209 (nds32_long_call_p): Modify.
5210 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
5211 * config/nds32/nds32-protos.h
5212 (symbolic_reference_mentioned_p): Declare.
5213 (nds32_legitimize_ict_address): Declare.
5214 (nds32_expand_ict_move): Declare.
5215 (nds32_indirect_call_referenced_p): Declare.
5216 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
5217 (nds32_relax_group): Use nds32_ict_const_p as condition.
5218 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
5219 (nds32_asm_file_start): Output ict_model directive in asm code.
5220 (nds32_legitimate_address_p): Consider indirect call.
5221 (nds32_print_operand): Consider indirect call.
5222 (nds32_print_operand_address): Consider indirect call.
5223 (nds32_insert_attributes): Handle "indirect_call" attribute.
5224 (TARGET_LEGITIMATE_ADDRESS_P): Define.
5225 (TARGET_LEGITIMATE_CONSTANT_P): Define.
5226 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
5227 (TARGET_DELEGITIMIZE_ADDRESS): Define.
5228 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
5229 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
5230 (TARGET_ICT_MODEL_SMALL): Define.
5231 (TARGET_ICT_MODEL_LARGE): Define.
5232 * config/nds32/nds32.md (movsi): Consider ict model.
5233 (call, call_value): Consider ict model.
5234 (sibcall, sibcall_value): Consider ict model.
5235 * config/nds32/nds32.opt (mict-model): New option.
5236 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
5239 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
5240 Monk Chiang <sh.chiang04@gmail.com>
5241 Jim Wilson <jimw@sifive.com>
5243 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
5244 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
5245 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
5246 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
5247 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
5248 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
5249 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
5250 compute save_libcall_adjustment properly.
5251 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
5252 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
5253 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
5254 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
5255 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
5256 (ABI_SPEC): Handle mabi=ilp32e.
5257 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
5258 (RVE): Add RVE mask.
5259 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
5260 <-march>: Add rv32e as an example.
5262 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
5265 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
5266 (intra_create_variable_infos): Handle C++ constructors.
5268 2018-05-18 Martin Liska <mliska@suse.cz>
5270 * passes.def: Remove a redundant pass.
5272 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
5275 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
5277 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5279 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
5281 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
5283 (armv2, armv2a, armv3, armv3m): Delete architectures.
5284 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
5285 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
5286 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
5288 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
5289 (*mulsidi3adddi): Likewise.
5290 (mulsidi3): Likewise.
5291 (*mulsidi3_nov6): Likewise.
5292 (umulsidi3): Likewise.
5293 (umulsidi3_nov6): Likewise.
5294 (umaddsidi4): Likewise.
5295 (*umulsidi3adddi): Likewise.
5296 (smulsi3_highpart): Likewise.
5297 (*smulsi3_highpart_nov6): Likewise.
5298 (umulsi3_highpart): Likewise.
5299 (*umulsi3_highpart_nov6): Likewise.
5300 * config/arm/arm.h (arm_arch3m): Delete.
5301 * config/arm/arm.c (arm_arch3m): Delete.
5302 (arm_option_override_internal): Update armv3-related comment.
5303 (arm_configure_build_target): Delete use of isa_bit_mode32.
5304 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
5305 (arm_rtx_costs_internal): Delete check of arm_arch3m.
5306 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
5308 (mulusa3): Likewise.
5309 * config/arm/arm-protos.h (arm_arch3m): Delete.
5310 * config/arm/arm-tables.opt: Regenerate.
5311 * config/arm/arm-tune.md: Likewise.
5312 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
5313 deleted architectures.
5315 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5317 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
5318 (armv5t, armv5te): New features.
5319 (ARMv5, ARMv5e): Delete fgroups.
5320 (ARMv5t, ARMv5te): Adjust for above changes.
5322 (armv5, armv5e): Delete arches.
5323 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
5325 (*call_reg_arm): Likewise.
5326 (*call_value_reg_armv5): Likewise.
5327 (*call_value_reg_arm): Likewise.
5328 (*call_symbol): Likewise.
5329 (*call_value_symbol): Likewise.
5330 (*sibcall_insn): Likewise.
5331 (*sibcall_value_insn): Likewise.
5333 (prefetch): Likewise.
5334 (define_split and define_peephole2 dependent on arm_arch5):
5336 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
5338 (TARGET_ARM_QBIT): Likewise.
5339 (TARGET_DSP_MULTIPLY): Likewise.
5340 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
5341 (arm_arch5, arm_arch5e): Delete.
5342 (arm_arch5t, arm_arch5te): Declare.
5343 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
5344 (arm_arch5t): Declare.
5345 (arm_option_reconfigure_globals): Update for the above.
5346 (arm_options_perform_arch_sanity_checks): Update comment, replace
5347 use of arm_arch5 with arm_arch5t.
5348 (use_return_insn): Likewise.
5349 (arm_emit_call_insn): Likewise.
5350 (output_return_instruction): Likewise.
5351 (arm_final_prescan_insn): Likewise.
5352 (arm_coproc_builtin_available): Likewise.
5353 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
5354 arm_arch5e with arm_arch5t and arm_arch5te.
5355 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
5356 (arm_arch5t, arm_arch5te): Declare.
5357 * config/arm/arm-tables.opt: Regenerate.
5358 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
5359 * config/arm/t-multilib: Likewise.
5360 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
5361 instead of arm_arch5.
5362 (*call_reg_thumb1): Likewise.
5363 (*call_value_reg_thumb1_v5): Likewise.
5364 (*call_value_reg_thumb1): Likewise.
5365 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
5367 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
5369 2018-05-18 Martin Liska <mliska@suse.cz>
5371 PR gcov-profile/84846
5372 * doc/gcov.texi: Document -t option of gcov tool.
5374 2018-05-18 Martin Liska <mliska@suse.cz>
5376 PR gcov-profile/84846
5377 * gcov.c (print_usage): Add new -t option.
5378 (process_args): Handle the option.
5379 (generate_results): Use stdout as output when requested by
5382 2018-05-18 Martin Liska <mliska@suse.cz>
5384 PR gcov-profile/84846
5385 * coverage.c (coverage_init): Write PWD to .gcno file.
5386 * doc/gcov.texi: Document how working directory is printed.
5387 * gcov-dump.c (dump_gcov_file): Print PWD.
5388 * gcov.c (output_intermediate_file): Likewise.
5389 (read_graph_file): Read PWD string.
5390 (output_lines): Print PWD.
5392 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5395 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
5396 for retval and return false if all args to phi are zero.
5398 2018-05-18 Richard Biener <rguenther@suse.de>
5400 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
5402 (evrp_dom_walker::before_dom_children): Call it.
5404 2018-05-18 Richard Biener <rguenther@suse.de>
5406 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
5407 results when processing array refs with variable index.
5409 2018-05-18 Toon Moene <toon@moene.org>
5411 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
5412 directly after that of -floop-interchange. Indicate that both
5413 options are enabled by default when specifying -O3.
5415 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5417 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
5418 iterator. Delete separate integer-mode vec_set<mode> expander.
5419 (aarch64_simd_vec_setv2di): Delete.
5420 (vec_setv2di): Delete.
5421 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
5422 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
5423 the "w, r" alternative.
5425 2018-05-18 Martin Liska <mliska@suse.cz>
5427 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
5428 * tree-pass.h (make_pass_lower_switch_O0): New function.
5429 * tree-switch-conversion.c (node_has_low_bound): Remove.
5430 (node_has_high_bound): Likewise.
5431 (node_is_bounded): Likewise.
5432 (class pass_lower_switch): Make it a template type and create
5434 (pass_lower_switch::execute): Add template argument.
5435 (make_pass_lower_switch): New function.
5436 (make_pass_lower_switch_O0): New function.
5437 (do_jump_if_equal): Remove.
5438 (emit_case_nodes): Simplify to just handle all 3 cases and leave
5439 all the hard work to tree optimization passes.
5441 2018-05-18 Martin Liska <mliska@suse.cz>
5443 * dbgcnt.c (limit_low): Renamed from limit.
5444 (limit_high): New variable.
5445 (dbg_cnt_is_enabled): Check for upper limit.
5446 (dbg_cnt): Adjust dumping.
5447 (dbg_cnt_set_limit_by_index): Add new argument for high
5449 (dbg_cnt_set_limit_by_name): Likewise.
5450 (dbg_cnt_process_single_pair): Parse new format.
5451 (dbg_cnt_process_opt): Use strtok.
5452 (dbg_cnt_list_all_counters): Remove 'value' and add
5454 * doc/invoke.texi: Document changes.
5456 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
5458 * doc/sourcebuild.texi (scalar_all_fma): Document.
5459 * tree.def (FMA_EXPR): Delete.
5460 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
5461 * internal-fn.c (ternary_direct): New macro.
5462 (expand_ternary_optab_fn): Likewise.
5463 (direct_ternary_optab_supported_p): Likewise.
5464 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
5465 * builtins.c (fold_builtin_fma): Delete.
5466 (fold_builtin_3): Don't call it.
5467 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
5468 * expr.c (expand_expr_real_2): Likewise.
5469 * fold-const.c (operand_equal_p): Likewise.
5470 (fold_ternary_loc): Likewise.
5471 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
5472 * gimple.c (DEFTREECODE): Likewise.
5473 * gimplify.c (gimplify_expr): Likewise.
5474 * optabs-tree.c (optab_for_tree_code): Likewise.
5475 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
5476 * tree-eh.c (operation_could_trap_p): Likewise.
5477 (stmt_could_throw_1_p): Likewise.
5478 * tree-inline.c (estimate_operator_cost): Likewise.
5479 * tree-pretty-print.c (dump_generic_node): Likewise.
5480 (op_code_prio): Likewise.
5481 * tree-ssa-loop-im.c (stmt_cost): Likewise.
5482 * tree-ssa-operands.c (get_expr_operands): Likewise.
5483 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
5484 * fold-const-call.h (fold_fma): Delete.
5485 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
5486 CFN_FNMA and CFN_FNMS.
5488 * genmatch.c (combined_fn): New enum.
5489 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
5490 (commutative_op): New function.
5491 (commutate): Use it. Handle more than 2 operands.
5492 (dt_operand::gen_gimple_expr): Use commutative_op.
5493 (parser::parse_expr): Allow :c to be used with non-binary
5494 operators if the commutative operand is known.
5495 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
5496 CFN_FMS, CFN_FNMA and CFN_FNMS.
5497 (backprop::process_assign_use): Remove FMA_EXPR handling.
5498 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
5499 (gen_hsa_fma): New function.
5500 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
5501 IFN_FNMA and IFN_FNMS.
5502 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
5503 * gimple-fold.h (follow_all_ssa_edges): Declare.
5504 * gimple-fold.c (follow_all_ssa_edges): New function.
5505 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
5506 gimple_build interface and use follow_all_ssa_edges to fold the result.
5507 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
5508 instead of checking for optabs directly.
5509 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
5510 rather than FMA_EXPRs.
5511 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
5512 call to IFN_FMA instead of an FMA_EXPR.
5514 2018-05-17 Jim Wilson <jimw@sifive.com>
5516 * expr.c (do_tablejump): When converting index to Pmode, if we have a
5517 sign extended promoted subreg, and the range does not have the sign bit
5518 set, then do a sign extend.
5520 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
5521 test, check for sign extended subreg and/or constant operands, and
5522 do a sign extend in that case.
5524 2018-05-17 Steve Ellcey <sellcey@cavium.com>
5526 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
5527 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
5529 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
5530 Change logics_shift_reg to logics_shift_imm.
5531 (thunderx2t99_fp_loadpair_basic): Delete.
5532 (thunderx2t99_fp_storepair_basic): Delete.
5533 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
5534 (thunderx2t99_asimd_polynomial): Delete.
5535 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
5536 and neon_fp_mul_d_scalar_q.
5537 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
5538 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
5539 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
5540 (thunderx2t99_asimd_lut): Add missing tbl types.
5541 (thunderx2t99_asimd_ext): Delete.
5542 (thunderx2t99_asimd_load1_1_mult): Delete.
5543 (thunderx2t99_asimd_load1_2_mult): Delete.
5544 (thunderx2t99_asimd_load1_ldp): New.
5545 (thunderx2t99_asimd_load1): New.
5546 (thunderx2t99_asimd_load2): Add missing *load2* types.
5547 (thunderx2t99_asimd_load3): New.
5548 (thunderx2t99_asimd_load4): New.
5549 (thunderx2t99_asimd_store1_1_mult): Delete.
5550 (thunderx2t99_asimd_store1_2_mult): Delete.
5551 (thunderx2t99_asimd_store2_mult): Delete.
5552 (thunderx2t99_asimd_store2_onelane): Delete.
5553 (thunderx2t99_asimd_store_stp): New.
5554 (thunderx2t99_asimd_store1): New.
5555 (thunderx2t99_asimd_store2): New.
5556 (thunderx2t99_asimd_store3): New.
5557 (thunderx2t99_asimd_store4): New.
5559 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
5561 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
5562 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
5564 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
5565 Segher Boessenkool <segher@kernel.crashing.org>
5568 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
5571 2018-05-17 Richard Biener <rguenther@suse.de>
5573 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
5574 for pruning loop and prune defs feeding only already visited PHIs.
5576 2018-05-17 Richard Biener <rguenther@suse.de>
5578 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
5580 2018-05-17 Bin Cheng <bin.cheng@arm.com>
5581 Richard Biener <rguenther@suse.de>
5583 PR tree-optimization/85793
5584 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
5585 for VMAT_ELEMENTWISE.
5587 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
5589 * internal-fn.h (lookup_internal_fn): Declare
5590 * internal-fn.c (lookup_internal_fn): New function.
5591 * gimple.c (gimple_build_call_from_tree): Handle calls to
5593 * gimple-pretty-print.c (dump_gimple_call): Print "." before
5594 internal function names.
5595 * tree-pretty-print.c (dump_generic_node): Likewise.
5596 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
5598 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
5600 * gimple-fold.h (gimple_build): Make the function forms take
5601 combined_fn rather than built_in_function.
5602 (gimple_simplify): Likewise.
5603 * gimple-match-head.c (gimple_simplify): Likewise.
5604 * gimple-fold.c (gimple_build): Likewise.
5605 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
5606 rather than gimple_build_call_internal.
5607 (get_initial_defs_for_reduction): Likewise.
5608 (vect_create_epilog_for_reduction): Likewise.
5609 (vectorizable_live_operation): Likewise.
5611 2018-05-17 Martin Liska <mliska@suse.cz>
5613 * gimple-ssa-sprintf.c (format_directive): Do not use
5614 space in between 'G_' and '('.
5616 2018-05-17 Jakub Jelinek <jakub@redhat.com>
5619 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
5620 even if the mask is not all ones.
5623 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
5625 (ix86_gimple_fold_builtin): Likewise.
5628 * config/i386/i386.c: Include tree-vector-builder.h.
5629 (ix86_vector_shift_count): New function.
5630 (ix86_fold_builtin): Fold shift builtins by scalar count.
5631 (ix86_gimple_fold_builtin): Likewise.
5633 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
5634 _mm512_setzero): New intrinsics.
5636 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
5637 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5639 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
5640 code generation for cases where splatting a value is not useful.
5641 * simplify-rtx.c (simplify_ternary_operation): Simplify
5642 vec_merge across a vec_duplicate and a paradoxical subreg forming
5643 a vector mode to a vec_concat.
5645 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
5647 * config.gcc: Support "goldmont-plus".
5648 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
5650 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
5651 PROCESSOR_GOLDMONT_PLUS.
5652 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
5653 (processor_target_table): Add "goldmont-plus".
5654 (PTA_GOLDMONT_PLUS): Define.
5655 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
5656 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
5657 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
5658 (fold_builtin_cpu): Add "goldmont-plus".
5659 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
5660 (ix86_option_override_internal): Add "goldmont-plus".
5661 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
5662 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
5663 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
5664 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
5666 2018-05-17 Richard Biener <rguenther@suse.de>
5668 PR tree-optimization/85757
5669 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
5670 remove defs that only feed that PHI from further processing.
5672 2018-05-16 Jim Wilson <jimw@sifive.com>
5674 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
5676 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
5678 2018-05-16 Mark Wielaard <mark@klomp.org>
5680 * dwarf2out.c (count_index_strings): New function.
5681 (output_indirect_strings): Call count_index_strings and generate
5682 header for dwarf_version >= 5.
5684 2018-05-16 Mark Wielaard <mark@klomp.org>
5686 * dwarf2out.c (dwarf_FORM): New function.
5687 (set_indirect_string): Use dwarf_FORM.
5688 (reset_indirect_string): Likewise.
5689 (size_of_die): Likewise.
5690 (value_format): Likewise.
5691 (output_die): Likewise.
5692 (add_skeleton_AT_string): Likewise.
5693 (output_macinfo_op): Likewise.
5694 (index_string): Likewise.
5695 (output_index_string_offset): Likewise.
5696 (output_index_string): Likewise.
5697 (count_index_strings): Likewise.
5699 2018-05-16 Carl Love <cel@us.ibm.com>
5701 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
5702 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
5704 2018-05-16 Martin Jambor <mjambor@suse.cz>
5706 * ipa-prop.c (ipa_free_all_edge_args): Remove.
5707 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
5709 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
5711 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
5712 (fnma<mode>4): Likewise.
5713 (fms<mode>4): Likewise.
5714 (fnms<mode>4): Likewise.
5715 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
5716 (aarch64_fnma<mode>4): Likewise.
5717 (aarch64_fms<mode>4): Likewise.
5718 (aarch64_fnms<mode>4): Likewise.
5719 (aarch64_fnmadd<mode>4): Likewise.
5721 2018-05-16 Jason Merrill <jason@redhat.com>
5723 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
5725 2018-05-16 Richard Biener <rguenther@suse.de>
5727 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
5728 (dump_stmt_cost): Declare.
5729 (add_stmt_cost): Dump cost we add.
5730 (add_stmt_costs): New function.
5731 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
5733 (vect_analyze_stmt): Adjust prototype.
5734 (vectorizable_condition): Likewise.
5735 (vectorizable_live_operation): Likewise.
5736 (vectorizable_reduction): Likewise.
5737 (vectorizable_induction): Likewise.
5738 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
5739 cost vector to pass to vectorizable_ and record afterwards.
5740 (vect_model_reduction_cost): Take cost vector argument and adjust.
5741 (vect_model_induction_cost): Likewise.
5742 (vectorizable_reduction): Likewise.
5743 (vectorizable_induction): Likewise.
5744 (vectorizable_live_operation): Likewise.
5745 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
5746 SLP_TREE_NUMBER_OF_VEC_STMTS.
5747 (vect_analyze_slp_cost_1): Remove.
5748 (vect_analyze_slp_cost): Likewise.
5749 (vect_slp_analyze_node_operations): Take visited args and
5750 a target cost vector. Avoid processing already visited stmt sets.
5751 (vect_slp_analyze_operations): Use a local cost vector to gather
5752 costs and register those of non-discarded instances.
5753 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
5754 (vect_schedule_slp_instance): Remove copying of
5755 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
5757 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
5758 adding cost. Record cost entry location.
5759 (vect_prologue_cost_for_slp_op): Function to compute cost of
5760 a constant or invariant generated for SLP vect in the prologue,
5761 split out from vect_analyze_slp_cost_1.
5762 (vect_model_simple_cost): Make static. Adjust for SLP costing.
5763 (vect_model_promotion_demotion_cost): Likewise.
5764 (vect_model_store_cost): Likewise, make static.
5765 (vect_model_load_cost): Likewise.
5766 (vectorizable_bswap): Add cost vector arg and adjust.
5767 (vectorizable_call): Likewise.
5768 (vectorizable_simd_clone_call): Likewise.
5769 (vectorizable_conversion): Likewise.
5770 (vectorizable_assignment): Likewise.
5771 (vectorizable_shift): Likewise.
5772 (vectorizable_operation): Likewise.
5773 (vectorizable_store): Likewise.
5774 (vectorizable_load): Likewise.
5775 (vectorizable_condition): Likewise.
5776 (vectorizable_comparison): Likewise.
5777 (can_vectorize_live_stmts): Likewise.
5778 (vect_analyze_stmt): Likewise.
5779 (vect_transform_stmt): Adjust calls to vectorizable_*.
5780 * tree-vectorizer.c: Include gimple-pretty-print.h.
5781 (dump_stmt_cost): New function.
5783 2018-05-16 Richard Biener <rguenther@suse.de>
5785 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
5786 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
5787 * tree-ssa-dse.c: Include tree-ssa-loop.h.
5788 (check_name): New callback.
5789 (dse_classify_store): Track cycles via a visited bitmap of PHI
5790 defs and simplify handling of in-loop and across loop dead stores
5791 and properly fail for loop-variant refs. Handle byte-tracking with
5792 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
5795 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
5797 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
5798 (vect_get_mask_type_for_stmt): Likewise.
5799 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
5801 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
5802 to determine the statement's vector type and the vector type that
5803 should be used for calculating nunits. Deal with cases in which
5804 the type has to be deferred.
5805 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
5806 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
5807 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
5808 (vect_determine_vf_for_stmt): New functions, split out from...
5809 (vect_determine_vectorization_factor): ...here.
5810 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
5811 (vect_get_mask_type_for_stmt): New functions, split out from
5812 vect_determine_vectorization_factor.
5814 2018-05-16 Richard Biener <rguenther@suse.de>
5816 * tree-cfg.c (verify_gimple_assign_ternary): Properly
5817 verify the [VEC_]COND_EXPR embedded comparison.
5819 2018-05-15 Martin Sebor <msebor@redhat.com>
5821 PR tree-optimization/85753
5822 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
5823 RECORD_TYPE in addition to ARRAY_TYPE.
5825 2018-05-15 Martin Sebor <msebor@redhat.com>
5828 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
5830 2018-05-15 Richard Biener <rguenther@suse.de>
5832 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
5833 add by_clobber_p one. Change algorithm to collect all defs
5834 representing uses we need to walk and try reducing them to
5835 a single one before failing.
5836 (dse_dom_walker::dse_optimize_stmt): Adjust.
5838 2018-05-13 Mark Wielaard <mark@klomp.org>
5840 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
5841 (size_of_loc_descr): Likewise.
5842 (output_loc_operands): Likewise.
5843 (output_loc_operands_raw): Likewise.
5844 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
5845 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
5846 (hash_loc_operands): Likewise.
5847 (compare_loc_operands): Likewise.
5849 2018-05-14 Mark Wielaard <mark@klomp.org>
5851 * dwarf2out.c (count_index_addrs): New function.
5852 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
5854 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5856 PR tree-optimization/83648
5857 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
5858 return value as malloc candidate.
5860 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5863 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
5864 param as true in call to suggest_attribute.
5866 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
5868 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
5869 -mreadonly-in-sdata.
5871 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5873 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
5875 (aarch64_crypto_aesd_fused): Likewise.
5877 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
5879 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
5880 (movsi_aarch64): Likewise.
5881 (load_pairsi): Likewise.
5882 (load_pairdi): Likewise.
5883 (store_pairsi): Likewise.
5884 (store_pairdi): Likewise.
5885 (load_pairsf): Likewise.
5886 (load_pairdf): Likewise.
5887 (store_pairsf): Likewise.
5888 (store_pairdf): Likewise.
5889 (zero_extend): Likewise.
5890 (trunc): Swap alternatives.
5891 (fcvt_target): Add '?' to prefer w over r.
5893 2018-05-14 Jakub Jelinek <jakub@redhat.com>
5896 * config/i386/i386.md: Disallow non-commutative arithmetics in
5897 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
5898 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
5899 in the peephole2 before it.
5901 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
5903 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
5904 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
5905 (ix86_handle_option): Handle -mcldemote.
5906 * config.gcc: New header.
5907 * config/i386/cldemoteintrin.h: New file.
5908 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
5909 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
5911 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
5912 OPTION_MASK_ISA_CLDEMOTE.
5913 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
5914 (ix86_valid_target_attribute_inner_p): Ditto.
5915 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
5916 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
5917 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
5918 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
5919 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
5921 * config/i386/i386.opt: Add -mcldemote.
5922 * config/i386/x86intrin.h: New header.
5923 * doc/invoke.texi: Add -mcldemote.
5925 2018-05-14 Richard Biener <rguenther@suse.de>
5927 * doc/match-and-simplify.texi: Adjust :s documentation.
5929 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
5931 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
5932 intended memcpy size.
5933 (REORDER_45): Likewise.
5935 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
5937 * sort.cc: New file.
5938 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
5939 * vec.c (qsort_chk): Use gcc_qsort.
5940 * Makefile.in (OBJS-libcommon): Add sort.o.
5941 (build/sort.o): New target. Use it...
5942 (BUILD_RTL): ... here, and...
5943 (build/gencfn-macros): ... here, and...
5944 (build/genmatch): ... here.
5946 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
5947 Chung-Ju Wu <jasonwucj@gmail.com>
5949 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
5950 * config/nds32/nds32-graywolf.md: New file.
5951 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
5952 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
5954 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
5955 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
5956 * config/nds32/nds32.md (pipeline_model): Add graywolf.
5957 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
5958 * config/nds32/pipelines.md: Include n15 settings.
5960 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
5961 Chung-Ju Wu <jasonwucj@gmail.com>
5963 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
5964 * config/nds32/nds32-n13.md: New file.
5965 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
5966 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
5968 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
5969 * config/nds32/nds32.md (pipeline_model): Add n13.
5970 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
5971 * config/nds32/pipelines.md: Include n13 settings.
5973 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
5974 Chung-Ju Wu <jasonwucj@gmail.com>
5976 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
5977 * config/nds32/nds32-n10.md: New file.
5978 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
5979 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
5981 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
5982 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
5983 * config/nds32/nds32.md (pipeline_model): Add n10.
5984 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
5985 * config/nds32/pipelines.md: Include n10 settings.
5987 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
5988 Kito Cheng <kito.cheng@gmail.com>
5989 Chung-Ju Wu <jasonwucj@gmail.com>
5991 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
5992 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
5993 Add enum values for DSP extension instructions.
5994 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
5996 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
5997 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
5999 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
6000 * config/nds32/nds32-dspext.md: New file for DSP implementation.
6001 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
6002 * config/nds32/nds32-intrinsic.md: Likewise.
6003 * config/nds32/nds32_intrinsic.h: Likewise.
6004 * config/nds32/nds32-md-auxiliary.c: Likewise.
6005 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
6006 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
6007 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
6008 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
6009 * config/nds32/nds32-protos.h: New declarations for DSP extension.
6010 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
6011 TYPE_DMAC in switch statement.
6012 * config/nds32/nds32.c: New checking and implementation for DSP
6013 extension instructions.
6014 * config/nds32/nds32.h: Likewise.
6015 * config/nds32/nds32.md: Likewise.
6016 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
6017 * config/nds32/predicates.md: Implement new predicates for DSP
6020 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
6022 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
6023 Reformat alternatives and attributes so it is easier to identify
6024 which constraints/attributes go with which instruction.
6025 (mov<mode>_hardfloat32, FMOVE64): Likewise.
6026 (mov<mode>_softfloat32, FMOVE64): Likewise.
6027 (mov<mode>_hardfloat64, FMOVE64): Likewise.
6028 (mov<mode>_softfloat64, FMOVE64): Likewise.
6030 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
6032 * doc/extend.texi (PowerPC Built-in Functions): Rename this
6034 (Basic PowerPC Built-in Functions): The new name of the
6035 subsection previously known as "PowerPC Built-in Functions".
6036 (Basic PowerPC Built-in Functions Available on all Configurations):
6038 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
6039 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
6040 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
6041 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
6043 2018-05-11 Martin Jambor <mjambor@suse.cz>
6046 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
6049 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
6052 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
6054 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
6056 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
6057 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
6058 (ix86_handle_option): Handle -mwaitpkg.
6059 * config.gcc: New header.
6060 * config/i386/cpuid.h (bit_WAITPKG): New bit.
6061 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
6062 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
6064 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
6065 OPTION_MASK_ISA_WAITPKG.
6066 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
6067 (ix86_option_override_internal): Add PTA_WAITPKG.
6068 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
6069 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
6070 IX86_BUILTIN_TPAUSE.
6071 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
6072 __builtin_ia32_umwait and __builtin_ia32_tpause.
6073 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
6074 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
6075 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
6076 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
6077 UNSPECV_TPAUSE): New.
6078 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
6079 * config/i386/i386.opt: Add -mwaitpkg.
6080 * config/i386/waitpkgintrin.h: New file.
6081 * config/i386/x86intrin.h: New header.
6082 * doc/invoke.texi: Add -mwaitpkg.
6084 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
6087 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
6089 (cortex-m0): Use armv6s-m isa.
6090 (cortex-m0plus): Likewise.
6091 (cortex-m1): Likewise.
6092 (cortex-m0.small-multiply): Likewise.
6093 (cortex-m0plus.small-multiply): Likewise.
6094 (cortex-m1.small-multiply): Likewise.
6096 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
6097 Jakub Jelinek <jakub@redhat.com>
6099 PR tree-optimization/85692
6100 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
6101 source permute as well.
6103 2018-05-11 Martin Liska <mliska@suse.cz>
6106 * doc/extend.texi: Document LLVM style format for no_sanitize
6109 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
6111 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
6112 mode_supports_vsx_dform_quad to mode_supports_dq_form.
6113 (mode_supports_vsx_dform_quad): Likewise.
6114 (mode_supports_vmx_dform): Move these functions to be next to the
6115 other mode_supports functions.
6116 (mode_supports_dq_form): Likewise.
6117 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
6118 mode_supports_dq_form.
6119 (reg_offset_addressing_ok_p): Likewise.
6120 (offsettable_ok_by_alignment): Likewise.
6121 (rs6000_legitimate_offset_address_p): Likewise.
6122 (legitimate_lo_sum_address_p): Likewise.
6123 (rs6000_legitimize_address): Likewise.
6124 (rs6000_legitimize_reload_address): Likewise.
6125 (rs6000_secondary_reload_inner): Likewise.
6126 (rs6000_preferred_reload_class): Likewise.
6127 (rs6000_output_move_128bit): Likewise.
6129 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
6131 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
6132 Generate SImode target register for null target.
6133 <case IX86_BUILTIN_XGETBV>: Ditto.
6134 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
6135 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
6137 2018-05-10 Carl Love <cel@us.ibm.com>
6139 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
6140 dcbtt and dcbtstt if operands[2] is 0.
6142 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
6145 * config/i386/sse.md (usadv64qi): New expander.
6147 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
6149 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
6150 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
6151 -maltivec=be support.
6152 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
6153 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
6154 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
6155 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
6156 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
6157 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
6158 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
6159 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
6160 altivec_vsumsws): Adjust.
6161 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
6162 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
6163 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
6165 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
6166 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
6167 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
6168 (altivec_lve<VI_char>x): Delete expand.
6169 (*altivec_lve<VI_char>x_internal): Rename to...
6170 (altivec_lve<VI_char>x): ... this.
6171 (altivec_lvxl_<mode>): Delete expand.
6172 (*altivec_lvxl_<mode>_internal): Rename to ...
6173 (altivec_lvxl_<mode>): ... this.
6174 (altivec_stvxl_<mode>): Delete expand.
6175 (*altivec_stvxl_<mode>_internal): Rename to ...
6176 (altivec_stvxl_<mode>): ... this.
6177 (altivec_stve<VI_char>x): Delete expand.
6178 (*altivec_stve<VI_char>x_internal): Rename to ...
6179 (altivec_stve<VI_char>x): ... this.
6180 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
6181 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
6182 reduc_plus_scal_<mode>): Adjust.
6183 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
6185 (rs6000_cpu_cpp_builtins): Adjust.
6186 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
6187 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
6188 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
6189 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
6190 -maltivec=be support.
6191 (rs6000_split_vec_extract_var): Adjust.
6192 (rs6000_split_v4si_init): Adjust.
6193 (swap_selector_for_mode): Delete.
6194 (altivec_expand_lvx_be, altivec_expand_stvx_be,
6195 altivec_expand_stvex_be): Delete.
6196 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
6197 -maltivec=be support.
6198 (rs6000_gimple_fold_builtin): Ditto.
6199 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
6201 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
6202 (TARGET_DIRECT_MOVE_64BIT): Adjust.
6203 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
6204 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
6205 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
6206 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
6207 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
6208 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
6209 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
6210 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
6211 anonymous split): Adjust.
6212 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
6213 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
6215 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
6217 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
6218 when --with-gxx-include-dir is also specified.
6219 * configure: Regenerate.
6221 2018-05-09 Jim Wilson <jimw@sifive.com>
6224 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
6225 * config/riscv/t-withmultilib: New.
6226 * config/riscv/withmultilib.h: New.
6227 * doc/install.texi: Document RISC-V --with-multilib-list support.
6229 2018-05-09 Richard Biener <rguenther@suse.de>
6231 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
6233 (vect_bb_vectorization_profitable_p): Adjust. Compute
6234 actual scalar cost using the cost vector and the add_stmt_cost
6237 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
6239 PR rtl-optimization/85645
6240 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
6241 in the REG_CFA_REGISTER note for LR, don't leave it empty.
6243 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
6245 PR rtl-optimization/85645
6246 * shrink-wrap.c (spread_components): Return a boolean saying if
6247 anything was changed.
6248 (try_shrink_wrapping_separate): Iterate spread_components until
6249 nothing changes anymore.
6251 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
6253 PR rtl-optimization/85645
6254 * regrename.c (build_def_use): Also kill the chains that include the
6255 destination of a REG_CFA_REGISTER note.
6257 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
6259 PR rtl-optimization/85645
6260 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
6261 insn that has a REG_CFA_REGISTER note.
6263 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
6265 * cfgexpand.c (expand_clobber): New function.
6266 (expand_gimple_stmt_1): Use it.
6267 * tree-vect-stmts.c (vect_clobber_variable): New function,
6269 (vectorizable_simd_clone_call): ...here.
6270 (vectorizable_store): Emit a clobber either side of an
6271 IFN_STORE_LANES sequence.
6272 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
6274 2018-05-09 Tom de Vries <tom@codesourcery.com>
6277 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
6278 (define_insn "trap_if_false"): Add exit after trap.
6280 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
6282 PR rtl-optimization/85638
6283 * bb-reorder.c: Include common/common-target.h.
6284 (create_forwarder_block): New function extracted from...
6285 (fix_up_crossing_landing_pad): ...here. Rename into...
6286 (dw2_fix_up_crossing_landing_pad): ...this.
6287 (sjlj_fix_up_crossing_landing_pad): New function.
6288 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
6289 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
6290 from both partitions and exit the loop after one iteration.
6292 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
6295 * doc/extend.texi (PowerPC Built-in Functions): Rename this
6297 (Basic PowerPC Built-in Functions): The new name of the
6298 subsection previously known as "PowerPC Built-in Functions".
6299 (Basic PowerPC Built-in Functions Available on all Configurations):
6301 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
6303 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
6305 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
6307 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
6310 2018-05-08 Jim Wilson <jimw@sifive.com>
6312 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
6313 (LD_EMUL_SUFFIX): New.
6314 (LINK_SPEC): Use it.
6316 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
6318 * doc/extend.texi (PowerPC Built-in Functions): Rename this
6320 (Basic PowerPC Built-in Functions): The new name of the
6321 subsection previously known as "PowerPC Built-in Functions".
6322 (Basic PowerPC Built-in Functions Available on all Configurations):
6324 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
6326 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
6328 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
6330 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
6333 2018-05-08 Jakub Jelinek <jakub@redhat.com>
6336 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
6337 after cmpelim optimization.
6339 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
6341 * config.gcc: Support "goldmont".
6342 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
6343 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
6345 * config/i386/i386.c (m_GOLDMONT): Define.
6346 (processor_target_table): Add "goldmont".
6347 (PTA_GOLDMONT): Define.
6348 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
6349 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
6350 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
6351 (fold_builtin_cpu): Add "goldmont".
6352 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
6353 (ix86_option_override_internal): Add "goldmont".
6354 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
6355 (processor_type): Add PROCESSOR_GOLDMONT.
6356 * config/i386/i386.md: Add CPU "glm".
6357 * config/i386/glm.md: New file.
6358 * config/i386/x86-tune.def: Add m_GOLDMONT.
6359 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
6361 2018-05-08 Jakub Jelinek <jakub@redhat.com>
6364 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
6366 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
6367 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
6368 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
6371 * config/i386/i386.c (ix86_fold_builtin): Handle
6372 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
6375 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
6376 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
6378 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
6381 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
6382 (check_arch): Likewise.
6383 (check_fpu): Return the result rather than printing it.
6384 (end arch): Fix operator precedence.
6385 (end cpu): Likewise.
6386 (END): Print the result from check_fpu.
6388 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
6389 Alan Hayward <alan.hayward@arm.com>
6390 David Sherwood <david.sherwood@arm.com>
6392 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
6393 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
6394 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
6395 (*fcmuo<mode>_and): New patterns.
6397 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
6399 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
6400 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
6401 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
6402 (cmp_op, sve_imm_con): New code attributes.
6403 (SVE_COND_INT_CMP, imm_con): Delete.
6404 (cmp_op): Remove above unspecs from int attribute.
6405 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
6407 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
6408 comparison-specific unspecs.
6409 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
6410 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
6411 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
6412 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
6413 (*vec_fcm<cmp_op><mode>): Rename to...
6414 (*fcm<cmp_op><mode>): ...this and adjust likewise.
6415 (*vec_fcmuo<mode>): Rename to...
6416 (*fcmuo<mode>): ...this and adjust likewise.
6417 (*pred_fcm<cmp_op><mode>): New pattern.
6418 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
6419 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
6421 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
6423 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
6424 (aarch64_emit_sve_predicated_cond): New function.
6425 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
6426 (aarch64_emit_unspec_cond_or): Replace with...
6427 (aarch64_emit_sve_or_conds): ...this new function. Use
6428 aarch64_emit_sve_ptrue_op for the individual comparisons and
6429 aarch64_emit_binop to OR them together.
6430 (aarch64_emit_inverted_unspec_cond): Replace with...
6431 (aarch64_emit_sve_inverted_cond): ...this new function. Use
6432 aarch64_emit_sve_ptrue_op for the comparison and
6433 aarch64_emit_unop to invert the result.
6434 (aarch64_expand_sve_vec_cmp_float): Update after the above
6435 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
6437 2018-05-07 Nathan Sidwell <nathan@acm.org>
6439 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
6440 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
6441 (Backwards Compatibility): Likewise.
6443 2018-05-07 Luis Machado <luis.machado@linaro.org>
6447 2018-05-07 Luis Machado <luis.machado@linaro.org>
6449 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
6450 <prefetch_dynamic_strides>: New const bool field.
6451 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
6452 prefetch_dynamic_strides.
6453 (exynosm1_prefetch_tune): Likewise.
6454 (thunderxt88_prefetch_tune): Likewise.
6455 (thunderx_prefetch_tune): Likewise.
6456 (thunderx2t99_prefetch_tune): Likewise.
6457 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
6459 (aarch64_override_options_internal): Update to set
6460 PARAM_PREFETCH_DYNAMIC_STRIDES.
6461 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
6462 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
6463 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
6464 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
6465 prefetch-dynamic-strides setting.
6467 2018-05-07 Luis Machado <luis.machado@linaro.org>
6469 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
6470 <minimum_stride>: New const int field.
6471 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
6472 minimum_stride field.
6473 (exynosm1_prefetch_tune): Likewise.
6474 (thunderxt88_prefetch_tune): Likewise.
6475 (thunderx_prefetch_tune): Likewise.
6476 (thunderx2t99_prefetch_tune): Likewise.
6477 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
6478 (aarch64_override_options_internal): Update to set
6479 PARAM_PREFETCH_MINIMUM_STRIDE.
6480 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
6481 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
6482 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
6483 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
6484 stride is constant and is below the minimum stride threshold.
6486 2018-05-07 Luis Machado <luis.machado@linaro.org>
6488 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
6491 2018-05-07 Luis Machado <luis.machado@linaro.org>
6493 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
6494 <prefetch_dynamic_strides>: New const bool field.
6495 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
6496 prefetch_dynamic_strides.
6497 (exynosm1_prefetch_tune): Likewise.
6498 (thunderxt88_prefetch_tune): Likewise.
6499 (thunderx_prefetch_tune): Likewise.
6500 (thunderx2t99_prefetch_tune): Likewise.
6501 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
6503 (aarch64_override_options_internal): Update to set
6504 PARAM_PREFETCH_DYNAMIC_STRIDES.
6505 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
6506 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
6507 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
6508 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
6509 prefetch-dynamic-strides setting.
6511 2018-05-07 Luis Machado <luis.machado@linaro.org>
6513 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
6514 <minimum_stride>: New const int field.
6515 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
6516 minimum_stride field.
6517 (exynosm1_prefetch_tune): Likewise.
6518 (thunderxt88_prefetch_tune): Likewise.
6519 (thunderx_prefetch_tune): Likewise.
6520 (thunderx2t99_prefetch_tune): Likewise.
6521 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
6522 (aarch64_override_options_internal): Update to set
6523 PARAM_PREFETCH_MINIMUM_STRIDE.
6524 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
6525 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
6526 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
6527 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
6528 stride is constant and is below the minimum stride threshold.
6530 2018-05-06 Jakub Jelinek <jakub@redhat.com>
6533 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
6534 the type is addressable. Don't force op into register if it has
6537 2018-05-05 Roland McGrath <mcgrathr@google.com>
6540 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
6541 any section for which we don't know a specific type it should have,
6542 regardless of name. Previously this was done only for the exact
6543 names ".init_array", ".fini_array", and ".preinit_array".
6544 (default_elf_asm_named_section): Add comment about
6545 relationship with default_section_type_flags and SECTION_NOTYPE.
6546 (get_section): Don't consider it a type conflict if one side has
6547 SECTION_NOTYPE and the other doesn't, as long as neither has the
6548 SECTION_BSS et al used in the default_section_type_flags logic.
6550 2018-05-05 Tom de Vries <tom@codesourcery.com>
6553 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
6554 (workaround_barsyncs): New function.
6555 (nvptx_reorg): Use workaround_barsyncs.
6556 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
6557 (define_expand "nvptx_membar_cta"): New define_expand.
6558 (define_insn "*nvptx_membar_cta"): New insn.
6560 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
6562 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
6563 To improve optimization opportunities.
6564 * builtin-types.def: The new needed builtin types for the above.
6566 2018-05-04 Richard Biener <rguenther@suse.de>
6568 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
6569 * gimple-ssa-store-merging.c
6570 (imm_store_chain_info::output_merged_store): Remove redundant create,
6571 release split_store vector contents on failure.
6572 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
6573 scalar stmt vector on cache hit.
6575 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
6577 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
6579 * config.gcc (powerpc-xilinx-eabi*): Remove.
6580 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
6582 (fusion_addis_mem_combo_load): Ditto.
6583 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
6585 (rs6000_cpu_cpp_builtins): Ditto.
6586 * config/rs6000/rs6000-linux.c
6587 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
6588 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
6589 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
6591 (rs6000_setup_reg_addr_masks): Ditto.
6592 (rs6000_init_hard_regno_mode_ok): Ditto.
6593 (rs6000_option_override_internal): Ditto.
6594 (legitimate_lo_sum_address_p): Ditto.
6595 (rs6000_legitimize_address): Ditto.
6596 (rs6000_legitimize_reload_address): Ditto.
6597 (rs6000_legitimate_address_p): Ditto.
6598 (abi_v4_pass_in_fpr): Ditto.
6599 (setup_incoming_varargs): Ditto.
6600 (rs6000_gimplify_va_arg): Ditto.
6601 (rs6000_split_multireg_move): Ditto.
6602 (rs6000_savres_strategy): Ditto.
6603 (rs6000_emit_prologue_components): Ditto.
6604 (rs6000_emit_epilogue_components): Ditto.
6605 (rs6000_emit_prologue): Ditto.
6606 (rs6000_emit_epilogue): Ditto.
6607 (rs6000_elf_file_end): Ditto.
6608 (rs6000_function_value): Ditto.
6609 (rs6000_libcall_value): Ditto.
6610 * config/rs6000/rs6000.h: Ditto.
6611 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
6612 (TARGET_MINMAX): ... this. New.
6613 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
6614 * config/rs6000/rs6000.md: Remove Xilinx FP support.
6615 (*movsi_internal1_single): Delete.
6616 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
6617 mfpu=, mxilinx-fpu): Delete.
6618 * config/rs6000/singlefp.h: Delete.
6619 * config/rs6000/sysv4.h: Remove Xilinx FP support.
6620 * config/rs6000/t-rs6000: Ditto.
6621 * config/rs6000/t-xilinx: Delete.
6622 * gcc/config/rs6000/titan.md: Adjust for fp_type removal.
6623 * gcc/config/rs6000/vsx.md: Remove Xilinx FP support.
6624 (VStype_simple): Delete.
6625 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
6626 * config/rs6000/xfpu.h: Delete.
6627 * config/rs6000/xfpu.md: Delete.
6628 * config/rs6000/xilinx.h: Delete.
6629 * config/rs6000/xilinx.opt: Delete.
6630 * gcc/doc/invoke.texi (RS/6000 and PowerPC Options): Remove
6631 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
6633 2018-05-04 Tom de Vries <tom@codesourcery.com>
6636 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
6639 2018-05-04 Richard Biener <rguenther@suse.de>
6642 * tree-complex.c (update_complex_assignment): We are always in SSA form.
6643 (expand_complex_div_wide): Likewise.
6644 (expand_complex_operations_1): Likewise.
6645 (expand_complex_libcall): Preserve EH info of the original stmt.
6646 (tree_lower_complex): Handle removed blocks.
6647 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
6648 on complex multiplication and division libcall builtins.
6650 2018-05-04 Richard Biener <rguenther@suse.de>
6653 * fold-const.c (negate_expr_p): Restrict negation of operand
6654 zero of a division to when we know that can happen without
6656 (fold_negate_expr_1): Likewise.
6658 2018-05-04 Jakub Jelinek <jakub@redhat.com>
6661 * real.h (real_nextafter): Declare.
6662 * real.c (real_nextafter): New function.
6663 * fold-const-call.c (fold_const_nextafter): New function.
6664 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
6665 CASE_CFN_NEXTTOWARD.
6666 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
6667 even when arg1_mode is different from arg0_mode.
6669 2018-05-03 Nathan Sidwell <nathan@acm.org>
6671 * doc/extend.texi (Deprecated Features): Remove
6673 (Backwards Compatibility): Likewise.
6674 * doc/invoke.texi (C++ Language Options): Likewise.
6675 (C++ Dialect Options): Likewise.
6677 2018-05-03 Jakub Jelinek <jakub@redhat.com>
6680 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
6681 _mm512_mask_mullox_epi64): New intrinsics.
6683 2018-05-03 Tom de Vries <tom@codesourcery.com>
6686 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
6687 dump files): Add offload-tree.
6689 2018-05-03 Richard Biener <rguenther@suse.de>
6691 PR tree-optimization/85615
6692 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
6693 to loops not nested in BBs loop father to avoid creating multi-entry
6696 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6698 PR tree-optimization/70291
6699 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
6700 arguments. Change return type to tree. Emit libcall as a new
6701 statement rather than replacing existing one when inplace_p is true.
6702 (expand_complex_multiplication_components): New function.
6703 (expand_complex_multiplication): Expand floating-point complex
6704 multiplication using the above.
6705 (expand_complex_division): Rename inner_type parameter to type.
6706 Update expand_complex_libcall call-site.
6707 (expand_complex_operations_1): Update expand_complex_multiplication
6708 and expand_complex_division call-sites.
6710 2018-05-02 Jakub Jelinek <jakub@redhat.com>
6713 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
6714 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
6715 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
6716 the highest significant bit of the shift count mask is clear. In
6717 check whether and[sq]i3 is needed verify that all significant bits
6718 of the shift count other than the highest are set.
6720 2018-05-02 Tom de Vries <tom@codesourcery.com>
6723 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
6724 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
6725 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
6726 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
6727 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
6728 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
6729 __builtin_goacc_parlevel_size.
6731 2018-05-02 Richard Biener <rguenther@suse.de>
6733 PR tree-optimization/85597
6734 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
6735 do not use split vect_get_vec_defs call but call vect_get_slp_defs
6738 2018-05-02 Tom de Vries <tom@codesourcery.com>
6741 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
6742 dump files): Add ltrans-tree.
6744 2018-05-02 Tom de Vries <tom@codesourcery.com>
6747 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
6748 dump files): Add wpa-ipa.
6750 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
6752 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
6753 powerpc*-*-linux*paired* target.
6754 * config/rs6000/750cl.h: Delete.
6755 * config/rs6000/paired.h: Delete.
6756 * config/rs6000/paired.md: Delete.
6757 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
6759 * config/rs6000/rs6000-builtin.def: Remove paired float support.
6760 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
6761 comment. Remove paired float support.
6762 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
6763 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
6765 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
6766 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
6768 * config/rs6000/rs6000.c: Remove paired float support.
6769 (paired_expand_vector_init, paired_expand_vector_move,
6770 paired_emit_vector_compare, paired_emit_vector_cond_expr,
6771 (paired_expand_lv_builtin, paired_expand_stv_builtin,
6772 paired_expand_builtin, paired_expand_predicate_builtin,
6773 paired_init_builtins): Delete.
6774 * config/rs6000/rs6000.h: Remove paired float support.
6775 * config/rs6000/rs6000.md: Remove paired float support.
6776 (move_from_CR_ov_bit): Delete.
6777 * config/rs6000/rs6000.opt (mpaired): Delete.
6778 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
6779 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
6781 2018-05-02 Richard Biener <rguenther@suse.de>
6784 * gimplify.c (gimplify_save_expr): When in SSA form allow
6785 SAVE_EXPRs to compute to SSA vars.
6787 2018-05-02 Jakub Jelinek <jakub@redhat.com>
6790 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
6791 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
6792 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
6793 clobber operands[2], instead use a new pseudo. Formatting fixes.
6795 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
6797 PR tree-optimization/85586
6798 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
6799 exit early for statements in the same group if the accesses are
6802 2018-05-02 Tom de Vries <tom@codesourcery.com>
6805 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
6808 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
6810 PR tree-optimization/85143
6811 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
6813 2018-05-01 Tom de Vries <tom@codesourcery.com>
6816 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
6817 not found" error message.
6819 2018-05-01 Tom de Vries <tom@codesourcery.com>
6822 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
6823 * vec.c (test_ordered_remove_if): New function.
6824 (vec_c_tests): Call test_ordered_remove_if.
6825 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
6826 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
6827 * tree-vect-patterns.c (vect_pattern_recog_1): Use
6828 VEC_ORDERED_REMOVE_IF.
6830 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6832 PR tree-optimization/82665
6833 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
6834 pointer subtraction where arguments come from a memchr call.
6836 2018-05-01 Jakub Jelinek <jakub@redhat.com>
6838 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
6839 --push-state --as-needed and --pop-state instead of --as-needed and
6840 --no-as-needed if ld supports it.
6841 * configure: Regenerated.
6844 * doc/install.texi2html: Replace _002d with - and _002a with * in
6845 generated html files using sed.
6847 2018-04-30 David Malcolm <dmalcolm@redhat.com>
6850 * gcc-rich-location.c (blank_line_before_p): New function.
6851 (use_new_line): New function.
6852 (gcc_rich_location::add_fixit_insert_formatted): New function.
6853 * gcc-rich-location.h
6854 (gcc_rich_location::add_fixit_insert_formatted): New function.
6856 2018-04-30 David Malcolm <dmalcolm@redhat.com>
6858 * selftest.c (assert_streq): Rename "expected" and "actual" to
6859 "val1" and "val2". Extend NULL-handling to cover both inputs
6860 symmetrically, while still requiring both to be non-NULL for a pass.
6861 * selftest.h (assert_streq): Rename "expected" and "actual" to
6863 (ASSERT_EQ): Likewise.
6864 (ASSERT_EQ_AT): Likewise.
6865 (ASSERT_KNOWN_EQ): Likewise.
6866 (ASSERT_KNOWN_EQ_AT): Likewise.
6867 (ASSERT_NE): Likewise.
6868 (ASSERT_MAYBE_NE): Likewise.
6869 (ASSERT_MAYBE_NE_AT): Likewise.
6870 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
6871 the assertion to pass.
6872 (ASSERT_STREQ_AT): Likewise.
6874 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
6876 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
6877 interaction with -pie.
6879 2018-04-30 David Malcolm <dmalcolm@redhat.com>
6881 * selftest.h: Fix alphabetization of per-source-file selftest
6884 2018-04-30 Jason Merrill <jason@redhat.com>
6886 PR c++/61982 - dead stores to destroyed objects.
6887 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
6890 2018-04-30 Jason Merrill <jason@redhat.com>
6892 * tree.c (build_clobber): New.
6893 * tree.h: Declare it.
6894 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
6896 2018-04-30 David Malcolm <dmalcolm@redhat.com>
6898 * diagnostic-show-locus.c (layout::layout): Update for
6899 location_get_source_line returning a char_span.
6900 (struct char_span): Move to input.h.
6901 (struct correction): Update for fields in char_span becoming
6903 (struct source_line): Update for location_get_source_line
6904 returning a char_span.
6905 (layout::print_line): Likewise.
6906 * edit-context.c (edited_file::print_content): Likewise.
6907 (edited_file::print_diff_hunk): Likewise.
6908 (edited_file::print_run_of_changed_lines): Likewise.
6909 (edited_file::get_num_lines): Likewise.
6910 (edited_line::edited_line): Likewise.
6911 * final.c (asm_show_source): Likewise.
6912 * input.c (location_get_source_line): Convert return type
6913 from const char * to char_span, losing the final "line_len"
6915 (dump_location_info): Update for the above.
6916 (get_substring_ranges_for_loc): Likewise. Use a char_span
6917 when handling the literal within the line.
6918 (test_reading_source_line): Update for location_get_source_line
6919 returning a char_span.
6920 * input.h (class char_span): Move here from
6921 diagnostic-show-locus.c, converting from a struct to a class.
6922 Make data members private.
6923 (char_span::operator bool): New.
6924 (char_span::length): New.
6925 (char_span::get_buffer): New.
6926 (char_span::operator[]): New.
6927 (char_span::subspan): Make const.
6928 (char_span::xstrdup): New.
6929 (location_get_source_line): Convert return type from const char *
6930 to char_span, losing the final "line_size" param.
6932 2018-04-30 Jan Hubicka <jh@suse.cz>
6934 * lto-wrapper.c (ltrans_priorities): New static var.
6935 (cmp_priority): New.
6936 (run_gcc): Read priorities and if doing parallel build order
6937 the Makefile by them.
6939 2018-04-30 David Malcolm <dmalcolm@redhat.com>
6941 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
6943 2018-04-30 Richard Biener <rguenther@suse.de>
6945 * tree-cfg.c (verify_address): Remove base argument, add
6946 flag whether to check TREE_ADDRESSABLE and do that.
6947 (verify_expr): Remove.
6948 (verify_types_in_gimple_reference): Add pieces from verify_expr.
6949 (verify_gimple_assign_single): Likewise.
6950 (verify_gimple_switch): Likewise.
6951 (verify_expr_location_1): Dereference tp once. Add (disabled)
6952 piece from verify_expr.
6953 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
6955 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
6957 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
6959 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
6961 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
6962 (small_data_pattern): Likewise.
6963 (arc_rewrite_small_data): Likewise.
6964 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
6965 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
6966 (get_symbol_alignment): New function.
6967 (legitimate_small_data_address_p): Likewise.
6968 (legitimate_scaled_address): Update, call
6969 legitimate_small_data_address_p.
6970 (output_sdata): New static variable.
6971 (arc_print_operand): Update how we handle small data operands.
6972 (arc_print_operand_address): Likewise.
6973 (arc_legitimate_address_p): Update, use
6974 legitimate_small_data_address_p.
6975 (arc_rewrite_small_data_p): Remove.
6976 (arc_rewrite_small_data_1): Likewise.
6977 (arc_rewrite_small_data): Likewise.
6978 (small_data_pattern): Likewise.
6979 (compact_sda_memory_operand): Update to use
6980 legitimate_small_data_address_p and get_symbol_alignment.
6981 (prepare_move_operands): Don't rewite sdata pattern.
6982 (prepare_extend_operands): Remove.
6983 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
6985 (zero_extendqisi2): Likewise.
6986 (zero_extendhisi2): Likewise.
6987 (extendqihi2): Likewise.
6988 (extendqisi2): Likewise.
6989 (extendhisi2): Likewise.
6993 * config/arc/constraints.md (Usd): Change it to memory constraint.
6995 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
6997 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
6998 as source of std instructions.
6999 * config/arc/arc.md (movsi_insn): Update pattern predicate to
7000 allow 6-bit constants as source for store instructions.
7001 (movdi_insn): Update instruction pattern to allow 6-bit constants
7002 as source for store instructions.
7004 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
7006 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
7008 2018-04-30 Nathan Sidwell <nathan@acm.org>
7009 Sandra Loosemore <sandra@codesourcery.com>
7011 * dumpfile.c (dump_open): Allow '-' for stdout.
7012 * doc/invoke.texi (Developer Options): Document dump filename
7013 determination early. Document stdin/stdout selection.
7015 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
7017 Microblaze Target: PIC data text relative
7019 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
7020 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
7022 * gcc/config/microblaze/microblaze.h (microblaze_constant_address_p):
7023 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
7024 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
7025 New addressing mode for data-text relative position indepenedent code.
7026 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
7027 'ADDRESS_SYMBOLIC_TXT_REL'.
7028 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
7029 (microblaze_legitimate_pic_operand): Exclude function calls from
7030 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
7031 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
7033 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
7034 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
7035 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
7036 for 'address + offset'.
7037 (microblaze_expand_prologue): Add new function prologue call for
7039 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
7040 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
7041 table in case of TARGET_PIC_DATA_TEXT_REL.
7042 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
7043 * gcc/config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
7044 Add new macros 'UNSPEC_TEXT',
7045 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
7046 + exclude function calls from 'UNSPEC_PLT' in case of data text
7048 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
7049 new target hook for generating address diff vector tables in case of
7051 * doc/tm.texi : Regenerate.
7052 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
7053 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
7054 of addr diff vector generation.
7055 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
7056 target hook definition.
7057 * targhooks.h, gcc/targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
7058 Add default function for generate_pic_addr_diff_vec -> flag_pic.
7059 * doc/invoke.texi (Add new pic option): Add new microblaze pic
7060 option for data text relative.
7062 2018-04-30 Richard Biener <rguenther@suse.de>
7064 * tree-chrec.h (evolution_function_is_constant_p): Remove
7066 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
7068 2018-04-30 Richard Biener <rguenther@suse.de>
7071 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
7073 2018-04-30 Richard Biener <rguenther@suse.de>
7075 PR tree-optimization/28364
7076 PR tree-optimization/85275
7077 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
7078 copying first exit test.
7080 2018-04-28 Mark Wielaard <mark@klomp.org>
7082 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
7084 (dwarf_AT): Handle DW_AT_addr_base.
7085 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
7087 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
7090 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
7091 (*ashl<dwi>3_doubleword_mask_1): Ditto.
7092 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
7093 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
7095 2018-04-28 Richard Biener <rguenther@suse.de>
7097 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
7098 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
7099 to reflect use. Only add interesting stmts.
7101 2018-04-27 Martin Jambor <mjambor@suse.cz>
7104 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
7105 the jump function allows for passing through aggregate values.
7107 2018-04-27 David Malcolm <dmalcolm@redhat.com>
7109 * input.h (in_system_header_at): Convert from macro to inline
7111 (from_macro_expansion_at): Likewise.
7112 (from_macro_definition_at): Likewise.
7114 2018-04-27 Jeff Law <law@redhat.com>
7116 * config.gcc: Mark tile* targets as deprecated/obsolete.
7118 2018-04-27 Richard Biener <rguenther@suse.de>
7120 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
7123 2018-04-27 Richard Biener <rguenther@suse.de>
7125 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
7127 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
7129 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
7130 with Yd constraint. Set "preferred_for_speed" attribute from
7131 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
7133 (*movdi_internal): Ditto.
7134 (movti_interunit splitters): Remove
7135 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
7136 (movdi_interunit splitters): Ditto.
7137 * config/i386/constraints.md (Ye): Remove.
7138 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
7140 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7143 * config/aarch64/constraints.md (Usg): Limit to 31.
7146 2018-04-27 Jakub Jelinek <jakub@redhat.com>
7148 PR tree-optimization/85529
7149 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
7150 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
7151 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
7152 zero extension or masking of the MSB bit.
7153 (optimize_range_tests): Add FIRST_BB argument, pass it through
7154 to optimize_range_tests_var_bound.
7155 (maybe_optimize_range_tests, reassociate_bb): Adjust
7156 optimize_range_tests callers.
7158 2018-04-26 Richard Biener <rguenther@suse.de>
7159 Jakub Jelinek <jakub@redhat.com>
7161 * cgraph.h (symbol_table): Just declare debug method here.
7162 * symtab.c (symbol_table::debug): Define.
7164 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
7166 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
7168 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
7170 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
7171 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
7172 (*movdi_internal): Substitute Yi and Yj constraint with x
7173 and Ym and Yn constraint with y constraint. Update "isa"
7174 attribute and set "preferred_for_speed" attribute from
7175 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
7176 (*movsi_internal): Ditto.
7177 (*movdf_internal): Ditto.
7178 (*movsf_internal): Ditto.
7179 (*zero_extendsidi2): Ditto.
7180 * config/i386/sse.md (vec_set<mode>_0): Ditto.
7181 (sse2_loadld): Ditto.
7182 (*vec_extract<ssevecmodelower>_0): Ditto.
7183 (*vec_extractv4si_0_zext_sse4): Ditto.
7184 (vec_concatv2di): Ditto.
7185 (*vec_dup<mode>): Ditto.
7186 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
7187 * config/i386/constraints.md (Yi): Remove.
7192 2018-04-26 Nathan Sidwell <nathan@acm.org>
7194 * dumpfile.c (dump_open): New.
7195 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
7196 (dump_finish): Detect stdio/stderr by value not name.
7198 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
7200 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
7202 2018-04-26 Tom de Vries <tom@codesourcery.com>
7205 * config/nvptx/nvptx.c (verify_neutering_jumps)
7206 (verify_neutering_labels): New function
7207 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
7209 2018-04-26 Tom de Vries <tom@codesourcery.com>
7212 * config/nvptx/nvptx.c (needs_neutering_p): New function.
7213 (nvptx_single): Use needs_neutering_p to skip over insns that do not
7216 2018-04-26 Richard Biener <rguenther@suse.de>
7217 Tom de Vries <tom@codesourcery.com>
7220 * lto-streamer-out.c (output_function): Fixup loops if required to match
7221 discovery done in the reader.
7223 2018-04-26 Richard Biener <rguenther@suse.de>
7225 PR tree-optimization/85116
7226 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
7227 have a loop exit from the single latch predecessor. Remove
7228 case of header with just condition.
7229 (ch_base::copy_headers): Exclude infinite loops from any
7231 (pass_ch::execute): Record exits.
7233 2018-04-26 Richard Biener <rguenther@suse.de>
7235 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
7236 prologue cost vector and pass it to vect_get_load_cost.
7237 (vect_get_peeling_costs_all_drs): Likewise.
7238 (vect_peeling_hash_get_lowest_cost): Likewise.
7239 (vect_enhance_data_refs_alignment): Likewise.
7241 2018-04-26 Richard Biener <rguenther@suse.de>
7244 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
7245 checking of integer<->pointer conversions.
7246 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
7247 sign-/zero-extending pointer types.
7248 (expand_omp_for_static_chunk): Likewise.
7250 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
7251 Jean Lee <xiaoyur347@gmail.com>
7253 * config/mips/mips.c (mips_asan_shadow_offset): New function.
7254 (TARGET_ASAN_SHADOW_OFFSET): Define.
7255 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
7256 true for -fsanitize=address.
7258 2018-04-25 Mark Wielaard <mark@klomp.org>
7260 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
7263 2018-04-25 Jakub Jelinek <jakub@redhat.com>
7265 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
7266 than "alu", remove explicit "memory" and "imm_disp" attributes.
7267 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
7270 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
7271 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
7272 gen_lowpart_no_emit.
7274 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
7277 * config/i386/i386.c (ix86_expand_builtin): Change memory
7278 operand to XI, extend p0 to Pmode.
7279 * config/i386/i386.md: Change unspec volatile and operand
7280 1 mode to XI, change operand 0 mode to P.
7282 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
7284 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
7285 GET_MODE_MASK before any checking.
7286 (nds32_can_use_bset_p): Likewise.
7287 (nds32_can_use_btgl_p): Likewise.
7289 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
7291 * config/nds32/nds32-doubleword.md: New define_split pattern for
7292 illegal register number.
7294 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
7296 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
7298 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
7300 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
7302 2018-04-25 Richard Biener <rguenther@suse.de>
7304 * lto-streamer.h (LTO_major_version): Bump to 8.
7306 2018-04-25 Jakub Jelinek <jakub@redhat.com>
7308 * BASE-VER: Set to 9.0.0.
7310 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
7312 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
7313 in __abskf2 and __powikf2.
7315 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7318 * config/aarch64/constraints.md (Usg, Usj): New constraints.
7319 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
7320 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
7321 Use the above on operand 2. Reindent.
7322 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
7324 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
7327 * common/config/i386/i386-common.c (ix86_handle_option): Don't
7329 * config/i386/i386.opt (mcet): Removed.
7330 * doc/install.texi: Remove -mcet documentation.
7331 * doc/invoke.texi: Likewise.
7333 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
7336 * doc/install.texi: Remove -mcet from bootstrap-cet.
7338 2018-04-24 Jakub Jelinek <jakub@redhat.com>
7341 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
7342 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
7346 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
7347 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
7348 containing a CONST_VECTOR.
7350 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
7352 * doc/install.texi: Update newlib dependency for nvptx.
7354 2018-04-24 Jakub Jelinek <jakub@redhat.com>
7357 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
7358 instead of INTVAL when shifting x left.
7360 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
7362 PR tree-optimization/85478
7363 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
7364 vect_grouped_store_supported for single element vectors.
7366 2018-04-24 Richard Biener <rguenther@suse.de>
7369 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
7370 load cost increase to the case of non-constant step.
7372 2018-04-24 Jakub Jelinek <jakub@redhat.com>
7375 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
7376 destination if any_malformed_asm.
7378 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
7381 * expr.c (store_field): In the bitfield case, if the value comes from
7382 a function call and is returned in registers by means of a PARALLEL,
7383 do not change the mode of the temporary unless BLKmode and VOIDmode.
7385 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
7387 PR rtl-optimization/85423
7388 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
7389 dependencies to debug insns when the previous insn is non-debug.
7391 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
7393 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
7394 enums into a single definition.
7395 (fls): Fix predicates and printing.
7398 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
7400 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
7401 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
7402 and short u6 immediate.
7403 (check_if_valid_sleep_operand): Remove.
7404 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
7406 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
7408 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
7409 flag_always_save_lp condition.
7410 * config/nds32/nds32.opt (malways-save-lp): New option.
7412 2018-04-22 Shiva Chen <shiva0217@gmail.com>
7414 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
7415 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
7416 * config/nds32/nds32.h
7417 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
7418 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
7420 2018-04-22 Shiva Chen <shiva0217@gmail.com>
7422 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
7423 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
7425 2018-04-22 Shiva Chen <shiva0217@gmail.com>
7426 Chung-Ju Wu <jasonwucj@gmail.com>
7428 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
7430 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
7431 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
7433 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
7435 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
7437 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
7439 * config/nds32/nds32-protos.h (nds32_data_alignment,
7440 nds32_local_alignment): Declare.
7441 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
7442 nds32_local_alignment): New functions.
7443 (TARGET_CONSTANT_ALIGNMENT): Define.
7444 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
7446 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
7448 * config/nds32/nds32.c
7449 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
7450 (TARGET_MODES_TIEABLE_P): Likewise.
7452 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
7454 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
7457 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
7458 Chung-Ju Wu <jasonwucj@gmail.com>
7460 * config/nds32/constants.md (unspec_volatile_element): Add enum values
7461 for unaligned access.
7462 * config/nds32/nds32-intrinsic.c: Implementation of expanding
7464 * config/nds32/nds32-intrinsic.md: Likewise.
7465 * config/nds32/nds32_intrinsic.h: Likewise.
7466 * config/nds32/nds32.h (nds32_builtins): Likewise.
7467 * config/nds32/nds32.opt (munaligned-access): New option.
7468 * config/nds32/nds32.c (nds32_asm_file_start): Display
7469 flag_unaligned_access status.
7471 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
7473 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
7474 -mno-relax is present.
7475 * config/riscv/linux.h (LINK_SPEC): Ditto.
7477 2018-04-20 Martin Sebor <msebor@redhat.com>
7480 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
7482 (gimple_fold_builtin_stxcpy_chk): Same.
7483 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
7485 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
7488 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
7489 __powikf2 when long double is IEEE 128-bit.
7491 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
7493 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
7494 step to make sure stack always aligned.
7496 2018-04-20 Carl Love <cel@us.ibm.com>
7499 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
7500 size check for arg0.
7502 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
7503 Tom de Vries <tom@codesourcery.com>
7506 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
7507 Emit insns for calls too.
7508 (nvptx_find_par): Always look for worker-level predecessor insn.
7509 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
7511 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
7512 (nvptx_process_pars): Propagate frames for calls.
7514 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
7517 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
7519 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
7520 (ix86_handle_option): Don't handle OPT_mibt.
7521 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
7523 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
7525 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
7526 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
7527 (ix86_target_macros): Define __CET__ with flag_cf_protection
7528 for -fcf-protection.
7529 * config/i386/i386.c (isa2_opts): Remove -mibt.
7530 * config/i386/i386.h (TARGET_IBT): Removed.
7531 (TARGET_IBT_P): Likewise.
7532 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
7533 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
7534 * config/i386/i386.opt (mcet): Update help message.
7537 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
7538 -mcet as an alias for -mshstk.
7540 2018-04-20 Richard Biener <rguenther@suse.de>
7543 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
7544 complexity by forcing a single use of the multiply operand.
7546 2018-04-20 Martin Jambor <mjambor@suse.cz>
7549 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
7550 recursion dependency to only apply to non-clones.
7552 2018-04-20 Martin Jambor <mjambor@suse.cz>
7555 * ipa-cp.c (create_specialized_node): Check that clones of
7556 self-recursive edges exist during IPA-CP.
7558 2018-04-19 Toon Moene <toon@moene.org>
7560 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
7563 2018-04-19 Jakub Jelinek <jakub@redhat.com>
7565 PR tree-optimization/85467
7566 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
7567 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
7568 VECTOR_CST element to type.
7570 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
7573 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
7574 * config/i386/i386.md (builtin_setjmp_setup): Removed.
7575 (builtin_longjmp): Likewise.
7576 (save_stack_nonlocal): New pattern.
7577 (restore_stack_nonlocal): Likewise.
7579 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
7582 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
7583 Replace ASM_OUTPUT_LABEL with fprintf.
7585 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
7588 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
7589 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
7590 * config/i386/i386-c.c (ix86_target_macros_internal): Also
7591 define __IBT__ and __SHSTK__ for -fcf-protection.
7592 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
7594 (ix86_trampoline_init): Likewise.
7595 (x86_output_mi_thunk): Likewise.
7596 (ix86_notrack_prefixed_insn_p): Likewise.
7597 (ix86_option_override_internal): Don't disallow -fcf-protection.
7598 * config/i386/i386.md (rdssp<mode>): Also enable for
7600 (incssp<mode>): Likewise.
7601 (nop_endbr): Likewise.
7602 * config/i386/i386.opt (mcet): Change help message to built-in
7606 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
7607 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
7608 enable CET built-in functions.
7610 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
7612 * common/config/i386/i386-common.c
7613 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
7614 OPTION_MASK_ISA_MOVDIRI_UNSET,
7615 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
7616 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
7617 * config.gcc (movdirintrin.h): New header.
7618 * config/i386/cpuid.h (bit_MOVDIRI,
7619 bit_MOVDIR64B): New bits.
7620 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
7622 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
7623 (VOID, PVOID, PCVOID)): New function types.
7624 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
7625 __builtin_ia32_directstoreu_u64,
7626 __builtin_ia32_movdir64b): New builtins.
7627 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
7628 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
7630 (ix86_valid_target_attribute_inner_p): Ditto.
7631 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
7632 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
7633 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
7634 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
7635 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
7636 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
7637 (movdiri<mode>, movdir64b_<mode>): New.
7638 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
7639 * config/i386/immintrin.h: Include movdirintrin.h.
7640 * config/i386/movdirintrin.h: New file.
7641 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
7643 2018-04-19 Richard Biener <rguenther@suse.de>
7646 * cfg.c (clear_bb_flags): When loop state says we have
7647 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
7649 2018-04-19 Richard Biener <rguenther@suse.de>
7651 PR tree-optimization/84737
7652 * tree-vect-data-refs.c (vect_copy_ref_info): New function
7653 copying restrict info.
7654 (vect_setup_realignment): Use it.
7655 * tree-vectorizer.h (vect_copy_ref_info): Declare.
7656 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
7657 the first DR to all generated stores.
7658 (vectorizable_load): Likewise for loads.
7660 2018-04-19 Jakub Jelinek <jakub@redhat.com>
7662 PR tree-optimization/85446
7663 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
7664 the integral and pointer types to have the same precision.
7666 * doc/install.texi: Document --disable-cet being the default and
7669 2018-04-18 Martin Liska <mliska@suse.cz>
7671 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
7674 2018-04-18 Martin Liska <mliska@suse.cz>
7677 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
7680 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
7681 arguments if they are comparable.
7683 2018-04-18 Martin Liska <mliska@suse.cz>
7686 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
7689 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
7692 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
7695 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
7696 ENDBR after calling __morestack.
7698 2018-04-18 David Malcolm <dmalcolm@redhat.com>
7701 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
7702 by using gcc_base_ver to generate a gcc_driver_version, and use
7703 it when generating GCC_DRIVER_NAME.
7704 * configure: Regenerate.
7706 2018-04-18 Jakub Jelinek <jakub@redhat.com>
7709 * config.gcc: Obsolete powerpc*-*-*spe*.
7711 2018-04-17 Jakub Jelinek <jakub@redhat.com>
7714 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
7715 (stabstr_D): Change type of unum from unsigned int to
7716 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
7719 2018-04-17 Jim Wilson <jimw@sifive.com>
7722 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
7723 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
7724 Set arg_pointer_offset after using pretend_args_size.
7726 2018-04-17 Jakub Jelinek <jakub@redhat.com>
7728 PR rtl-optimization/85431
7729 * dse.c (record_store): Ignore zero width stores.
7732 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
7733 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
7734 __builtin_stack_restore rather than after it.
7735 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
7736 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
7737 argument instead of virtual_dynamic_stack_rtx.
7739 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
7741 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
7743 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7744 Add note to error message to explain internal mapping of overloaded
7745 built-in function name to non-overloaded built-in function name.
7746 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
7749 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
7752 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
7753 where the inputs overlap with the output.
7755 2018-04-17 Jakub Jelinek <jakub@redhat.com>
7757 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
7758 (=v, v) alternative and explicit "memory" attribute.
7759 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
7760 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
7762 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
7763 "sselog1" type instead of "sselog".
7764 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
7765 "sselog". Remove explicit "memory" attribute.
7766 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
7767 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
7769 (vec_extract_hi_v32hi): Merge all alternatives into one, use
7770 "sselog1" type instead of "sselog". Remove explicit "memory"
7772 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
7773 use "sselog1" type instead of "sselog". Remove explicit "memory"
7775 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
7776 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
7778 (vec_extract_hi_v64qi): Merge all alternatives into one, use
7779 "sselog1" type instead of "sselog". Remove explicit "memory"
7781 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
7782 use "sselog1" type instead of "sselog". Remove explicit "memory"
7786 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
7789 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
7792 2018-04-17 Martin Jambor <mjambor@suse.cz>
7795 * ipa-cp.c (create_specialized_node): Call
7796 expand_all_artificial_thunks if necessary.
7798 2018-04-17 Martin Liska <mliska@suse.cz>
7801 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
7802 in message, remote space in between '_G' and '('.
7804 2018-04-17 Jakub Jelinek <jakub@redhat.com>
7807 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
7808 avx512f_vmcmp<mode>3<round_saeonly_name>,
7809 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
7810 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
7811 avx512f_rndscale<mode><round_saeonly_name>,
7812 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
7813 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
7814 Use %<iptr>2 instead of %2 for -masm=intel.
7815 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
7816 avx512f_vcvttss2usi<round_saeonly_name>,
7817 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
7819 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
7820 avx512f_vcvttsd2usi<round_saeonly_name>,
7821 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
7822 Use %q1 instead of %1 for -masm=intel.
7823 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
7824 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
7825 of %3 for -masm=intel.
7826 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
7828 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
7830 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
7832 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
7833 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
7835 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
7837 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
7838 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
7839 %g1 and one with %0 and %1.
7840 (avx512er_vmrcp28<mode><round_saeonly_name>,
7841 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
7843 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
7844 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
7845 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
7846 of %0 and %{%4%} for -masm=intel.
7847 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
7848 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
7849 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
7850 order of %0 and %{%5%}%{z%} for -masm=intel.
7852 2018-04-17 Jan Hubicka <jh@suse.cz>
7855 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
7857 2018-04-17 Martin Liska <mliska@suse.cz>
7860 * multiple_target.c (create_dispatcher_calls): Set apostrophes
7861 for target_clone error message. Make default implementation
7862 clone to be a local declaration.
7863 (separate_attrs): Add new argument and check for an empty
7865 (expand_target_clones): Handle it.
7866 (ipa_target_clone): Make redirection just for target_clones
7869 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
7870 Tom de Vries <tom@codesourcery.com>
7873 * omp-expand.c (expand_oacc_for): Add dummy false branch for
7874 tiled basic blocks without omp continue statements.
7876 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
7879 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
7880 vec_extract expression as having side effects to make sure it gets
7883 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
7886 * config/i386/i386.c (get_builtin_code_for_version): Check
7889 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
7892 * gcc/config.gcc: Support "skylake".
7893 * gcc/config/i386/i386-c.c (ix86_target_macros_internal): Handle
7895 * gcc/config/i386/i386.c (m_SKYLAKE): Define.
7896 (processor_target_table): Add "skylake".
7897 (ix86_option_override_internal): Add "skylake".
7898 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
7899 PROCESSOR_CANNONLAKE.
7900 (get_builtin_code_for_version): Fix priority for
7901 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
7902 PROCESSOR_SKYLAKE-AVX512.
7903 * gcc/config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
7904 (processor_type): Add PROCESSOR_SKYLAKE.
7906 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
7907 Jason Merrill <jason@redhat.com>
7910 * convert.c (convert_to_integer_1): Use direct recursion for
7911 enumeral types and types with a precision less than the number
7912 of bits in their mode.
7914 2018-04-16 Julia Koval <julia.koval@intel.com>
7917 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
7918 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
7920 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
7923 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
7924 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
7925 and -mno-direct-move.
7927 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
7930 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
7931 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
7932 Ensure negative shifts result in {0}.
7934 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
7936 PR rtl-optimization/79916
7937 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
7938 regs (if any) to define how to gnerate SD moves when LRA is in
7941 2018-04-13 Jakub Jelinek <jakub@redhat.com>
7943 PR rtl-optimization/85393
7944 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
7945 * except.c (expand_dw2_landing_pad_for_region): Make static.
7946 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
7947 a label and unconditional jump to old_bb, rather than
7948 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
7951 PR rtl-optimization/85376
7952 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
7953 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
7954 instead of a specific value.
7956 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
7957 Bin Cheng <bin.cheng@arm.com>
7959 PR tree-optimization/82965
7960 PR tree-optimization/83991
7961 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
7962 by_profile_only parameter.
7963 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
7964 information if the loop was predicted to iterate too many times.
7965 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
7967 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
7970 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
7973 2018-04-13 Martin Liska <mliska@suse.cz>
7974 Jakub Jelinek <jakub@redhat.com>
7977 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
7978 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
7979 * builtins.c (expand_builtin_memory_copy_args): Use
7980 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
7981 handle dest_addr == pc_rtx.
7983 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
7986 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
7987 asked to not generate direct moves.
7988 (fix_trunc<mode>si2_stfiwx): Similar.
7989 (fix_trunc<mode>si2_internal): Similar.
7991 2018-04-12 Jakub Jelinek <jakub@redhat.com>
7994 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
7995 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
7996 lookup if dest in some wider mode is known to be const0_rtx and
7997 if so, record permanent equivalence for it to be ZERO_EXTEND of
7998 the narrower mode destination.
8000 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
8002 * lto-streamer-out.c (output_function): Revert 259346.
8003 * omp-expand.c (expand_oacc_for): Likewise.
8005 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
8007 PR rtl-optimization/85354
8008 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
8009 * sel-sched.c (sel_global_init): ... here.
8011 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
8014 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
8015 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
8016 mode for PE-COFF targets.
8017 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
8018 (i386_pe_asm_lto_end): Likewise.
8019 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
8020 (TARGET_ASM_LTO_END): Likewise.
8021 * config/i386/winnt.c (saved_debug_info_level): New static variable.
8022 (i386_pe_asm_lto_start): New function.
8023 (i386_pe_asm_lto_end): Likewise.
8025 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
8026 Richard Biener <rguenther@suse.de>
8029 * lto-streamer-out.c (output_function): Fix CFG loop state before
8031 * omp-expand.c (expand_oacc_for): Handle calls to internal
8032 functions like regular functions.
8034 2018-04-12 Richard Biener <rguenther@suse.de>
8037 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
8038 for the early LTO debug to properly generate references to it
8039 during DIE emission. Do not re-use that for the skeleton for
8041 (dwarf2out_early_finish): Likewise.
8043 2018-04-12 Jakub Jelinek <jakub@redhat.com>
8046 * config/i386/sse.md
8047 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
8048 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
8049 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
8050 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
8051 and output is a reg, avoid creating invalid lowpart subreg, but
8052 instead split into a 512-bit move. Don't split if not AVX512VL,
8053 input is xmm16+ reg and output is a mem.
8054 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
8055 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
8056 xmm16+ reg and output is a mem.
8058 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8060 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
8061 also for flag_dwarf2_cfi_asm.
8063 2018-04-12 Jakub Jelinek <jakub@redhat.com>
8065 PR rtl-optimization/85342
8066 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
8067 a bool scalar var inside of the loop instead. Don't try to update
8068 recog_data.operand after failed apply_change_group.
8070 2018-04-12 Tom de Vries <tom@codesourcery.com>
8073 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
8074 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
8075 array with flexible array member as array without given dimension.
8076 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
8077 argument for undefined param to true.
8079 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
8082 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
8083 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
8084 from PowerPC section.
8085 * config/rs6000/sysv4.opt (mcall-): Improve help text.
8086 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
8087 help text that is too long.
8088 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
8089 help text that is too long.
8090 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
8091 help text that is too long.
8093 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
8095 * config/alpha/alpha.md (stack_probe_internal): Rename
8096 from "probe_stack". Update all callers.
8098 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
8100 PR rtl-optimization/84566
8101 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
8102 sched_macro_fuse_insns.
8104 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
8107 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
8108 (compute_block_dependences): ... from here.
8110 2018-04-11 Jakub Jelinek <jakub@redhat.com>
8112 PR tree-optimization/85331
8113 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
8114 from int to HOST_WIDE_INT.
8116 2018-04-11 Martin Jambor <mjambor@suse.cz>
8119 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
8120 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
8121 not the same as the source val.
8122 (cgraph_edge_brings_value_p): New parameter.
8123 (gather_edges_for_value): Pass destination value to
8124 cgraph_edge_brings_value_p.
8125 (perhaps_add_new_callers): Likewise.
8126 (get_info_about_necessary_edges): Likewise and exclude values brought
8127 only by self-recursive edges.
8128 (create_specialized_node): Redirect only clones of self-calling edges.
8129 (+self_recursive_pass_through_p): New function.
8130 (find_more_scalar_values_for_callers_subset): Use it.
8131 (find_aggregate_values_for_callers_subset): Likewise.
8132 (known_aggs_to_agg_replacement_list): Removed.
8133 (decide_whether_version_node): Re-calculate known constants for all
8134 remaining context clones.
8136 2018-04-11 Richard Biener <rguenther@suse.de>
8139 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
8140 from early DWARF output.
8141 (dwarf2out_early_finish): Output line info unconditionally into
8142 early DWARF and add reference to it.
8144 2018-04-11 Jakub Jelinek <jakub@redhat.com>
8147 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
8148 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
8149 other than V2DFmode using iptr mode attribute.
8150 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
8152 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
8154 PR rtl-optimization/84659
8155 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
8157 2018-04-11 Jakub Jelinek <jakub@redhat.com>
8160 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
8162 (output_loc_list): Pass address of a dummy size variable even in the
8163 locview handling loop.
8164 (index_location_lists): Add comment on why skip_loc_list_entry can't
8167 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
8170 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
8173 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
8176 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
8177 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
8178 and -mstring-compare-inline-limit.
8180 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
8183 * gcc/config/rs6000/rs6000.md (allocate_stack): Put the residual size
8184 for stack clash protection in a register whenever we need it to be in
8187 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
8189 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
8190 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
8192 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
8195 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
8197 (mlong-double-): Ditto.
8198 * config/rs6000/sysv4.opt (msdata=): Ditto.
8199 (mtls-size=): Ditto.
8201 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
8203 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
8204 erroneous entries for
8205 "vector int vec_ldl (int, long int *)", and
8206 "vector unsigned int vec_ldl (int, unsigned long int *)".
8207 Add comments and entries for
8208 "vector bool char vec_ldl (int, bool char *)",
8209 "vector bool short vec_ldl (int, bool short *)",
8210 "vector bool int vec_ldl (int, bool int *)",
8211 "vector bool long long vec_ldl (int, bool long long *)",
8212 "vector pixel vec_ldl (int, pixel *)",
8213 "vector long long vec_ldl (int, long long *)",
8214 "vector unsigned long long vec_ldl (int, unsigned long long *)".
8215 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
8216 type tree bool_long_long_type_node and correct definition of
8217 bool_V2DI_type_node to make reference to this new type tree.
8218 (rs6000_mangle_type): Replace erroneous reference to
8219 bool_long_type_node with bool_long_long_type_node.
8220 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
8221 comments to emphasize sign distinctions for char and int types and
8222 replace RS6000_BTI_bool_long constant with
8223 RS6000_BTI_bool_long_long constant. Also add comment to restrict
8224 use of RS6000_BTI_pixel.
8225 (bool_long_type_node): Remove this macro definition.
8226 (bool_long_long_type_node): New macro definition
8228 2018-04-10 Jakub Jelinek <jakub@redhat.com>
8230 PR rtl-optimization/85300
8231 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
8232 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
8233 simplify_unary_operation fails.
8235 2018-04-10 Martin Liska <mliska@suse.cz>
8237 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
8238 cgraph_edge and ipa_ref.
8240 2018-04-10 Jakub Jelinek <jakub@redhat.com>
8244 * config/i386/sse.md
8245 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
8246 computation of the VEC_MERGE selector from mask.
8247 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
8248 Fix decoding of the VEC_MERGE selector into mask.
8250 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
8252 PR tree-optimization/85286
8253 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
8255 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
8257 * final.c (final_1): Set insn_last_address as well as
8258 insn_current_address.
8260 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8263 * explow.c (emit_stack_probe): Call validize_mem on memory location
8264 before passing it to gen_probe_stack. Create address operand and
8265 legitimize it for the probe_stack_address case.
8267 2018-04-09 Jan Hubicka <jh@suse.cz>
8270 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
8271 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
8272 * tree.c (free_lang_data_in_type): Fix handling of binfos;
8274 (free_lang_data): Rebuild type inheritance graph.
8276 2018-04-09 Martin Sebor <msebor@redhat.com>
8278 * invoke.texi (-finline-small-functions): Mention other optimization
8280 (-findirect-inlining, -fpartial-inlining): Same.
8281 (-finline-functions-called-once): Same.
8282 (-freorder-blocks-and-partition): Same.
8284 2018-04-09 Jan Hubicka <jh@suse.cz>
8287 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
8288 jumps; choose last target that matches the criteria (i.e.
8289 no partition changes for non-crossing jumps).
8290 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
8291 support for redirecting crossing jumps to non-crossing.
8293 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
8295 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
8296 also for naked functions.
8298 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
8300 * config/arc/arc.md (add_shift): New pattern.
8301 (add_shift2): Likewise.
8302 (sub_shift): Likewise.
8303 (sub_shift_cmp0_noout): Likewise.
8304 (compare_si_ashiftsi): Likewise.
8305 (xbfu_cmp0_noout): New combine pattern.
8306 (xbfu_cmp0"): Likewise.
8307 (movsi_set_cc_insn): Place the predicable variant first.
8308 (commutative_binary_cmp0_noout): Remove clobber.
8309 (commutative_binary_cmp0): New pattern.
8310 (noncommutative_binary_cmp0): Likewise.
8311 (noncommutative_binary_cmp0_noout): Likewise.
8312 (noncommutative_binary_comparison_result_used): Removed.
8313 (rsub_cmp0): New pattern.
8314 (rsub_cmp0_noout): Likewise.
8315 (extzvsi): Changed, keep only meaningful variants.
8316 (SQH, SEZ): New iterators.
8317 (SQH_postfix): New mode attribute.
8318 (SEZ_prefix): New code attribute.
8319 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
8320 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
8321 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
8323 (noncommutative_operator): Check the availability of barrel
8326 2018-04-09 Richard Biener <rguenther@suse.de>
8328 PR tree-optimization/85284
8329 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
8330 Only use the niter constraining form of simple_iv when the exit
8333 2018-04-09 Tom de Vries <tom@codesourcery.com>
8336 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
8337 (define_expand "*memory_barrier"): New define_expand.
8338 (define_insn "memory_barrier"): New insn.
8340 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
8342 PR rtl-optimization/80463
8343 PR rtl-optimization/83972
8344 PR rtl-optimization/83480
8346 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
8347 correct producer for the insn.
8348 (tidy_control_flow): Fixup seqnos in case of debug insns.
8350 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
8352 PR rtl-optimization/83913
8354 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
8355 different sched-times when merging exprs.
8357 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
8359 PR rtl-optimization/83962
8361 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
8362 tidy_fallthru_edge and tidy_control_flow.
8364 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
8366 PR rtl-optimization/83530
8368 * sel-sched.c (force_next_insn): New global variable.
8369 (remove_insn_for_debug): When force_next_insn is true, also leave only
8370 next insn in the ready list.
8371 (sel_sched_region): When the region wasn't scheduled, make another pass
8372 over it with force_next_insn set to 1.
8374 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
8376 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
8378 * config/nds32/constants.md (unspec_volatile_element): Add enum values
8379 for interrupt control.
8380 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
8381 functions for interrupt control.
8382 * config/nds32/nds32-intrinsic.md: Likewise.
8383 * config/nds32/nds32_intrinsic.h: Likewise.
8384 * config/nds32/nds32.h (nds32_builtins): Likewise.
8386 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
8388 * config/nds32/nds32.c (nds32_init_machine_status,
8389 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
8390 strict_aligned_p field.
8391 (nds32_expand_to_rtl_hook): New function.
8392 (TARGET_EXPAND_TO_RTL_HOOK): Define.
8393 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
8395 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
8396 Chung-Ju Wu <jasonwucj@gmail.com>
8398 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
8399 * config/nds32/nds32-n7.md: New file.
8400 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
8401 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
8403 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
8404 * config/nds32/nds32.md (pipeline_model): Add n7.
8405 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
8406 * config/nds32/pipelines.md: Include n7 settings.
8408 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
8409 Chung-Ju Wu <jasonwucj@gmail.com>
8411 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
8412 * config/nds32/nds32-e8.md: New file.
8413 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
8414 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
8416 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
8417 * config/nds32/nds32.md (pipeline_model): Add e8.
8418 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
8419 * config/nds32/pipelines.md: Include e8 settings.
8421 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
8422 Chung-Ju Wu <jasonwucj@gmail.com>
8424 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
8425 * config/nds32/nds32-n8.md: New file.
8426 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
8427 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
8429 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
8430 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
8431 * config/nds32/nds32.md (pipeline_model): Add n8.
8432 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
8433 * config/nds32/pipelines.md: Include n8 settings.
8435 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
8436 Chung-Ju Wu <jasonwucj@gmail.com>
8438 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
8439 * config/nds32/nds32-n9-2r1w.md: New file.
8440 * config/nds32/nds32-n9-3r2w.md: New file.
8441 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
8442 nds32_register_ports): New or modify for cpu n9.
8443 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
8445 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
8446 * config/nds32/nds32-utils.c: New file.
8447 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
8448 TARGET_MUL_SLOW): Define.
8449 * config/nds32/nds32.md (pipeline_model): New attribute.
8450 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
8451 New options that support cpu n9.
8452 * config/nds32/pipelines.md: Include n9 settings.
8453 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
8455 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
8457 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
8458 information if necessary.
8459 (output_cond_branch_compare_zero): Likewise.
8460 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
8461 (nds32_target_alignment): Refine for alignment.
8462 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
8463 (FUNCTION_BOUNDARY): Modify.
8464 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
8466 * config/nds32/nds32.opt (malways-align, malign-functions): New.
8468 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
8470 * config/nds32/constants.md (unspec_volatile_element): Add values for
8471 TLB operation and data prefetch.
8472 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
8473 functions for TLB operation and data prefetch.
8474 * config/nds32/nds32-intrinsic.md: Likewise.
8475 * config/nds32/nds32_intrinsic.h: Likewise.
8476 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
8477 (nds32_print_operand): Likewise.
8478 * config/nds32/nds32.h (nds32_builtins): Likewise.
8480 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
8481 Andrew Pinski <pinsika@gcc.gnu.org>
8484 * match.pd: Use constant_boolean_node of correct type instead of
8485 boolean_true_node or boolean_false_node for simplifying
8486 pointer comparisons to zero.
8488 2018-04-07 Jakub Jelinek <jakub@redhat.com>
8490 PR tree-optimization/80021
8491 * tree.c (verify_type_variant): Make error call in verify_variant_match
8492 translatable and remove final full stop.
8494 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
8496 * config/nds32/constants.md (unspec_volatile_element): Add
8497 UNSPEC_VOLATILE_EH_RETURN.
8498 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
8499 nds32_output_stack_pop): Support dwarf exception handling process.
8500 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
8501 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
8502 exception handling process.
8503 (nds32_compute_stack_frame): Likewise.
8504 (nds32_return_addr_rtx): Likewise.
8505 (nds32_initial_elimination_offset): Likewise.
8506 (nds32_expand_prologue): Likewise.
8507 (nds32_expand_epilogue): Likewise.
8508 (nds32_dynamic_chain_address): New function.
8509 * config/nds32/nds32.h (machine_function): Add fields for dwarf
8511 (DYNAMIC_CHAIN_ADDRESS): Define.
8512 (EH_RETURN_DATA_REGNO): Define.
8513 (EH_RETURN_STACKADJ_RTX): Define.
8514 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
8515 patterns for dwarf exception handling.
8517 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
8519 * config/nds32/nds32.h: Clean up obsolete macros.
8521 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
8523 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
8524 Add enum values for particular instructions.
8525 * config/nds32/nds32-intrinsic.c: Implementation of expanding
8526 particular intrinsic functions.
8527 * config/nds32/nds32-intrinsic.md: Likewise.
8528 * config/nds32/nds32_intrinsic.h: Likewise.
8529 * config/nds32/nds32.h (nds32_builtins): Likewise.
8530 * config/nds32/nds32.md (type): Add pbsad and pbsada.
8531 (btst, ave): New patterns for particular instructions.
8533 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
8535 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
8536 Add enum values for atomic load/store and memory sync.
8537 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
8539 * config/nds32/nds32-intrinsic.md: Likewise.
8540 * config/nds32/nds32_intrinsic.h: Likewise.
8541 * config/nds32/nds32.h (nds32_builtins): Likewise.
8543 2018-04-07 Jakub Jelinek <jakub@redhat.com>
8545 PR tree-optimization/85257
8546 * fold-const.c (native_encode_vector): If not all elts could fit
8547 and off is -1, return 0 rather than offset.
8548 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
8549 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
8550 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
8551 adjust buffer in native_interpret_expr call.
8553 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
8555 * config/nds32/constants.md (unspec_volatile_element): Add cache
8556 control enum values.
8557 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
8558 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
8559 * config/nds32/nds32.c (nds32_cctl_names): New.
8560 (nds32_print_operand): Handle cache control register names.
8561 * config/nds32/nds32.h (nds32_builtins): New enum values.
8562 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
8564 * config/nds32/nds32.md (type): Add mmu.
8565 * config/nds32/pipelines.md (simple_insn): Add mmu.
8567 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
8569 * config/nds32/nds32.md (type): Remove call.
8570 * config/nds32/pipelines.md (simple_insn): Likewise.
8572 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
8574 * config/nds32/constants.md (unspec_volatile_element): Add
8575 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
8576 UNSPEC_VOLATILE_FMFCFG.
8577 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
8578 description for fmfcfg and fmfcsr.
8579 (bdesc_1arg): Add fmtcsr.
8580 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
8581 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
8582 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
8583 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
8584 unspec_fmfcfg): New patterns.
8585 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
8586 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
8587 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
8588 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
8589 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
8590 __nds32__fmfcfg): Define.
8592 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
8594 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
8595 intrinsic register names.
8596 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
8597 intrinsic register enum values and macros.
8599 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
8601 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
8602 for load/store addressing form.
8603 (nds32_print_operand_address): Likewise.
8605 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
8608 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
8609 based on LABEL_REF. Remove useless assertion.
8610 (pic_address_needs_scratch): Fix formatting.
8611 (sparc_legitimize_pic_address): Minor tweaks.
8612 (sparc_delegitimize_address): Adjust assertion accordingly.
8613 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
8614 into symbolic_operand.
8615 (movsi_high_pic_label_ref): Likewise.
8616 (movsi_lo_sum_pic_label_ref): Likewise.
8617 (movdi_pic_label_ref): Likewise.
8618 (movdi_high_pic_label_ref): Likewise.
8619 (movdi_lo_sum_pic_label_ref): Likewise.
8621 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
8623 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
8624 custom LIB_SPEC setup.
8626 2018-04-06 Ruslan Bukin <br@bsdpad.com>
8627 Kito Cheng <kito.cheng@gmail.com>
8629 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
8630 * config/riscv/freebsd.h: New.
8632 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
8634 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
8635 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
8638 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
8639 Kito Cheng <kito.cheng@gmail.com>
8641 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
8642 nds32_output_call, nds32_symbol_binds_local_p): New functions.
8643 * config/nds32/nds32-protos.h (nds32_output_call,
8644 nds32_output_return): Declare.
8645 * config/nds32/nds32.md: Refine all the call and return patterns.
8647 2018-04-06 Jakub Jelinek <jakub@redhat.com>
8650 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
8651 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
8653 PR rtl-optimization/84872
8654 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
8655 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
8658 2018-04-06 Tamar Christina <tamar.christina@arm.com>
8660 * expr.c (copy_blkmode_to_reg): Revert 254862.
8661 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
8663 2018-04-06 Richard Biener <rguenther@suse.de>
8666 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
8667 after seeing a component reference with an adjacent field. Treat
8668 refs to arrays at struct end of external decls similar to
8669 refs to unconstrained commons.
8671 2018-04-06 Jakub Jelinek <jakub@redhat.com>
8674 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
8675 look through SAVE_EXPRs with non-side-effects argument. Adjust
8677 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
8680 2018-04-06 Richard Biener <rguenther@suse.de>
8683 * alias.c (find_base_term): New wrapper around find_base_term
8684 unwinding CSELIB_VAL_PTR changes.
8685 (find_base_term): Do not restore CSELIB_VAL_PTR during the
8688 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8690 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
8692 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
8693 constant definitions.
8694 ("nop"): lr 0,0 -> nopr r0
8695 ("nop_lr0", "nop_lr1"): New insn definitions.
8697 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
8699 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
8700 NDS32_V3PUSH_AVAILABLE_P macro.
8702 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
8703 Chung-Ju Wu <jasonwucj@gmail.com>
8705 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
8706 (nds32*-*-*): Add float and fpu_config into supported_defaults.
8707 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
8708 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
8709 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
8710 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
8711 * config/nds32/constraints.md: New constraints and checking for hard
8712 float configuration.
8713 * config/nds32/iterators.md: New mode iterator and attribute for hard
8714 float configuration.
8715 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
8717 * config/nds32/nds32-fpu.md: New file.
8718 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
8719 deal with hard float code generation.
8720 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
8722 (abi_type, float_reg_number): New enum type.
8723 * config/nds32/nds32-predicates.c: New predicates for hard float.
8724 * config/nds32/nds32-protos.h: Declare functions for hard float.
8725 * config/nds32/nds32.c: Implementation for hard float configuration.
8726 * config/nds32/nds32.h: Definitions for hard float configuration.
8727 * config/nds32/nds32.md: Include hard float machine description and
8728 modify patterns for hard float configuration.
8729 * config/nds32/nds32.opt: New options for hard float configuration.
8730 * config/nds32/predicates.md: New predicates for hard float
8733 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
8735 * common/config/nds32/nds32-common.c
8736 (nds32_option_optimization_table): Enable -mreleax-hint by default.
8738 2018-04-05 Jakub Jelinek <jakub@redhat.com>
8741 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
8742 CONSTRUCTOR_ELT (ctor, ...)->value.
8744 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
8747 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
8749 2018-04-05 Tom de Vries <tom@codesourcery.com>
8752 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
8755 2018-04-05 Shiva Chen <shiva0217@gmail.com>
8756 Kito Cheng <kito.cheng@gmail.com>
8758 * config/nds32/constraints.md (U33): Fine-tune checking condition.
8759 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
8760 * config/nds32/nds32.h (nds32_16bit_address_type): Add
8761 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
8763 2018-04-05 Shiva Chen <shiva0217@gmail.com>
8764 Kito Cheng <kito.cheng@gmail.com>
8766 * config/nds32/constraints.md (Ufe): New memory constraint.
8767 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
8768 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
8769 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
8771 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
8772 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
8774 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
8776 * config/nds32/nds32.md: Use optimize_size in the condition for
8777 alu-shift instructions.
8779 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
8781 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
8783 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
8785 * config/nds32/nds32.md (negsi2): Refine pattern.
8787 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
8788 Chung-Ju Wu <jasonwucj@gmail.com>
8790 * config/nds32/iterators.md (shift_rotate): New code iterator.
8791 (shift): New code attribute.
8792 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
8793 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
8794 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
8795 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
8796 bit-wise operations.
8797 (andsi3, *andsi3): Ditto.
8798 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
8799 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
8800 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
8801 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
8802 nds32_ior_operand, nds32_xor_operand): New predicates.
8804 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
8806 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
8807 (addsi3, subsi3): ... this.
8809 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
8811 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
8813 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
8815 * config/nds32/nds32.md: Adjust indention.
8817 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
8819 * config/nds32/nds32.md (feature): New attribute.
8821 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
8823 * config/nds32/nds32.md (subtype): New attribute.
8825 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
8828 * config/arm/arm-builtins.c (arm_expand_builtin): Change
8829 expansion to perform a bitwise AND of the argument followed by a
8830 boolean negation of the result.
8832 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
8834 PR rtl-optimization/84878
8835 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
8836 the basic block. Assert the use reference is not artificial and that
8837 it has an associated insn.
8839 2018-04-04 Michael Matz <matz@suse.de>
8841 * builtins.c (compute_objsize): Pass correct operand
8842 to array_at_struct_end_p.
8844 2018-04-04 Richard Biener <rguenther@suse.de>
8847 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
8848 from contexts for DINFO_LEVEL_TERSE and below.
8850 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
8852 * config/nds32/nds32-doubleword.md (move_<mode>): Require
8853 resiter_operand condition.
8854 * config/nds32/nds32.md (*move<mode>): Ditto.
8856 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
8857 Monk Chiang <sh.chiang04@gmail.com>
8859 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
8861 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
8863 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
8865 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
8866 Kito Cheng <kito.cheng@gmail.com>
8868 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
8869 nds32_cond_code_str, output_cond_branch,
8870 output_cond_branch_compare_zero, nds32_expand_cbranch,
8871 nds32_expand_cstore, nds32_expand_movcc,
8872 nds32_output_cbranchsi4_equality_zero,
8873 nds32_output_cbranchsi4_equality_reg,
8874 nds32_output_cbranchsi4_equality_reg_or_const_int,
8875 nds32_output_cbranchsi4_greater_less_zero: New functions.
8876 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
8877 nds32_expand_cstore, nds32_expand_movcc,
8878 nds32_output_cbranchsi4_equality_zero,
8879 nds32_output_cbranchsi4_equality_reg,
8880 nds32_output_cbranchsi4_equality_reg_or_const_int,
8881 nds32_output_cbranchsi4_greater_less_zero): Declare.
8882 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
8883 nds32_rimm11s_operand): New predicates.
8884 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
8885 * config/nds32/nds32.md: Rewrite all the branch and conditional move
8888 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
8890 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
8891 * config/nds32/nds32.md: Ditto.
8892 * config/nds32/pipelines.md: Ditto.
8894 2018-04-04 Richard Biener <rguenther@suse.de>
8896 PR tree-optimization/85168
8897 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
8898 propagating abnormals.
8900 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
8902 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
8904 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
8905 Kito Cheng <kito.cheng@gmail.com>
8907 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
8908 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
8909 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
8910 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
8911 * config/nds32/nds32.md (sibcall_internal): New.
8912 (sibcall_register): Remove.
8913 (sibcall_immediate): Remove.
8914 (sibcall_value_internal): New.
8915 (sibcall_value_register): Remove.
8916 (sibcall_value_immediate): Remove.
8917 * config/nds32/predicates.md (nds32_general_register_operand): New.
8918 (nds32_call_address_operand): New.
8920 2018-04-03 Jakub Jelinek <jakub@redhat.com>
8922 PR rtl-optimization/85167
8923 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
8924 bb_defs if *split_p, instead preinitialize it to NULL.
8926 PR tree-optimization/85156
8927 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
8928 evaluating the argument multiple times.
8930 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
8932 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
8934 (_mm_cvtpd_ps): Likewise.
8935 (_mm_cvttpd_epi32): Likewise.
8936 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
8937 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
8938 vector, pixel, and bool following altivec.h include.
8940 2018-04-03 Martin Sebor <msebor@redhat.com>
8942 * doc/extend.texi (Common Function Attributes): Clarify.
8943 (const attribute): Likewise.
8944 (pure attribute): Likewise.
8946 2018-04-03 Jakub Jelinek <jakub@redhat.com>
8949 * config/i386/i386.c (ix86_expand_vector_set): Use
8950 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
8952 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
8954 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
8955 instructions when changing rounding bits to preserve precision bits
8956 in the x87 control word.
8958 2018-04-03 Martin Liska <mliska@suse.cz>
8960 PR tree-optimization/82491
8961 * rtl.h (strip_offset_and_add): Replace += suboffset with
8962 poly_uint64 () + suboffset.
8964 2018-03-29 Martin Liska <mliska@suse.cz>
8965 Martin Jambor <mjambor@suse.cz>
8968 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
8969 param_type is not an integral or pointer type.
8971 2018-04-03 Richard Biener <rguenther@suse.de>
8973 * sese.h (recompute_all_dominators): Remove.
8975 2018-04-02 Martin Sebor <msebor@redhat.com>
8977 * doc/invoke.texi (-Wrestrict): Fix typos.
8979 2018-04-02 Jim Wilson <jimw@sifive.com>
8981 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
8982 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
8983 (<optab>di3, <optab>si3_extend): Likewise.
8984 (<optab>si3_mask, <optab>si3_mask_1): New.
8985 (<optab>di3_mask, <optab>di3_mask_1): New.
8986 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
8987 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
8988 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
8990 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
8992 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
8995 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
8997 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
8998 (nds32_canonicalize_comparison): New function.
9000 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
9001 Kito Cheng <kito.cheng@gmail.com>
9002 Kuan-Lin Chen <kuanlinchentw@gmail.com>
9004 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
9005 * config/nds32/constants.md (unspec_volatile_element): Add
9006 UNSPEC_VOLATILE_RELAX_GROUP.
9007 * config/nds32/nds32-relax-opt.c: New file.
9008 * config/nds32/nds32-predicates.c
9009 (nds32_symbol_load_store_p): New function.
9010 * config/nds32/nds32-protos.h
9011 (nds32_symbol_load_store_p): Declare function.
9012 (make_pass_nds32_relax_opt): Declare new rtl pass function.
9013 * config/nds32/nds32.c
9014 (nds32_register_pass): New function to register pass.
9015 (nds32_register_passes): New function to register passes.
9016 * config/nds32/nds32.md (relax_group): New pattern.
9017 * config/nds32/nds32.opt (mrelax-hint): New option.
9018 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
9020 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
9022 * config/nds32/t-nds32: Modify files dependency.
9024 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
9026 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
9027 (PROFILE_HOOK): Define its implementation.
9029 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
9031 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
9032 type and 32-bit size.
9034 2018-04-01 Jakub Jelinek <jakub@redhat.com>
9037 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
9038 (V_128_256): New mode iterator.
9039 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
9040 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
9041 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
9043 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
9044 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
9046 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
9049 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
9050 NaN inputs correctly.
9052 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
9055 * config/rs6000/vsx.md (??r): New mode attribute.
9056 (*vsx_mov<mode>_64bit): Use it.
9057 (*vsx_mov<mode>_32bit): Likewise.
9059 2018-03-30 Martin Sebor <msebor@redhat.com>
9061 PR tree-optimization/84818
9062 * builtins.c (check_access): Use warning_n.
9064 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9067 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
9069 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
9072 2018-03-30 Julia Koval <julia.koval@intel.com>
9075 * x86-tune.def (movx, partial_reg_dependency): Enable for
9078 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
9081 * lra-constraints.c (process_alt_operands): Move setting
9082 this_alternative_matches below.
9084 2018-03-29 Martin Liska <mliska@suse.cz>
9087 * doc/invoke.texi: Document how LTO works with debug info.
9088 Describe auto-load support of binutils. Mention 'x86-64'
9089 as valid option value of -march option.
9091 2018-03-29 Jakub Jelinek <jakub@redhat.com>
9093 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
9096 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
9097 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
9098 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
9101 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
9104 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
9105 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
9106 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
9107 for RS6000_BTM_POWERPC64.
9108 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
9109 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
9110 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
9115 2018-03-28 Carl Love <cel@us.ibm.com>
9118 2017-09-27 Carl Love <cel@us.ibm.com>
9120 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
9121 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
9122 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
9125 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9127 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
9128 instead of __vector bool.
9129 (_mm_max_pu8): Likewise.
9130 (_mm_min_pi16): Likewise.
9132 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
9135 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
9136 (DIVWEUO): Likewise.
9138 (DIVDEUO): Likewise.
9139 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
9140 DIVWEUO and DIVDEUO.
9141 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
9142 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
9143 (div_extend): Likewise.
9144 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
9146 (__builtin_divweuo): Likewise.
9147 (__builtin_divdeo): Likewise.
9148 (__builtin_divdeuo): Likewise.
9150 2018-03-28 Jakub Jelinek <jakub@redhat.com>
9153 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
9154 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
9156 PR tree-optimization/82004
9157 * gimple-match-head.c (optimize_pow_to_exp): New function.
9158 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
9159 Don't fold to exp if optimize_pow_to_exp is false.
9161 2018-03-28 Martin Liska <mliska@suse.cz>
9164 * calls.c (initialize_argument_information): Fix trailing space.
9165 * common.opt: Fix typo and provide better explanation for
9166 -fsanitize-coverage option.
9167 * config/i386/i386.opt: Fix typo.
9169 2018-03-28 Jakub Jelinek <jakub@redhat.com>
9170 Martin Liska <mliska@suse.cz>
9173 * gimplify.c (asan_poison_variable): Don't do the check for
9174 gimplify_omp_ctxp here.
9175 (gimplify_decl_expr): Do it here.
9176 (gimplify_target_expr): Likewise.
9178 2018-03-28 Martin Liska <mliska@suse.cz>
9181 * config/i386/i386.c (ix86_function_arg_advance): Do not call
9182 chkp_type_bounds_count if MPX is not enabled.
9184 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
9186 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
9188 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
9191 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
9192 function to create the function decl for complex long double
9193 multiply and divide for -mabi=ieeelongdouble.
9194 (init_float128_ieee): Call it.
9196 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
9199 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
9200 -fcf-protection=branch -mibt.
9201 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
9203 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9206 * config/arm/arm.c (arm_valid_symbolic_address): Handle
9207 arm_word_relocations.
9209 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
9212 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
9213 extern array declarations.
9215 2018-03-27 Richard Biener <rguenther@suse.de>
9218 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
9219 explicit single_use checks.
9221 2018-03-27 Richard Biener <rguenther@suse.de>
9223 PR tree-optimization/85082
9224 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
9227 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9229 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
9230 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
9231 Turn on fasynchronous-unwind-tables and funwind-tables.
9233 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
9236 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
9237 (*bmi_blsr_<mode>_ccz): Ditto.
9239 2018-03-26 Tom de Vries <tom@codesourcery.com>
9241 PR tree-optimization/85063
9242 * omp-general.c (offloading_function_p): New function. Factor out
9244 * omp-offload.c (pass_omp_target_link::gate): ... here.
9245 * omp-general.h (offloading_function_p): Declare.
9246 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
9247 with attribute omp declare target for offloading functions.
9249 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
9251 PR tree-optimization/84005
9252 * tree-data-ref.h (get_base_for_alignment): Declare.
9253 * tree-data-ref.c (get_base_for_alignment_1): New function.
9254 (get_base_for_alignment): Likewise.
9255 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
9256 get_base_for_alignment to find a suitable base object, instead
9257 of always using drb->base_address.
9259 2018-03-23 Jakub Jelinek <jakub@redhat.com>
9262 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
9263 known size by default.
9265 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
9268 * lra-constraints.c (process_alt_operands): Don't match BLKmode
9269 and non BLKmode operands.
9271 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9274 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
9275 Clean up attributes.
9277 2018-03-23 Richard Biener <rguenther@suse.de>
9280 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
9281 we are going to emit early debug for LTO.
9283 2018-03-23 Jakub Jelinek <jakub@redhat.com>
9286 * function.c (match_asm_constraints_1): Don't optimize if input
9287 doesn't satisfy general_operand predicate for output's mode.
9290 * alias.c (write_dependence_p): Don't require for x_canonicalized
9291 non-VOIDmode if x has VOIDmode.
9294 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
9295 just don't try to optimize it rather than assert it never happens.
9297 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
9299 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
9300 macro expansions for definition of ST_INTERNAL_<mode> and
9301 LD_INTERNAL_<mode> builtins.
9302 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
9304 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
9306 (altivec_expand_st_builtin): Likewise.
9307 (altivec_expand_builtin): Remove calls to deleted functions.
9308 (rs6000_address_for_altivec): Delete this function.
9309 * config/rs6000/vector.md: Remove expands for
9310 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
9312 2018-03-22 Sudakshina Das <sudi.das@arm.com>
9315 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
9316 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
9317 re-computing once computed.
9318 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
9319 (arm_init_machine_status): Initialize
9320 machine->static_chain_stack_bytes.
9322 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
9325 * doc/extend.texi: Add four new prototypes for vec_ld.
9326 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
9327 definitions for more logical presentation.
9328 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
9329 entries for V1TI variants of __builtin_altivec_ld builtin.
9330 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
9331 handling of V1TI variant of LVX icode pattern.
9332 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
9333 (rs6000_gimple_fold_builtin): Likewise.
9334 (altivec_init_builtins): Add code to define
9335 __builtin_altivec_lvx_v1ti function.
9337 2018-03-22 Jakub Jelinek <jakub@redhat.com>
9340 * function.c (match_asm_constraints_1): Don't do the optimization
9341 if input isn't a REG, SUBREG, MEM or constant.
9343 2018-03-22 Tom de Vries <tom@codesourcery.com>
9345 PR tree-optimization/84956
9346 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
9347 bb_has_abnormal_pred.
9349 2018-03-22 Jakub Jelinek <jakub@redhat.com>
9352 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
9353 DECL_INITIAL (decl) to decl at the end.
9354 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
9357 2018-03-21 Joseph Myers <joseph@codesourcery.com>
9359 * doc/extend.texi (__builtin_tgmath): Document when complex
9360 integer types are treated as _Complex _Float64.
9362 2018-03-21 Tom de Vries <tom@codesourcery.com>
9364 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
9366 2018-03-21 Jakub Jelinek <jakub@redhat.com>
9368 PR tree-optimization/84960
9369 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
9370 if it is ENTRY block, move them into single succ of ENTRY in that case.
9372 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
9374 PR tree-optimization/84811
9375 * poly-int.h (poly_span_traits): Remove the T3 parameter and
9376 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
9377 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
9378 (known_subrange_p): Update accordingly. Cast each value involved
9379 in the size comparison, rather than casting the result of the
9382 2018-03-21 Jakub Jelinek <jakub@redhat.com>
9384 PR tree-optimization/84982
9385 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
9386 by flipping the least significant bit rather than all bits from
9387 bitpos to bitpos + bitsize - 1.
9389 2018-03-21 Nathan Sidwell <nathan@acm.org>
9391 * doc/extend.texi (Deprecated Features): Remove mention of
9392 long-deleted deprecations.
9394 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9397 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
9398 * configure: Regenerate.
9400 2018-03-21 Tom de Vries <tom@codesourcery.com>
9402 PR tree-optimization/83126
9403 * tree-parloops.c (num_phis): New function.
9404 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
9406 2018-03-21 Nathan Sidwell <nathan@acm.org>
9408 * doc/extend.texi (Deprecated Features): Update deprecated flags,
9409 mention anon-struct/union members and trailing attributes.
9411 2018-03-21 Bin Cheng <bin.cheng@arm.com>
9413 PR tree-optimization/84969
9414 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
9415 builtin memset partitions if they set different rhs values.
9417 2018-03-21 Jakub Jelinek <jakub@redhat.com>
9419 PR rtl-optimization/84989
9420 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
9421 VEC_DUPLICATE with scalar result mode.
9423 2018-03-21 Martin Liska <mliska@suse.cz>
9426 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
9427 not intended return statement.
9429 2018-03-21 Martin Liska <mliska@suse.cz>
9432 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
9433 (chkp_find_bound_slots_1): Limit number of iterations.
9435 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
9438 * Minor grammar fixes for x86 options.
9440 2018-03-20 Jakub Jelinek <jakub@redhat.com>
9443 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
9444 holding REG_CFA_RESTORE notes, instead turn them into a USE.
9446 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
9449 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
9450 (altivec_lvx_<mode>_1op): Likewise.
9451 (altivec_stvx_<mode>_2op): Likewise.
9452 (altivec_stvx_<mode>_1op): Likewise.
9453 (altivec_lvx_<VM2:mode>): New define_expand.
9454 (altivec_stvx_<VM2:mode>): Likewise.
9455 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
9456 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
9457 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
9458 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
9459 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
9460 (rs6000_gen_lvx): Likewise.
9461 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
9462 (altivec_expand_stv_builtin): Likewise.
9463 (altivec_expand_builtin): Likewise.
9464 * config/rs6000/vector.md: Likewise.
9466 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9469 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
9472 2018-03-20 Richard Biener <rguenther@suse.de>
9475 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
9476 sign-conversions as zero, fall back to standard scalar_stmt
9479 2018-03-20 Martin Liska <mliska@suse.cz>
9482 * predict.c (rebuild_frequencies): Handle case when we have
9483 PROFILE_ABSENT, but flag_guess_branch_prob is false.
9485 2018-03-20 Jakub Jelinek <jakub@redhat.com>
9488 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
9489 flag_section_anchors.
9490 * varasm.c (use_blocks_for_decl_p): Remove hack for
9491 dw2_force_const_mem.
9494 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
9496 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
9497 be created, use lowpart_subreg of operands[0] rather than operands[0]
9499 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
9500 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
9501 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
9502 and n constraint instead of aarch64_shift_imm_di and Usd.
9503 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
9504 (*aarch64_<optab>_reg_minus<mode>3): ... this.
9506 2018-03-20 Sudakshina Das <sudi.das@arm.com>
9509 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
9510 to favor GPR over NEON registers.
9511 (<shift>di3_neon): Likewise.
9513 2018-03-20 Tom de Vries <tom@codesourcery.com>
9516 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
9517 (nvptx_process_pars): Emit bar.sync asap and alap.
9519 2018-03-20 Tom de Vries <tom@codesourcery.com>
9522 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
9523 seen_label if seen_label is already set.
9525 2018-03-20 Jakub Jelinek <jakub@redhat.com>
9528 * config/i386/i386.c (fold_builtin_cpu): For features above 31
9529 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
9530 Use 1U instead of 1. Formatting fixes.
9533 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
9534 instead of TREE_TYPE (s1) for the return value.
9536 2018-03-19 Jakub Jelinek <jakub@redhat.com>
9538 PR tree-optimization/84946
9539 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
9540 bitsize + bitsize in poly_uint64 rather than poly_int64.
9543 * dwarf2asm.c: Include fold-const.c.
9544 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
9545 of decl rather than decl itself.
9547 PR rtl-optimization/84643
9548 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
9550 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
9553 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
9554 calling assemble_variable.
9556 2018-03-19 Sudakshina Das <sudi.das@arm.com>
9559 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
9560 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
9562 2018-03-19 Jim Wilson <jimw@sifive.com>
9565 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
9566 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
9567 (riscv_first_stack_step): Likewise.
9568 (riscv_option_override): Use STACK_BOUNDARY instead of
9570 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
9572 (BIGGEST_ALIGNMENT): Set to 128.
9573 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
9574 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
9577 2018-03-19 Richard Biener <rguenther@suse.de>
9579 PR tree-optimization/84933
9580 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
9581 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
9583 2018-03-19 Richard Biener <rguenther@suse.de>
9585 PR tree-optimization/84859
9586 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
9587 (cond_if_else_store_replacement): Perform sinking operation on
9588 single-store BBs regardless of MAX_STORES_TO_SINK setting.
9589 Generalize what a BB with a single eligible store is.
9591 2018-03-19 Richard Biener <rguenther@suse.de>
9593 PR tree-optimization/84929
9594 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
9595 chrec_is_positive against non-chrec arg.
9597 2018-03-19 Tamar Christina <tamar.christina@arm.com>
9600 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
9602 2018-03-18 Martin Liska <mliska@suse.cz>
9604 PR rtl-optimization/84635
9605 * regrename.c (build_def_use): Use matches_mode only when
9608 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
9610 PR tree-optimization/84913
9611 * tree-vect-loop.c (vectorizable_reduction): Don't try to
9612 vectorize chains of COND_EXPRs.
9614 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
9616 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
9618 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
9620 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
9622 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
9624 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
9626 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
9627 Kito Cheng <kito.cheng@gmail.com>
9629 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
9630 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
9631 (nds32_adjust_reg_alloc_order): New function.
9632 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
9634 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
9636 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
9637 nds32_print_operand, nds32_print_operand_address): Use
9638 HOST_WIDE_INT_PRINT_DEC instead.
9640 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
9642 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
9644 2018-03-17 Jakub Jelinek <jakub@redhat.com>
9647 * config/i386/i386.c (initial_ix86_tune_features,
9648 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
9650 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
9651 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
9652 rather than 1u << ix86_tune. Formatting fix.
9653 (ix86_option_override_internal): Change ix86_arch_mask from
9654 unsigned int to unsigned HOST_WIDE_INT, initialize to
9655 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
9656 (ix86_function_specific_restore): Likewise.
9658 2018-03-16 Jakub Jelinek <jakub@redhat.com>
9661 * postreload.c (reload_combine_recognize_pattern): Perform
9662 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
9663 truncate_int_for_mode the result for the destination's mode.
9666 * hsa-gen.c (mem_type_for_type): Fix comment typo.
9667 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
9669 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
9672 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
9675 * lra-assigns.c (lra_split_hard_reg_for): Don't use
9676 regno_allocno_class_array and sorted_pseudos.
9677 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
9678 insns where regno is used.
9680 2018-03-16 Martin Liska <mliska@suse.cz>
9683 * multiple_target.c (create_dispatcher_calls): Redirect
9684 reference in the symbol table.
9686 2018-03-16 Martin Liska <mliska@suse.cz>
9689 * multiple_target.c (create_dispatcher_calls): Redirect also
9692 2018-03-16 Jakub Jelinek <jakub@redhat.com>
9696 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
9697 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
9698 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
9700 2018-03-16 Julia Koval <julia.koval@intel.com>
9702 * doc/invoke.texi (Skylake Server): Add CLWB.
9703 Cannonlake): Remove CLWB.
9705 2018-03-16 Jakub Jelinek <jakub@redhat.com>
9707 PR tree-optimization/84841
9708 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
9710 (FLOAT_ONE_CONST_TYPE): Define.
9711 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
9712 (sort_by_operand_rank): Put entries with higher constant_type last
9713 rather than first to match comments.
9715 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
9717 * config/nios2/nios2.md (movsi_internal): Fix thinko in
9720 2018-03-15 Jakub Jelinek <jakub@redhat.com>
9723 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
9724 check and use address of target always.
9726 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
9729 * config/i386/i386.c (indirect_thunk_needed): Update comments.
9730 (indirect_thunk_bnd_needed): Likewise.
9731 (indirect_thunks_used): Likewise.
9732 (indirect_thunks_bnd_used): Likewise.
9733 (indirect_return_needed): New.
9734 (indirect_return_bnd_needed): Likewise.
9735 (output_indirect_thunk_function): Add a bool argument for
9737 (output_indirect_thunk_function): Don't generate alias for
9738 function return thunk.
9739 (ix86_code_end): Call output_indirect_thunk_function to generate
9740 function return thunks.
9741 (ix86_output_function_return): Set indirect_return_bnd_needed
9742 and indirect_return_needed instead of indirect_thunk_bnd_needed
9743 and indirect_thunk_needed.
9745 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
9747 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
9748 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
9749 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
9751 2018-03-15 David Malcolm <dmalcolm@redhat.com>
9752 Paul Hua <paul.hua.gm@gmail.com>
9755 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
9757 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
9759 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
9760 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
9763 2018-03-15 Tamar Christina <tamar.christina@arm.com>
9766 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
9767 instead of GET_MODE_SIZE when comparing Units.
9769 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
9772 * varasm.c (hash_section): Return an unchangeble hash value
9773 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
9774 Return !aarch64_can_use_per_function_literal_pools_p ().
9776 2018-03-15 Jakub Jelinek <jakub@redhat.com>
9779 * optabs.c (emit_conditional_move): Pass address of cmode's copy
9780 rather than address of cmode as last argument to prepare_cmp_insn.
9782 2018-03-15 Julia Koval <julia.koval@intel.com>
9784 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
9785 F_AVX512VNNI, F_AVX512BITALG): New.
9787 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
9790 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
9791 insn for floating-point loads and stores.
9793 2018-03-14 Carl Love <cel@us.ibm.com>
9795 * config/rs6000/rs6000-c.c: Add macro definitions for
9796 ALTIVEC_BUILTIN_VEC_PERMXOR.
9797 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
9798 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
9799 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
9800 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
9802 * config/doc/extend.texi: Add prototypes for vec_permxor.
9804 2018-03-14 David Malcolm <dmalcolm@redhat.com>
9807 * diagnostic-show-locus.c (class layout_point): Convert m_line
9808 from int to linenum_type.
9809 (line_span::comparator): Use linenum "compare" function when
9810 comparing line numbers.
9811 (test_line_span): New function.
9812 (layout_range::contains_point): Convert param "row" from int to
9814 (layout_range::intersects_line_p): Likewise.
9815 (layout::will_show_line_p): Likewise.
9816 (layout::print_source_line): Likewise.
9817 (layout::should_print_annotation_line_p): Likewise.
9818 (layout::print_annotation_line): Likewise.
9819 (layout::print_leading_fixits): Likewise.
9820 (layout::annotation_line_showed_range_p): Likewise.
9821 (struct line_corrections): Likewise for field m_row.
9822 (line_corrections::line_corrections): Likewise for param "row".
9823 (layout::print_trailing_fixits): Likewise.
9824 (layout::get_state_at_point): Likewise.
9825 (layout::get_x_bound_for_row): Likewise.
9826 (layout::print_line): Likewise.
9827 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
9828 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
9829 * input.c (selftest::test_linenum_comparisons): New function.
9830 (selftest::input_c_tests): Call it.
9831 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
9832 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
9833 * selftest.h (ASSERT_GT): New macro.
9834 (ASSERT_GT_AT): New macro.
9835 (ASSERT_LT): New macro.
9836 (ASSERT_LT_AT): New macro.
9838 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
9840 PR rtl-optimization/84780
9841 * combine.c (distribute_links): Don't make a link based on pc_rtx.
9843 2018-03-14 Martin Liska <mliska@suse.cz>
9845 * tree.c (record_node_allocation_statistics): Use
9846 get_stats_node_kind.
9847 (get_stats_node_kind): New function extracted from
9848 record_node_allocation_statistics.
9849 (free_node): Use get_stats_node_kind.
9851 2018-03-14 Richard Biener <rguenther@suse.de>
9853 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
9854 that the value-set of ANTIC_IN doesn't grow.
9857 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
9859 (BB_VISITED_WITH_VISITED_SUCCS): New define.
9860 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
9862 2018-03-14 Julia Koval <julia.koval@intel.com>
9864 * config.gcc (icelake-client, icelake-server): New.
9866 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
9867 (initial_ix86_arch_features): Ditto.
9868 (PTA_SKYLAKE): Add SGX.
9869 (PTA_ICELAKE): Remove.
9870 (PTA_ICELAKE_CLIENT): New.
9871 (PTA_ICELAKE_SERVER): New.
9872 (ix86_option_override_internal): Split up icelake on icelake client and
9874 (get_builtin_code_for_version): Ditto.
9875 (fold_builtin_cpu): Ditto.
9876 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
9877 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
9878 * config/i386/i386.h (processor_type): Ditto.
9879 * doc/invoke.texi: Ditto.
9881 2018-03-14 Jakub Jelinek <jakub@redhat.com>
9884 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
9885 INTEGER_CST offset, add it together with bitpos / 8 and
9886 sign extend based on POINTER_SIZE.
9890 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
9893 * config/i386/constraints.md (Yc): New register constraint.
9894 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
9895 Use Yc constraint for alternative 2 of operand 0. Remove
9896 preferred_for_speed attribute.
9898 2018-03-14 Richard Biener <rguenther@suse.de>
9900 PR tree-optimization/84830
9901 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
9902 with the old one to avoid oscillations.
9904 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
9907 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
9909 (assign_by_spills): Return a flag of reload assignment failure.
9910 Do not process the reload assignment failures. Do not spill other
9911 reload pseudos if they has the same reg class. Update n if
9913 (lra_assign): Add a return arg. Set up from the result of
9914 assign_by_spills call.
9915 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
9916 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
9917 usage_insns if it is not NULL.
9918 (spill_hard_reg_in_range): New function.
9919 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
9920 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
9921 function prototypes.
9922 (lra_assign): Change prototype.
9923 * lra.c (lra): Add code to deal with fails by splitting hard reg
9926 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
9928 * config/riscv/riscv.opt (mrelax): New option.
9929 * config/riscv/riscv.c (riscv_file_start): Emit ".option
9930 "norelax" when riscv_mrelax is disabled.
9931 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
9933 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9936 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
9937 reassociation for int modes.
9939 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
9941 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
9942 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
9944 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
9945 * config/aarch64/aarch64-sve.md
9946 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
9947 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
9948 (*extend<mode><Vwide>2): Rename to...
9949 (aarch64_sve_extend<mode><Vwide>2): ...this.
9950 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
9951 renaming the old pattern to...
9952 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
9954 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
9955 define_expand, renaming the old pattern to...
9956 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
9957 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
9958 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
9959 account when deciding which SVE instruction the optab should use.
9960 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
9962 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
9964 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
9965 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
9966 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
9967 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
9968 (tlsdesc_small_<mode>): Turn a define_expand and use
9969 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
9970 (tlsdesc_small_advsimd_<mode>): ...this.
9971 (tlsdesc_small_sve_<mode>): New pattern.
9973 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
9975 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
9976 (UNSPEC_UMUL_HIGHPART): New constants.
9977 (MUL_HIGHPART): New int iteraor.
9978 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
9979 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
9981 (*<su>mul<mode>3_highpart): New define_insn.
9983 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
9986 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
9989 2018-03-13 Martin Liska <mliska@suse.cz>
9992 * (sem_item_optimizer::sem_item_optimizer): Initialize new
9994 (sem_item_optimizer::~sem_item_optimizer): Release it.
9995 (sem_item_optimizer::merge_classes): Register variable aliases.
9996 (sem_item_optimizer::fixup_pt_set): New function.
9997 (sem_item_optimizer::fixup_points_to_sets): Likewise.
9998 * ipa-icf.h: Declare new variables and functions.
10000 2018-03-13 Jakub Jelinek <jakub@redhat.com>
10002 PR middle-end/84834
10003 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
10004 integer_pow2p@2 and test integer_pow2p in condition.
10005 (A < 0 ? C : 0): Similarly for @1.
10007 PR middle-end/84831
10008 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
10009 characters starting at p contain '\0' character, don't look beyond
10013 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
10014 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
10017 * reg-stack.c (change_stack): Change update_end var from int to
10018 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
10019 also call set_block_for_insn on the newly added insns and rescan.
10022 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
10023 on the last operand.
10026 * tree.c (stabilize_reference_1): Return save_expr (e) for
10027 STATEMENT_LIST even if it doesn't have side-effects.
10029 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
10031 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
10033 2018-03-12 Renlin Li <renlin.li@arm.com>
10035 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
10036 aarch64_output_scalar_simd_mov_immediate.
10038 2018-03-12 Martin Sebor <msebor@redhat.com>
10040 PR tree-optimization/83456
10041 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
10042 for perfectly overlapping calls to memcpy.
10043 (gimple_fold_builtin_memory_chk): Same.
10044 (gimple_fold_builtin_strcpy): Handle no-warning.
10045 (gimple_fold_builtin_stxcpy_chk): Same.
10046 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
10048 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
10050 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
10051 parameter. Use it for SFmode.
10052 (rs6000_function_arg_advance_1): Adjust.
10053 (rs6000_function_arg): Adjust.
10054 (rs6000_gimplify_va_arg): Pass false for that new parameter.
10056 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
10058 PR rtl-optimization/84169
10059 PR rtl-optimization/84780
10060 * combine.c (can_combine_p): Check for a 2-insn combination whether
10061 the destination register is used between the two insns, too.
10063 2018-03-12 Richard Biener <rguenther@suse.de>
10065 PR tree-optimization/84803
10066 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
10067 for refs DR analysis didn't process.
10069 2018-03-12 Richard Biener <rguenther@suse.de>
10071 PR tree-optimization/84777
10072 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
10073 force-vectorize loops ignore whether we are optimizing for size.
10075 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
10077 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
10078 (TARGET_MD_ASM_ADJUST): Define.
10080 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
10081 Kito Cheng <kito.cheng@gmail.com>
10082 Chung-Ju Wu <jasonwucj@gmail.com>
10084 * config/nds32/nds32.c (nds32_compute_stack_frame,
10085 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
10086 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
10087 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
10088 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
10089 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
10090 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
10091 * config/nds32/nds32.md (prologue, epilogue): Use macro
10092 NDS32_V3PUSH_AVAILABLE_P to do checking.
10094 2018-03-11 Jakub Jelinek <jakub@redhat.com>
10097 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
10098 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
10099 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
10100 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
10101 addition of most attributes on !orig_type_die or the attribute not
10102 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
10104 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
10105 Chung-Ju Wu <jasonwucj@gmail.com>
10107 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
10108 __NDS32_VH__ macro.
10109 * config/nds32/nds32.opt (mvh): New option.
10111 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
10112 Chung-Ju Wu <jasonwucj@gmail.com>
10114 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
10116 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
10117 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
10120 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
10121 Chung-Ju Wu <jasonwucj@gmail.com>
10123 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
10125 * config/nds32/nds32-multiple.md (strlensi): New pattern.
10126 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
10128 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
10129 Kito Cheng <kito.cheng@gmail.com>
10130 Chung-Ju Wu <jasonwucj@gmail.com>
10132 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
10133 UNSPEC_FFMISM and UNSPEC_FLMISM.
10134 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
10135 for ffb, ffmism and flmism.
10136 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
10137 (unspec_ffmism): Ditto.
10138 (unspec_flmism): Ditto.
10139 (nds32_expand_builtin_impl): Check if string extension is available.
10140 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
10141 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
10143 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
10146 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
10149 * lra-assigns.c (assign_by_spills): Return a flag of reload
10150 assignment failure. Do not process the reload assignment
10151 failures. Do not spill other reload pseudos if they has the same
10153 (lra_assign): Add a return arg. Set up from the result of
10154 assign_by_spills call.
10155 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
10156 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
10157 usage_insns if it is not NULL.
10158 (spill_hard_reg_in_range): New function.
10159 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
10160 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
10161 function prototypes.
10162 (lra_assign): Change prototype.
10163 * lra.c (lra): Add code to deal with fails by splitting hard reg
10166 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
10169 * config/i386/i386.opt: Replace Enforcment with Enforcement.
10171 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
10174 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
10175 (dw_val_node): Add val_symbolic_view.
10176 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
10177 (symview_upper_bound): New.
10178 (new_line_info_table): Initialize symviews_since_reset.
10179 (dwarf2out_source_line): Count symviews_since_reset and set
10180 symview_upper_bound.
10181 (dw_val_equal_p): Handle symview.
10182 (add_AT_symview): New.
10183 (print_dw_val): Handle symview.
10184 (attr_checksum, attr_checksum_ordered): Likewise.
10185 (same_dw_val_p, size_of_die): Likewise.
10186 (value_format, output_die): Likewise.
10187 (add_high_low_attributes): Use add_AT_symview for entry_view.
10188 (dwarf2out_finish): Reset symview_upper_bound, clear
10191 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
10194 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
10195 Add strict argument and use it.
10196 (rs6000_split_multireg_move): Update for new strict argument.
10197 (mem_operand_gpr): Disallow all non-offsettable addresses.
10198 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
10200 2018-03-09 Jakub Jelinek <jakub@redhat.com>
10203 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
10204 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
10205 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
10208 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
10209 decl, use remap_type if we want to use the type.
10211 2018-03-09 Martin Sebor <msebor@redhat.com>
10213 PR tree-optimization/84526
10214 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
10216 (builtin_access::generic_overlap): Be prepared to handle non-array
10219 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
10221 PR rtl-optimization/84682
10222 * lra-constraints.c (process_address_1): Check is_address flag
10223 for address constraints.
10224 (process_alt_operands): Likewise.
10225 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
10226 preprocess_constraints.
10227 * recog.h (preprocess_constraints): Add oploc parameter.
10229 * recog.c (preprocess_constraints): Test address_operand for
10230 CT_ADDRESS constraints.
10232 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
10235 * lra-assigns.c (assign_by_spills): Return a flag of reload
10236 assignment failure. Do not process the reload assignment
10237 failures. Do not spill other reload pseudos if they has the same
10239 (lra_assign): Add a return arg. Set up from the result of
10240 assign_by_spills call.
10241 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
10242 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
10243 usage_insns if it is not NULL.
10244 (spill_hard_reg_in_range): New function.
10245 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
10246 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
10247 function prototypes.
10248 (lra_assign): Change prototype.
10249 * lra.c (lra): Add code to deal with fails by splitting hard reg
10252 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10255 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
10256 Accept complain bool parameter. Only emit errors if it is true.
10257 (arm_parse_cpu_option_name): Likewise.
10258 (arm_target_thumb_only): Adjust callers of the above.
10259 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
10260 prototype to take a default true bool parameter.
10261 (arm_parse_arch_option_name): Likewise.
10263 2018-03-09 David Malcolm <dmalcolm@redhat.com>
10264 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
10268 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
10269 * configure: Regenerate.
10270 * configure.ac ("linker --version-script option"): New.
10271 ("linker soname option"): New.
10273 2018-03-09 Richard Biener <rguenther@suse.de>
10275 PR tree-optimization/84775
10276 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
10277 immediate uses of predicate stmts and mark them modified.
10280 PR tree-optimization/84178
10281 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
10283 (version_loop_for_if_conversion): Delay update_ssa call.
10284 (tree_if_conversion): Delay update_ssa until after predicate
10287 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
10290 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
10291 when the function accesses prior frames.
10293 2018-03-08 Jakub Jelinek <jakub@redhat.com>
10296 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
10297 gen_llsym, otherwise call maybe_gen_llsym.
10299 PR inline-asm/84742
10300 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
10301 has ',' character inside of it.
10303 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10306 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
10307 as clobbering CC_REGNUM.
10309 2018-03-08 Richard Biener <rguenther@suse.de>
10311 PR middle-end/84552
10312 * tree-scalar-evolution.c: Include tree-into-ssa.h.
10313 (follow_copies_to_constant): Do not follow SSA names registered
10316 2018-03-08 Richard Biener <rguenther@suse.de>
10318 PR tree-optimization/84178
10319 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
10321 (version_loop_for_if_conversion): Delay update_ssa call.
10322 (tree_if_conversion): Delay update_ssa until after predicate
10325 2018-03-08 David Malcolm <dmalcolm@redhat.com>
10327 PR tree-optimization/84178
10328 * tree-if-conv.c (release_bb_predicate): Remove the
10329 the assertion that the stmts have NULL use_ops.
10330 Discard the statements, asserting that they haven't
10331 yet been added to a BB.
10333 2018-03-08 Richard Biener <rguenther@suse.de>
10335 PR tree-optimization/84746
10336 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
10337 (phi_translate): Pass in destination ANTIC_OUT set.
10338 (phi_translate_1): Likewise. For a simplified result lookup
10339 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
10340 (phi_translate_set): Adjust.
10341 (do_pre_regular_insertion): Likewise.
10342 (do_pre_partial_partial_insertion): Likewise.
10344 2018-03-08 Martin Liska <mliska@suse.cz>
10346 PR gcov-profile/84735
10347 * doc/gcov.texi: Document usage of profile files.
10348 * gcov-io.h: Document changes in the format.
10350 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
10354 * dwarf2out.c (struct dw_line_info_table): Update comments for
10356 (FORCE_RESET_NEXT_VIEW): New.
10357 (FORCE_RESETTING_VIEW_P): New.
10358 (RESETTING_VIEW_P): Check for -1 too.
10359 (ZERO_VIEW_P): Likewise.
10360 (new_line_info_table): Force-reset next view.
10361 (dwarf2out_begin_function): Likewise.
10362 (dwarf2out_source_line): Simplify zero_view_p initialization.
10363 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
10364 view directly. Omit view when omitting .loc at line 0.
10366 2018-03-08 Jakub Jelinek <jakub@redhat.com>
10368 PR tree-optimization/84740
10369 * tree-switch-conversion.c (process_switch): Call build_constructors
10370 only if info.phi_count is non-zero.
10372 PR tree-optimization/84739
10373 * tree-tailcall.c (find_tail_calls): Check call arguments against
10374 DECL_ARGUMENTS (current_function_decl) rather than
10375 DECL_ARGUMENTS (func) when checking for tail recursion.
10377 2018-03-07 Jakub Jelinek <jakub@redhat.com>
10379 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
10380 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
10381 Volker Reichelt's entry and add entries for people that perform
10382 GCC fuzzy testing and report numerous bugs.
10384 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
10387 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
10388 readonly data in sdata, if that is disabled.
10389 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
10390 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
10391 -mreadonly-in-sdata option.
10393 2018-03-07 Martin Sebor <msebor@redhat.com>
10395 PR tree-optimization/84468
10396 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
10397 basic block when looking for nul assignment.
10399 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
10402 * except.h (output_function_exception_table): Adjust prototype.
10403 * except.c (output_function_exception_table): Remove FNNAME parameter
10404 and add SECTION parameter. Ouput one part of the table at a time.
10405 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
10406 the first part of the exception table and emit unwind directives.
10407 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
10408 (i386_pe_seh_cold_init): Likewise.
10409 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
10410 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
10411 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
10412 (ix86_output_call_insn): Emit a nop in one more case for SEH.
10413 * config/i386/winnt.c: Include except.h.
10414 (struct seh_frame_state): Add reg_offset, after_prologue and
10415 in_cold_section fields.
10416 (i386_pe_seh_end_prologue): Set seh->after_prologue.
10417 (i386_pe_seh_cold_init): New function.
10418 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
10419 to seh->in_cold_section.
10420 (seh_emit_push): Record the offset of the push.
10421 (seh_emit_save): Record the offet of the save.
10422 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
10423 Test seh->after_prologue to disregard the epilogue.
10424 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
10425 (i386_pe_end_cold_function): New function.
10427 2018-03-07 Jakub Jelinek <jakub@redhat.com>
10430 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
10431 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
10434 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
10436 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
10437 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
10439 PR middle-end/84723
10440 * multiple_target.c: Include tree-inline.h and intl.h.
10441 (expand_target_clones): Diagnose and fail if node->definition and
10442 !tree_versionable_function_p (node->decl).
10444 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
10446 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
10448 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
10449 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10450 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
10452 2018-03-06 Jakub Jelinek <jakub@redhat.com>
10455 * combine.c (try_combine): Use reg_or_subregno instead of handling
10456 just paradoxical SUBREGs and REGs.
10458 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
10460 * config/arc/arc.c (arc_finalize_pic): Remove function.
10461 (arc_must_save_register): We use single base PIC register, remove
10462 checks to save/restore the PIC register.
10463 (arc_expand_prologue): Likewise.
10464 * config/arc/arc-protos.h (arc_set_default_type_attributes):
10466 (arc_verify_short): Likewise.
10467 (arc_attr_type): Likewise.
10468 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
10469 (walk_stores): Likewise.
10470 (arc_address_cost): Make it static.
10471 (arc_verify_short): Likewise.
10472 (branch_dest): Likewise.
10473 (arc_attr_type): Likewise.
10474 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
10475 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
10476 (arc_final_prescan_insn): Remove inserting the nops due to
10477 hardware hazards. It is done in reorg step.
10478 (insn_length_variant_t): Remove.
10479 (insn_length_parameters_t): Likewise.
10480 (arc_insn_length_parameters): Likewise.
10481 (arc_get_insn_variants): Likewise.
10482 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
10484 2018-03-06 Jakub Jelinek <jakub@redhat.com>
10486 PR inline-asm/84683
10487 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
10490 PR tree-optimization/84687
10491 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
10493 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
10495 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10497 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
10498 Rename to ppc_speculation_barrier.
10499 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
10500 __builtin_ppc_speculation_barrier.
10502 2018-03-05 Jakub Jelinek <jakub@redhat.com>
10505 * combine.c (combine_simplify_rtx): Don't try to simplify if
10506 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
10509 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
10511 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
10512 to 32 bytes when compiling for POWER9.
10514 2018-03-05 Jakub Jelinek <jakub@redhat.com>
10517 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
10518 regparm >= 3 with no arg reg available also for calls with
10519 flag_force_indirect_call. Pass decl to ix86_function_regparm.
10522 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
10524 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
10526 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
10529 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
10531 2018-03-05 Richard Biener <rguenther@suse.de>
10533 PR tree-optimization/84486
10534 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
10535 When inserting a __builtin_assume_aligned call set the LHS
10536 SSA name alignment info accordingly.
10538 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
10540 PR tree-optimization/84114
10541 * config/aarch64/aarch64.c (aarch64_reassociation_width)
10542 Avoid reassociation of FLOAT_MODE addition.
10544 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
10546 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
10547 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
10548 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
10549 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
10550 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
10551 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
10552 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
10554 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
10555 __builtin_ia32_wbinvd): New builtins.
10556 (SPECIAL_ARGS2): New.
10557 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
10558 (SPECIAL_ARGS2): New.
10559 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
10560 (ix86_valid_target_attribute_inner_p): Ditto.
10561 (ix86_init_mmx_sse_builtins): Add special_args2.
10562 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
10563 TARGET_WBNOINVD_P): New.
10564 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
10565 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
10566 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
10567 * config/i386/immintrin.h (_wbinvd): New intrinsic.
10568 * config/i386/pconfigintrin.h: New file.
10569 * config/i386/wbnoinvdintrin.h: Ditto.
10570 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
10572 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
10574 2018-03-05 Richard Biener <rguenther@suse.de>
10576 PR tree-optimization/84670
10577 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
10579 (BB_VISITED_WITH_VISITED_SUCCS): New define.
10580 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
10581 (compute_antic_aux): Only assert the number of values in ANTIC_IN
10582 doesn't grow if all successors (recursively) were visited at least
10585 2018-03-05 Richard Biener <rguenther@suse.de>
10587 PR tree-optimization/84650
10588 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
10589 if executed in the loop pipeline.
10591 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
10593 * doc/configfiles.texi (Configuration Files): Move info about
10594 conditionalizing $target-protos.h to...
10595 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
10596 differs from $target-protos.h.
10598 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
10599 Chung-Ju Wu <jasonwucj@gmail.com>
10601 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
10602 * config/nds32/nds32-multiple.md (setmemsi): Define.
10603 * config/nds32/nds32-memory-manipulation.c
10604 (nds32_gen_dup_4_byte_to_word_value): New.
10605 (emit_setmem_word_loop): New.
10606 (emit_setmem_byte_loop): New.
10607 (nds32_expand_setmem_loop): New.
10608 (nds32_expand_setmem_loop_v3m): New.
10609 (nds32_expand_setmem_unroll): New.
10610 (nds32_expand_setmem): New.
10612 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
10613 Chung-Ju Wu <jasonwucj@gmail.com>
10615 * config/nds32/nds32-memory-manipulation.c
10616 (nds32_emit_load_store): New.
10617 (nds32_emit_post_inc_load_store): New.
10618 (nds32_emit_mem_move): New.
10619 (nds32_emit_mem_move_block): New.
10620 (nds32_expand_movmemsi_loop_unknown_size): New.
10621 (nds32_expand_movmemsi_loop_known_size): New.
10622 (nds32_expand_movmemsi_loop): New.
10623 (nds32_expand_movmemsi_unroll): New.
10624 (nds32_expand_movmemqi): Rename ...
10625 (nds32_expand_movmemsi): ... to this.
10626 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
10627 (movmemsi): ... to this.
10628 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
10629 (nds32_expand_movmemsi): ... to this.
10631 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
10632 Monk Chiang <sh.chiang04@gmail.com>
10633 Chung-Ju Wu <jasonwucj@gmail.com>
10635 * config/nds32/nds32-protos.h
10636 (nds32_expand_load_multiple): New arguments.
10637 (nds32_expand_store_multiple): Ditto.
10638 (nds32_valid_multiple_load_store): Rename ...
10639 (nds32_valid_multiple_load_store_p): ... to this.
10640 * config/nds32/nds32-memory-manipulation.c
10641 (nds32_expand_load_multiple): Refine implementation.
10642 (nds32_expand_store_multiple): Ditto.
10643 * config/nds32/nds32-multiple.md
10644 (load_multiple): Update nds32_expand_load_multiple interface.
10645 (store_multiple): Update nds32_expand_store_multiple interface.
10646 * config/nds32/nds32-predicates.c
10647 (nds32_valid_multiple_load_store): Rename ...
10648 (nds32_valid_multiple_load_store_p): ... to this and refine
10650 * config/nds32/predicates.md
10651 (nds32_load_multiple_and_update_address_operation): New predicate.
10652 (nds32_store_multiple_and_update_address_operation): New predicate.
10654 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
10655 Chung-Ju Wu <jasonwucj@gmail.com>
10657 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
10658 (combo): New attribute.
10659 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
10661 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
10663 * config/nds32/nds32.opt: Change -mcmodel= default value.
10665 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
10666 Monk Chiang <sh.chiang04@gmail.com>
10667 Chung-Ju Wu <jasonwucj@gmail.com>
10669 * config/nds32/constants.md (unspec_element): New enum.
10670 * config/nds32/constraints.md (Umw): New constraint.
10671 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
10672 * config/nds32/nds32-intrinsic.md: Likewise.
10673 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
10674 (nds32_valid_smw_lwm_base_p): New.
10675 (nds32_output_smw_single_word): New.
10676 (nds32_output_lmw_single_word): New.
10677 (nds32_expand_unaligned_load): New.
10678 (nds32_expand_unaligned_store): New.
10679 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
10680 (nds32_output_smw_single_word): Declare.
10681 (nds32_output_lmw_single_word): Declare.
10682 (nds32_expand_unaligned_load): Declare.
10683 (nds32_expand_unaligned_store): Declare.
10684 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
10685 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
10686 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
10687 NDS32_BUILTIN_UASTORE_DW.
10688 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
10691 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
10692 Kito Cheng <kito.cheng@gmail.com>
10693 Chung-Ju Wu <jasonwucj@gmail.com>
10695 * config/nds32/nds32-intrinsic.c
10696 (nds32_expand_builtin_null_ftype_reg): Delete.
10697 (nds32_expand_builtin_reg_ftype_imm): Ditto.
10698 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
10699 (nds32_read_argument): New.
10700 (nds32_legitimize_target): Ditto.
10701 (nds32_legitimize_argument): Ditto.
10702 (nds32_check_constant_argument): Ditto.
10703 (nds32_expand_unop_builtin): Ditto.
10704 (nds32_expand_unopimm_builtin): Ditto.
10705 (nds32_expand_binop_builtin): Ditto.
10706 (nds32_builtin_decl_impl): Ditto.
10707 (builtin_description): Ditto.
10708 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
10709 (nds32_init_builtins_impl): Ditto.
10710 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
10711 (nds32_builtin_decl): New.
10712 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
10713 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
10715 2018-03-02 Jeff Law <law@redhat.com>
10717 * reorg.c (stop_search_p): Handle DEBUG_INSN.
10718 (redundant_insn, fill_simple_delay_slots): Likewise.
10719 (fill_slots_from_thread): Likewise.
10720 * resource.c (mark_referenced_resources): Likewise.
10721 (mark_set_resources, find_dead_or_set_registers): Likewise.
10723 2018-03-02 Jakub Jelinek <jakub@redhat.com>
10725 * substring-locations.h (format_warning_va): Formatting fix for
10726 ATTRIBUTE_GCC_DIAG.
10727 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
10729 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
10730 * substring-locations.c: Include intl.h.
10731 (format_warning_va): Turned into small wrapper around
10732 format_warning_n_va, renamed to ...
10733 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
10734 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
10736 (format_warning_at_substring_n): New function.
10737 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
10738 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
10739 format_warning_at_substring with just a shorter name instead of
10740 const function pointer.
10741 (fmtwarn_n): New function.
10742 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
10743 appropriate, get rid of all the fmtstr temporaries, move conditionals
10744 with G_() wrapped string literals directly into fmtwarn arguments,
10745 cast dir.len to (int), formatting fixes.
10747 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
10749 * doc/invoke.texi: Remove "Cilk Plus" references.
10751 2018-03-02 Jakub Jelinek <jakub@redhat.com>
10752 Richard Biener <rguenther@suse.de>
10755 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
10756 for error or warning attributes if CALL_FROM_THUNK_P is set.
10759 2018-03-02 Jakub Jelinek <jakub@redhat.com>
10762 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
10763 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
10766 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
10767 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
10769 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
10770 instead of -1U in last predictors element's probability member.
10772 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
10775 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
10776 arguments if they are comparable.
10778 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
10780 PR tree-optimization/84634
10781 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
10782 masks and masked_loop_p with a single loop_masks, making sure it's
10783 null for bb vectorization.
10785 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
10787 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
10788 (vect_analyze_data_ref_access): Use loop->safe_len rather than
10789 loop->force_vectorize to check whether there is no alias.
10791 2018-03-02 Jakub Jelinek <jakub@redhat.com>
10794 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
10796 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
10798 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
10799 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
10800 instead of a loop around prev_real_insn.
10801 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
10804 PR inline-asm/84625
10805 * config/i386/i386.c (ix86_print_operand): Use conditional
10806 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
10809 2018-03-02 Richard Biener <rguenther@suse.de>
10811 PR tree-optimization/84427
10812 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
10813 (bitmap_set_subtract_values): Rewrite to handle multiple
10816 (prune_clobbered_mems): Likewise.
10817 (phi_translate): Take edge instead of pred/phiblock.
10818 (phi_translate_1): Likewise.
10819 (phi_translate_set): Likewise. Insert all translated
10820 exprs for a value into the set, keeping possibly multiple
10821 expressions per value.
10822 (compute_antic_aux): Adjust for phi_translate changes.
10823 When intersecting union the expressions and prune those
10824 not in the final value set, keeping possibly multiple
10825 expressions per value. Do not use value-insertion
10826 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
10827 all expressions. Add verification that the value-sets
10828 only shrink during iteration.
10829 (compute_partial_antic_aux): Adjust for the phi_translate changes.
10830 (do_pre_regular_insertion): Likewise.
10831 (do_pre_partial_partial_insertion): Likewise.
10833 2018-03-02 Richard Biener <rguenther@suse.de>
10836 * config/darwin.c (saved_debug_info_level): New static global.
10837 (darwin_asm_lto_start): Disable debug info generation for LTO out.
10838 (darwin_asm_lto_end): Restore debug info generation settings.
10840 2018-03-01 Martin Liska <mliska@suse.cz>
10843 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
10844 volatile arguments.
10846 2018-03-01 Richard Biener <rguenther@suse.de>
10849 * dwarf2out.c (gen_variable_die): Properly handle late VLA
10850 type annotation with LTO when debug was disabled at compile-time.
10852 2018-03-01 Matthew Fortune <mfortune@gmail.com>
10854 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
10856 (mips_final_postscan_insn): Likewise.
10858 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
10860 PR rtl-optimization/84528
10861 * alias.c (init_alias_target): Add commentary.
10862 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
10863 a unique base value if the frame pointer is not eliminated
10864 to the stack pointer.
10866 2018-03-01 Tom de Vries <tom@codesourcery.com>
10868 PR rtl-optimization/83327
10869 * lra-int.h (hard_regs_spilled_into): Declare.
10870 * lra.c (hard_regs_spilled_into): Define.
10871 (init_reg_info): Init hard_regs_spilled_into.
10872 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
10873 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
10874 (process_bb_lives): Handle hard_regs_spilled_into.
10875 (lra_create_live_ranges_1): Before doing liveness propagation, clear
10876 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
10878 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
10880 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
10881 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
10882 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
10883 * config/rs6000/aix72.h: New file.
10885 2018-02-28 Jakub Jelinek <jakub@redhat.com>
10887 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
10888 instead of warning_at with conditional singular and plural messages
10892 * stor-layout.c (update_alignment_for_field): For
10893 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
10894 && !DECL_PACKED (field), do the alignment update, just use
10895 only desired_align instead of MAX (type_align, desired_align)
10897 (place_field): Don't do known_align < desired_align handling
10898 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
10899 is non-NULL, instead do it after rli->prev_field handling and
10900 only if not within a bitfield word. For DECL_PACKED (field)
10901 use type_align of BITS_PER_UNIT.
10903 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
10905 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
10906 superfluous parentheses and trailing spaces.
10908 2018-02-28 Richard Biener <rguenther@suse.de>
10910 PR tree-optimization/84584
10911 * graphite-scop-detection.c (scop_detection::add_scop): Discard
10912 SCoPs with fake exit edge.
10914 2018-02-28 Martin Liska <mliska@suse.cz>
10917 * timevar.c (timer::print): Fix format to properly print 100%
10920 2018-02-28 Richard Biener <rguenther@suse.de>
10922 PR middle-end/84607
10923 * genmatch.c (capture_info::walk_match): Do not mark
10924 captured expressions without operands as expr_p given
10925 they act more like predicates and should be subject to
10926 "lost tail" side-effect preserving.
10928 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
10930 PR rtl-optimization/81611
10931 * auto-inc-dec.c (attempt_change): Move dead note from
10932 mem_insn if it's the next use of regno
10933 (find_address): Take address use of reg holding
10934 non-incremented value. Add parm to limit search to the named
10936 (merge_in_block): Attempt to use a mem insn that is the next
10937 use of the original regno.
10939 2018-02-27 Martin Sebor <msebor@redhat.com>
10942 * gcc/doc/invoke.texi (-Wmissing-attributes): New option.
10943 * gcc/print-tree.c (print_node): Handle DECL_UNINLINABLE.
10945 2018-02-27 Martin Sebor <msebor@redhat.com>
10947 PR translation/84207
10948 * diagnostic-core.h (warning_n, error_n, inform_n): Change
10949 n argument to unsigned HOST_WIDE_INT.
10950 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
10951 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
10952 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
10953 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
10955 2018-02-27 Richard Biener <rguenther@suse.de>
10957 PR tree-optimization/84512
10958 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
10959 Do not use the estimate returned from record_stmt_cost for
10960 the scalar iteration cost but sum properly using add_stmt_cost.
10962 2018-02-27 Richard Biener <rguenther@suse.de>
10964 PR tree-optimization/84466
10965 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
10966 Adjust last change to less strictly validate use operands.
10968 2018-02-27 Martin Liska <mliska@suse.cz>
10970 PR gcov-profile/84548
10971 * gcov.c (process_file): Allow partial overlap and consider it
10972 also as group functions.
10973 (output_lines): Properly calculate range of lines for a group.
10975 2018-02-27 Martin Liska <mliska@suse.cz>
10977 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
10978 'ggc' suffixes. Change first column width.
10979 (timer::print): Fix formatting of the column.
10981 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
10983 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
10984 preserve inline entry blocks for the sake of debug inline
10985 entry point markers alone.
10986 (remove_unused_locals): Suggest in comments a better place to
10987 force the preservation of inline entry blocks that are
10988 otherwise unused, but do not preserve them.
10990 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
10992 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
10994 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
10997 * config/i386/constraints.md (Bs): Replace
10998 ix86_indirect_branch_register with
10999 TARGET_INDIRECT_BRANCH_REGISTER.
11001 * config/i386/i386.md (indirect_jump): Likewise.
11002 (tablejump): Likewise.
11003 (*sibcall_memory): Likewise.
11004 (*sibcall_value_memory): Likewise.
11005 Peepholes of indirect call and jump via memory: Likewise.
11006 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
11007 (*sibcall_value_GOT_32): Likewise.
11008 * config/i386/predicates.md (indirect_branch_operand): Likewise.
11009 (GOT_memory_operand): Likewise.
11010 (call_insn_operand): Likewise.
11011 (sibcall_insn_operand): Likewise.
11012 (GOT32_symbol_operand): Likewise.
11013 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
11015 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
11017 PR rtl-optimization/83496
11018 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
11019 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
11020 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
11021 redundant insn, if any.
11022 (relax_delay_slots): Likewise.
11023 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
11025 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
11027 PR tree-optimization/83965
11028 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
11029 that grouped statements are part of a reduction chain. Return
11030 true if the statement is not marked as a reduction itself but
11031 is part of a group.
11032 (vect_recog_dot_prod_pattern): Don't check whether the statement
11033 is part of a group here.
11034 (vect_recog_sad_pattern): Likewise.
11035 (vect_recog_widen_sum_pattern): Likewise.
11037 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
11040 * final.c (rest_of_clean_state): Also look for calls inside sequences.
11042 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
11045 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
11047 (ix86_output_indirect_function_return): New prototype.
11048 (ix86_split_simple_return_pop_internal): Likewise.
11049 * config/i386/i386.c (indirect_return_via_cx): New.
11050 (indirect_return_via_cx_bnd): Likewise.
11051 (indirect_thunk_name): Handle return va CX_REG.
11052 (output_indirect_thunk_function): Create alias for
11053 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
11054 (ix86_output_indirect_jmp): Remove the bool argument.
11055 (ix86_output_indirect_function_return): New function.
11056 (ix86_split_simple_return_pop_internal): Likewise.
11057 * config/i386/i386.md (*indirect_jump): Don't pass false
11058 to ix86_output_indirect_jmp.
11059 (*tablejump_1): Likewise.
11060 (simple_return_pop_internal): Change it to define_insn_and_split.
11061 Call ix86_split_simple_return_pop_internal to split it for
11062 -mfunction-return=.
11063 (simple_return_indirect_internal): Call
11064 ix86_output_indirect_function_return instead of
11065 ix86_output_indirect_jmp.
11067 2018-02-26 Jakub Jelinek <jakub@redhat.com>
11070 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
11071 memset and value initialization afterwards.
11073 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
11075 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
11077 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11080 * common/config/aarch64/aarch64-common.c
11081 (aarch_option_optimization_table[]): Switch
11082 off fomit-frame-pointer
11084 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
11085 Chung-Ju Wu <jasonwucj@gmail.com>
11087 * config/nds32/nds32-multiple.md (load_multiple): Disallow
11089 (store_multiple): Ditto.
11091 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
11093 * config.gcc: Add --with-cpu support for nds32 target.
11094 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
11095 * config/nds32/nds32.opt: Add -mcpu= option.
11097 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
11099 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
11100 isel=yes): Warn for these deprecated options.
11102 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
11104 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
11105 ISA_2_5_MASKS_EMBEDDED.
11107 2018-02-23 Jakub Jelinek <jakub@redhat.com>
11109 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
11110 p->max as pointers rather than using iterative_hash_expr.
11112 2018-02-23 Carl Love <cel@us.ibm.com>
11114 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
11115 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
11117 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
11118 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
11119 P8V_BUILTIN_VEC_VUNSIGNED2.
11121 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
11124 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
11125 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
11127 * lra-constraints.c (curr_insn_transform): Set up
11128 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
11130 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
11131 * lra-eliminations.c (spill_pseudos): Ditto.
11132 (process_insn_for_elimination): Ditto.
11133 * lra-lives.c (reg_early_clobber_p): Use the new macros.
11134 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
11135 LRA_NON_CLOBBERED_ALT.
11137 2018-02-22 Martin Sebor <msebor@redhat.com>
11139 PR tree-optimization/84480
11140 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
11141 to maybe_diag_stxncpy_trunc. Call it.
11142 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
11143 from gimple_fold_builtin_strcpy. Print inlining stack.
11144 (handle_builtin_stxncpy): Print inlining stack.
11145 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
11147 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
11150 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
11151 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
11152 and -fcheck-pointer-bounds are used together.
11153 (indirect_thunk_prefix): New enum.
11154 (indirect_thunk_need_prefix): New function.
11155 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
11156 "_nt" instead of "_bnd" for NOTRACK prefix.
11157 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
11158 (output_indirect_thunk_function): Likewise.
11160 (ix86_code_end): Update output_indirect_thunk_function calls.
11161 (ix86_output_indirect_branch_via_reg): Replace
11162 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
11163 (ix86_output_indirect_branch_via_push): Likewise.
11164 (ix86_output_function_return): Likewise.
11165 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
11166 incompatible with -fcf-protection=branch and
11167 -fcheck-pointer-bounds.
11169 2018-02-22 Steve Ellcey <sellcey@cavium.com>
11172 * config/aarch64/aarch64.c (aarch64_print_address_internal):
11173 Change gcc_assert call to output_operand_lossage.
11175 2018-02-22 Steve Ellcey <sellcey@cavium.com>
11177 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
11179 2018-02-22 DJ Delorie <dj@redhat.com>
11180 Sebastian Perta <sebastian.perta@renesas.com>
11181 Oleg Endo <olegendo@gcc.gnu.org>
11183 * config/rx/rx.c (rx_rtx_costs): New function.
11184 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
11186 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
11188 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
11190 2018-02-22 Martin Liska <mliska@suse.cz>
11193 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
11194 Add "native" as a possible value.
11196 2018-02-22 Martin Liska <mliska@suse.cz>
11199 * config/i386/i386.c (ix86_option_override_internal):
11200 Add "native" as a possible value for -march and -mtune.
11202 2018-02-22 Jakub Jelinek <jakub@redhat.com>
11205 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
11206 to all type variants.
11208 PR tree-optimization/84503
11209 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
11210 width as info->bitpos + info->bitsize - start.
11211 (merged_store_group::merge_overlapping): Simplify width computation.
11212 (check_no_overlap): New function.
11213 (imm_store_chain_info::try_coalesce_bswap): Compute expected
11214 start + width and last_order of the group, fail if check_no_overlap
11216 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
11217 to group if check_no_overlap fails.
11219 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
11221 * config/rs6000/altivec.md: Delete contraint arguments to
11222 define_expand, define_split, and define_peephole2, and in
11223 define_insn_and_split if always unused.
11224 * config/rs6000/darwin.md: Ditto.
11225 * config/rs6000/dfp.md: Ditto.
11226 * config/rs6000/rs6000.md: Ditto.
11227 * config/rs6000/sync.md: Ditto.
11228 * config/rs6000/vector.md: Ditto.
11229 * config/rs6000/vsx.md: Ditto.
11231 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
11233 * config/rs6000/altivec.md: Write output control strings as braced
11234 blocks instead of double-quoted strings.
11235 * config/rs6000/darwin.md: Ditto.
11236 * config/rs6000/rs6000.md: Ditto.
11237 * config/rs6000/vector.md: Ditto.
11238 * config/rs6000/vsx.md: Ditto.
11240 2018-02-21 Jason Merrill <jason@redhat.com>
11242 PR c++/84314 - ICE with templates and fastcall attribute.
11243 * attribs.c (build_type_attribute_qual_variant): Remove assert.
11245 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
11247 * ipa-cp.c (determine_versionability): Fix comment typos.
11249 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
11252 * ipa-cp.c (determine_versionability): Do not version functions caling
11255 2018-02-21 Martin Liska <mliska@suse.cz>
11258 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
11259 Add "native" as a possible value.
11260 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
11261 the macro when native cpu detection is available.
11263 2018-02-21 Martin Liska <mliska@suse.cz>
11266 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
11267 Add "native" as a possible value.
11268 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
11269 when native cpu detection is available.
11271 2018-02-21 Jakub Jelinek <jakub@redhat.com>
11272 Martin Sebor <msebor@redhat.com>
11274 PR tree-optimization/84478
11275 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
11277 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
11278 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
11279 support which is conservatively correct, for 2 only stay conservative
11280 for maxlen. Formatting and comment capitalization fixes. Add STRICT
11281 argument to the 2 argument get_range_strlen, adjust 6 arg
11282 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
11284 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
11285 (gimple_fold_builtin_strlen): Pass true as last argument to
11288 2018-02-20 Martin Sebor <msebor@redhat.com>
11290 PR middle-end/84095
11291 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
11292 (builtin_memref::set_base_and_offset): Same. Handle inner references.
11293 (builtin_memref::builtin_memref): Factor out parts into
11294 set_base_and_offset and call it.
11296 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
11298 PR middle-end/84406
11299 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
11300 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
11301 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
11302 search at the associated MODE_INT.
11304 2018-02-20 Jeff Law <law@redhat.com>
11306 PR middle-end/82123
11307 PR tree-optimization/81592
11308 PR middle-end/79257
11309 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
11310 for range data rather than using global data.
11311 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
11312 range data rather than using global data.
11313 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
11314 pass it to children as needed.
11315 (struct directive::fmtresult): Similarly.
11316 (struct directive::set_width): Similarly.
11317 (struct directive::set_precision): Similarly.
11318 (format_integer, format_directive, parse_directive): Similarly.
11319 (format_none): Accept unnamed vr_values parameter.
11320 (format_percent, format_floating, format_character): Similarly.
11321 (format_string, format_plain): Similarly.
11322 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
11323 the EVRP range analyzer for range data rather than using global data.
11324 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
11325 gimple-ssa-evrp-analyze.h
11326 (class sprintf_dom_walker): Add after_dom_children member function.
11327 Add evrp_range_analyzer member.
11328 (sprintf_dom_walker::before_dom_children): Call into the EVRP
11329 range analyzer as needed.
11330 (sprintf_dom_walker::after_dom_children): New member function.
11331 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
11333 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
11334 (evrp_range_analyzer::pop_to_marker): Likewise.
11336 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
11338 PR tree-optimization/84419
11339 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
11340 with the required type if its current type is compatible but
11343 2018-02-20 Jakub Jelinek <jakub@redhat.com>
11345 PR middle-end/82004
11346 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
11347 after vectorization.
11349 2018-02-20 Martin Liska <mliska@suse.cz>
11352 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
11353 possible values if we don't have a hint.
11355 2018-02-20 Martin Liska <mliska@suse.cz>
11359 * final.c (shorten_branches): Build align_tab array with one
11361 * opts.c (finish_options): Add alignment option limit check.
11362 (MAX_CODE_ALIGN): Likewise.
11363 (MAX_CODE_ALIGN_VALUE): Likewise.
11364 * doc/invoke.texi: Document maximum allowed option value for
11365 all -falign-* options.
11367 2018-02-19 Jakub Jelinek <jakub@redhat.com>
11370 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
11371 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
11372 * var-tracking.c (emit_note_insn_var_location): Remove all references
11373 to NOTE_INSN_CALL_ARG_LOCATION.
11374 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
11375 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
11376 Use copy_rtx_if_shared.
11377 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
11378 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
11379 (dwarf2out_var_location): Remove handling of
11380 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
11382 * final.c (final_scan_insn): Remove all references to
11383 NOTE_INSN_CALL_ARG_LOCATION.
11384 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
11385 before dumping final insns.
11386 * except.c (emit_note_eh_region_end): Remove all references to
11387 NOTE_INSN_CALL_ARG_LOCATION.
11388 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
11389 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
11390 * config/arc/arc.c (hwloop_optimize): Likewise.
11391 * config/arm/arm.c (create_fix_barrier): Likewise.
11392 * config/s390/s390.c (s390_chunkify_start): Likewise.
11393 * config/sh/sh.c (find_barrier): Likewise.
11394 * config/i386/i386.c (rest_of_insert_endbranch,
11395 ix86_seh_fixup_eh_fallthru): Likewise.
11396 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
11397 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
11398 * config/frv/frv.c (frv_function_prologue): Likewise.
11399 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
11401 (note_outside_basic_block_p): Remove all references to
11402 NOTE_INSN_CALL_ARG_LOCATION.
11403 * gengtype.c (adjust_field_rtx_def): Likewise.
11404 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
11406 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
11407 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
11410 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
11413 PR tree-optimization/84452
11414 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
11415 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
11418 2018-02-19 Martin Liska <mliska@suse.cz>
11421 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
11423 2018-02-19 Martin Liska <mliska@suse.cz>
11424 Richard Sandiford <richard.sandiford@linaro.org>
11426 PR tree-optimization/82491
11427 * gimple-fold.c (get_base_constructor): Make earlier bail out
11430 2018-02-19 Carl Love <cel@us.ibm.com>
11432 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
11433 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
11435 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
11436 P8V_BUILTIN_VEC_NEG.
11438 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
11440 * config/rl78/rl78.md (movdf): New define expand.
11442 2018-02-19 Martin Liska <mliska@suse.cz>
11445 * doc/invoke.texi: Fix typo.
11446 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
11448 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
11450 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
11451 handle rs6000_single_float and rs6000_double_float specially for
11454 2018-02-16 Jeff Law <law@redhat.com>
11456 * config/rx/rx.c (add_pop_cfi_notes): New function.;
11457 (pop_regs): Use it.
11459 2018-02-16 Jakub Jelinek <jakub@redhat.com>
11462 * ipa-inline.c (inline_small_functions): Fix a typo.
11464 2018-02-16 Nathan Sidwell <nathan@acm.org>
11466 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
11468 2018-02-16 Carl Love <cel@us.ibm.com>
11470 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
11471 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
11472 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
11473 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
11474 expansion to P8V_BUILTIN_VEC_FLOAT2.
11476 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
11478 PR rtl-optimization/70023
11479 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
11480 src_regno into account.
11482 2018-02-16 Carl Love <cel@us.ibm.com>
11484 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
11485 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
11486 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
11487 * config/rs6000/rs6000.c: Remove case statements for
11488 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
11489 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
11490 and P9V_BUILTIN_VEC_VINSERT4B.
11491 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
11492 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
11493 * config/rs6000/vsx.md:
11494 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
11497 2018-02-16 Carl Love <cel@us.ibm.com>
11499 * config/rs6000/altivec.h: Add builtin names vec_extract4b
11501 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
11503 * config/rs6000/rs6000-c.c: Add the definitions for
11504 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
11505 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
11506 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
11507 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
11508 definition for insert4b and define insn *insert3b_internal.
11509 * doc/extend.texi: Add documentation for vec_extract4b.
11511 2018-02-16 Nathan Sidwell <nathan@acm.org>
11513 * doc/extend.texi (Backwards Compatibility): Mention friend
11514 injection. Note for-scope is deprecated.
11515 * doc/invoke.texi (-ffriend-injection): Deprecate.
11517 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
11519 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
11520 that moved to I2, also allow destinations that are a paradoxical
11521 subreg (instead of a normal reg).
11523 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
11526 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
11529 2018-02-16 Richard Biener <rguenther@suse.de>
11531 PR tree-optimization/84037
11532 PR tree-optimization/84016
11534 * config/i386/i386.c (ix86_builtin_vectorization_cost):
11535 Adjust vec_construct for the fact we need additional higher latency
11536 128bit inserts for AVX256 and AVX512 vector builds.
11537 (ix86_add_stmt_cost): Scale vector construction cost for
11540 2018-02-16 Richard Biener <rguenther@suse.de>
11542 PR tree-optimization/84417
11543 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
11544 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
11545 (non_rewritable_lvalue_p): Likewise, use poly-ints.
11547 2018-02-16 Martin Liska <mliska@suse.cz>
11550 * internal-fn.def (ASAN_CHECK): Set proper flags.
11551 (ASAN_MARK): Likewise.
11553 2018-02-16 Julia Koval <julia.koval@intel.com>
11555 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
11556 from PTA_CANNONLAKE.
11558 2018-02-16 Jakub Jelinek <jakub@redhat.com>
11561 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
11562 Use ++iter rather than iter++ for std::list iterators.
11563 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
11564 defer deleting them until all nodes in the forest are processed. Do
11565 free even leaf nodes. Change to_process into auto_vec.
11568 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
11569 * vec.h (vec_default_construct): Use memset instead of placement new
11570 if BROKEN_VALUE_INITIALIZATION is defined.
11571 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
11572 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
11575 PR rtl-optimization/83723
11576 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
11577 * lra.c (lra_substitute_pseudo): Likewise. If true, use
11578 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
11580 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
11582 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
11584 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
11586 PR rtl-optimization/81443
11587 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
11588 from inner REGs to paradoxical SUBREGs.
11590 2018-02-16 Richard Biener <rguenther@suse.de>
11592 PR tree-optimization/84399
11593 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
11594 For operands we can analyze at their definition make sure we can
11595 analyze them at each use as well.
11597 2018-02-16 Richard Biener <rguenther@suse.de>
11599 PR tree-optimization/84190
11600 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
11601 volatile accesses if the decl isn't volatile.
11603 2018-02-15 Jason Merrill <jason@redhat.com>
11605 PR c++/84314 - ICE with templates and fastcall attribute.
11606 * attribs.c (build_type_attribute_qual_variant): Don't clobber
11607 TYPE_CANONICAL on an existing type.
11609 2018-02-15 Jakub Jelinek <jakub@redhat.com>
11611 PR tree-optimization/84383
11612 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
11613 dstoff nor call operand_equal_p if dstbase is NULL.
11615 PR tree-optimization/84334
11616 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
11617 also a CONSTANT_CLASS_P, punt.
11619 2018-02-14 Jim Wilson <jimw@sifive.com>
11621 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
11622 first SMALL_OPERAND check. New local min_second_step. Move assert
11623 to where locals are set. Add TARGET_RVC support.
11624 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
11626 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
11628 * doc/invoke.texi: Correct -Wformat-overflow code sample.
11630 2018-02-14 Martin Sebor <msebor@redhat.com>
11632 PR tree-optimization/83698
11633 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
11634 arrays constrain the offset range to their bounds.
11635 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
11636 (builtin_access::overlap): Avoid setting the size of overlap if it's
11638 (maybe_diag_overlap): Also consider arrays when deciding what values
11639 of offsets to include in diagnostics.
11641 2018-02-14 Martin Sebor <msebor@redhat.com>
11644 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
11645 that correspond to the kind of a declaration.
11647 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
11650 * config/pa/pa.md: Load address of PIC label using the linkage table
11651 if the label is nonlocal.
11653 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
11655 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
11656 warning message if user requests -maltivec=be.
11657 * doc/invoke.texi: Document deprecation of -maltivec=be.
11659 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
11662 * config/rs6000/rs6000-c.c: Update definitions for
11663 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
11664 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
11666 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11669 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
11670 add _get_ssp intrinsics. Remove argument from
11671 __builtin_ia32_rdssp[d|q].
11672 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
11673 * config/i386/i386-builtin.def: Remove argument from
11674 __builtin_ia32_rdssp[d|q].
11675 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
11676 ix86_expand_special_args_builtin for _rdssp[d|q].
11677 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
11678 Clear register before usage.
11679 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
11680 Add documentation for new _get_ssp and _inc_ssp intrinsics.
11682 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
11684 PR tree-optimization/84357
11685 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
11686 operand 1 of an ARRAY_REF too.
11688 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
11691 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
11692 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
11694 (set_of_reg): New struct.
11695 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
11696 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
11697 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
11699 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
11700 Split into bitclr, bitset, bitinvert patterns if appropriate.
11701 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
11702 use rx_fuse_in_memory_bitop.
11703 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
11704 to named insn, correct maximum insn length.
11706 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
11709 * machmode.def: Define a complex mode for PARTIAL_INT.
11710 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
11712 * doc/rtl.texi: Document CSPImode.
11713 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
11715 (msp430_hard_regno_nregs_with_padding): Likewise.
11717 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
11720 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
11722 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
11724 PR rtl-optimization/84169
11725 * combine.c (try_combine): New variable split_i2i3. Set it to true if
11726 we generated a parallel as new i3 and we split that to new i2 and i3
11727 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
11728 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
11729 those to i2, not i1. Partially rewrite this scan code.
11731 2018-02-13 Jakub Jelinek <jakub@redhat.com>
11734 * stor-layout.c (place_field): For variable length fields, adjust
11735 offset_align afterwards not just based on the field's alignment,
11736 but also on the size.
11738 PR middle-end/84309
11739 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
11740 of exps and logs in the use_exp2 case.
11742 2018-02-13 Jeff Law <law@redhat.com>
11744 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
11745 entry for "vector".
11747 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
11750 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
11754 * common.opt (gas-loc-support, gas-locview-support): New.
11755 (ginline-points, ginternal-reset-location-views): New.
11756 * doc/invoke.texi: Document them. Use @itemx where intended.
11757 (gvariable-location-views): Adjust.
11758 * target.def (reset_location_view): New.
11759 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
11760 (TARGET_RESET_LOCATION_VIEW): New.
11761 * doc/tm.texi: Rebuilt.
11762 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
11763 (dwarf2out_default_as_locview_support): New.
11764 (output_asm_line_debug_info): Use option variables.
11765 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
11766 (output_loc_list): Likewise.
11767 (add_high_low_attributes): Check option variables.
11768 Don't output entry view attribute in strict mode.
11769 (gen_inlined_subroutine_die): Check option variables.
11770 (dwarf2out_inline_entry): Likewise.
11771 (init_sections_and_labels): Likewise.
11772 (dwarf2out_early_finish): Likewise.
11773 (maybe_reset_location_view): New, from...
11774 (dwarf2out_var_location): ... here. Call it.
11775 * debug.h (dwarf2out_default_as_loc_support): Declare.
11776 (dwarf2out_default_as_locview_support): Declare.
11777 * hooks.c (hook_int_rtx_insn_0): New.
11778 * hooks.h (hook_int_rtx_insn_0): Declare.
11779 * toplev.c (process_options): Take -gas-loc-support and
11780 -gas-locview-support from dwarf2out. Enable
11781 -gvariable-location-views by default only with locview
11782 assembler support. Enable -ginternal-reset-location-views by
11783 default only if the target defines the corresponding hook.
11784 Enable -ginline-points by default if location views are
11785 enabled; force it disabled if statement frontiers are
11787 * tree-inline.c (expand_call_inline): Check option variables.
11788 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
11790 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
11792 PR tree-optimization/84321
11793 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
11794 handling. Also check whether the anti-range contains any values
11795 that satisfy the mask; switch to a VR_RANGE if not.
11797 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
11800 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
11802 2018-02-13 Martin Jambor <mjambor@suse.cz>
11805 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
11806 of call statements, also set location of a load to a temporary.
11808 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
11810 * config/rl78/rl78.c (add_vector_labels): New function.
11811 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
11812 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
11813 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
11814 which checks that no arguments are passed.
11815 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
11816 * doc/extend.texi: Documentation for the new attribute.
11818 2018-02-13 Andreas Schwab <schwab@suse.de>
11820 * config/riscv/linux.h (CPP_SPEC): Define.
11822 2018-02-13 Jakub Jelinek <jakub@redhat.com>
11825 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
11826 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
11827 OPTION_MASK_ISA_AES as first argument to def_builtin_const
11828 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
11829 instead of OPTION_MASK_ISA_PCLMUL as first argument to
11830 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
11831 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
11832 temporarily for AES and PCLMUL builtins.
11834 PR tree-optimization/84339
11835 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
11836 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
11839 PR middle-end/84309
11840 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
11841 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
11842 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
11844 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
11846 * omp-simd-clone.h: New file.
11847 * omp-simd-clone.c: Include omp-simd-clone.h.
11848 (expand_simd_clones): No longer static.
11849 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
11850 cgraph.h and omp-simd-clone.h.
11851 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
11852 (vect_recog_widen_shift_pattern): Formatting fix.
11853 (vect_pattern_recog_1): Don't check optab for calls.
11856 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
11857 operands[2] into a REG before using gen_lowpart on it.
11859 2018-02-12 Jeff Law <law@redhat.com>
11862 * config/sh/sh.c (find_barrier): Consider a sibling call
11865 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
11866 successfully back substituting a reg.
11868 2018-02-12 Richard Biener <rguenther@suse.de>
11870 PR tree-optimization/84037
11871 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
11872 parameter, move visited init to caller.
11873 (vect_slp_analyze_operations): Separate cost from validity
11874 check, initialize visited once for all instances.
11875 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
11877 * tree-vect-stmts.c (vect_model_simple_cost): Make early
11879 (vect_model_promotion_demotion_cost): Likewise.
11880 (vectorizable_bswap): Guard cost modeling with !slp_node
11881 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
11883 (vectorizable_call): Likewise.
11884 (vectorizable_conversion): Likewise.
11885 (vectorizable_assignment): Likewise.
11886 (vectorizable_shift): Likewise.
11887 (vectorizable_operation): Likewise.
11888 (vectorizable_store): Likewise.
11889 (vectorizable_load): Likewise.
11890 (vectorizable_condition): Likewise.
11891 (vectorizable_comparison): Likewise.
11893 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
11896 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
11897 (ASAN_MARK): Fix fnspec to account for return value, change pointer
11898 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
11900 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
11902 PR middle-end/83665
11903 * params.def (inline-min-speedup): Increase from 8 to 15.
11904 (max-inline-insns-auto): Decrease from 40 to 30.
11905 * ipa-split.c (consider_split): Add some buffer for function to
11906 be considered inlining candidate.
11907 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
11910 2018-02-12 Richard Biener <rguenther@suse.de>
11912 PR tree-optimization/84037
11913 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
11914 matched stmts if we cannot swap the non-matched ones.
11916 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
11918 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
11919 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
11920 _mm_maskz_scalef_round_ss): New intrinsics.
11921 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
11922 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
11923 __builtin_ia32_scalefss_round): Remove.
11924 (__builtin_ia32_scalefsd_mask_round,
11925 __builtin_ia32_scalefss_mask_round): New intrinsics.
11926 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
11927 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
11928 ((match_operand:VF_128 2 "<round_nimm_predicate>"
11929 "<round_constraint>")): Changed to ...
11930 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
11931 "<round_scalar_constraint>")): ... this.
11932 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
11933 %0, %1, %2<round_op3>}"): Changed to ...
11934 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
11935 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
11936 %2<round_scalar_mask_op3>}"): ... this.
11937 * config/i386/subst.md (round_scalar_nimm_predicate): New.
11939 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
11941 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
11942 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
11943 (_mm_maskz_sqrt_round_ss): New intrinsics.
11944 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
11945 (__builtin_ia32_sqrtsd_mask_round)
11946 (__builtin_ia32_sqrtss_mask_round): New builtins.
11947 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
11948 (__builtin_ia32_sqrtss_round): Remove.
11949 (__builtin_ia32_sqrtsd_mask_round)
11950 (__builtin_ia32_sqrtss_mask_round): New builtins.
11951 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
11952 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
11953 ((match_operand:VF_128 1 "vector_operand"
11954 "xBm,<round_constraint>")): Changed to ...
11955 ((match_operand:VF_128 1 "vector_operand"
11956 "xBm,<round_scalar_constraint>")): ... this.
11957 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
11958 %0, %2, %<iptr>1<round_op3>}): Changed to ...
11959 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
11960 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
11961 %<iptr>1<round_scalar_mask_op3>}): ... this.
11962 ((set_attr "prefix" "<round_prefix>")): Changed to ...
11963 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
11965 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
11968 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
11969 Cast vec_cmpeq result to correct type.
11970 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
11971 Cast vec_cmpgt result to correct type.
11973 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
11975 * final.c (final_scan_insn_1): Renamed from...
11976 (final_scan_insn): ... this. New wrapper, to recover
11977 seen from the outermost call in recursive ones.
11978 * config/sparc/sparc.c (output_return): Drop seen from call.
11979 (output_sibcall): Likewise.
11980 * config/visium/visium.c (output_branch): Likewise.
11982 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
11984 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
11987 2018-02-10 Alan Modra <amodra@gmail.com>
11990 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
11991 Specify LR as an input.
11993 2018-02-10 Jakub Jelinek <jakub@redhat.com>
11996 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
11997 remove_member_access_dummy_vars): New functions.
11998 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
11999 lower_omp_1, execute_lower_omp): Use them.
12001 PR rtl-optimization/84308
12002 * shrink-wrap.c (spread_components): Release todo vector.
12004 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
12006 PR rtl-optimization/57193
12007 * ira-color.c (struct allocno_color_data): Add member
12008 conflict_allocno_hard_prefs.
12009 (update_conflict_allocno_hard_prefs): New.
12010 (bucket_allocno_compare_func): Add a preference based on
12011 conflict_allocno_hard_prefs.
12012 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
12013 (color_allocnos): Remove a dead code. Initiate
12014 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
12016 2018-02-09 Jakub Jelinek <jakub@redhat.com>
12019 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
12020 constraint from =wa to wa. Avoid a subreg on the output operand,
12021 instead use a pseudo and subreg it in a move.
12022 (p9_xxbrd_<mode>): Changed to ...
12023 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
12024 (p9_xxbrd_v2df): New expander.
12025 (p9_xxbrw_<mode>): Changed to ...
12026 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
12027 (p9_xxbrw_v4sf): New expander.
12029 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
12031 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
12033 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
12036 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
12037 multiply in 32-bit mode.
12038 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
12039 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
12042 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
12044 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
12045 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
12046 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
12047 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
12049 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
12052 * dwarf2out.c (is_trivial_indirect_ref): New function.
12053 (dwarf2out_late_global_decl): Do not generate a location
12054 attribute for variables that have a non-trivial DECL_VALUE_EXPR
12055 and that are not defined in the current unit.
12057 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
12059 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
12060 instead of a libcall for UNORDERED.
12062 2018-02-09 Tamar Christina <tamar.christina@arm.com>
12065 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
12066 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
12068 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12071 * config/s390/s390.c (s390_set_current_function): Invoke
12072 s390_indirect_branch_settings also if fndecl didn't change.
12074 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
12076 * config/rs6000/rs6000.md (blockage): Set length to zero.
12078 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
12080 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
12082 2018-02-09 Jakub Jelinek <jakub@redhat.com>
12085 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
12086 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
12090 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
12091 PARALLEL incoming that failed vt_get_decl_and_offset check.
12093 PR middle-end/84237
12094 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
12095 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
12097 (get_variable_section): For decls in named .bss* sections pass true as
12098 second argument to bss_initializer_p.
12100 2018-02-09 Marek Polacek <polacek@redhat.com>
12101 Jakub Jelinek <jakub@redhat.com>
12104 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
12105 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
12106 Sync some changes from cxx_fold_indirect_ref.
12108 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
12110 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
12112 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
12113 (BLOCK_INLINE_ENTRY_LABEL): New.
12114 (dwarf2out_var_location): Disregard inline entry markers.
12115 (inline_entry_data): New struct.
12116 (inline_entry_data_hasher): New hashtable type.
12117 (inline_entry_data_hasher::hash): New.
12118 (inline_entry_data_hasher::equal): New.
12119 (inline_entry_data_table): New variable.
12120 (add_high_low_attributes): Add DW_AT_entry_pc and
12121 DW_AT_GNU_entry_view attributes if a pending entry is found
12122 in inline_entry_data_table. Add old entry_pc attribute only
12123 if debug nonbinding markers are disabled.
12124 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
12125 markers are enabled.
12126 (block_within_block_p, dwarf2out_inline_entry): New.
12127 (dwarf2out_finish): Check that no entries remained in
12128 inline_entry_data_table.
12129 * final.c (reemit_insn_block_notes): Handle inline entry notes.
12130 (final_scan_insn, notice_source_line): Likewise.
12131 (rest_of_clean_state): Skip inline entry markers.
12132 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
12134 * gimple.c (gimple_build_debug_inline_entry): New.
12135 * gimple.h (enum gimple_debug_subcode): Add
12136 GIMPLE_DEBUG_INLINE_ENTRY.
12137 (gimple_build_debug_inline_entry): Declare.
12138 (gimple_debug_inline_entry_p): New.
12139 (gimple_debug_nonbind_marker_p): Adjust.
12140 * insn-notes.def (INLINE_ENTRY): New.
12141 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
12142 inline entry marker notes.
12143 (print_insn): Likewise.
12144 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
12145 (INSN_DEBUG_MARKER_KIND): Likewise.
12146 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
12147 * tree-inline.c (expand_call_inline): Build and insert
12148 debug_inline_entry stmt.
12149 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
12150 inline entry blocks early, if nonbind markers are enabled.
12151 (dump_scope_block): Dump fragment info.
12152 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
12153 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
12154 (gimple_build_debug_inline_entry): New.
12155 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
12156 Enable/disable inline entry points too.
12157 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
12158 (DEBUG_INSN): Describe inline entry markers.
12160 * common.opt (gvariable-location-views): New.
12161 (gvariable-location-views=incompat5): New.
12162 * config.in: Rebuilt.
12163 * configure: Rebuilt.
12164 * configure.ac: Test assembler for view support.
12165 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
12166 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
12167 * dwarf2out.c (var_loc_view): New typedef.
12168 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
12169 (dwarf2out_locviews_in_attribute): New.
12170 (dwarf2out_locviews_in_loclist): New.
12171 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
12172 (enum dw_line_info_opcode): Add LI_adv_address.
12173 (struct dw_line_info_table): Add view.
12174 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
12175 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
12176 (zero_view_p): New variable.
12177 (ZERO_VIEW_P): New macro.
12178 (output_asm_line_debug_info): New.
12179 (struct var_loc_node): Add view.
12180 (add_AT_view_list, AT_loc_list): New.
12181 (add_var_loc_to_decl): Add view param. Test it against last.
12182 (new_loc_list): Add view params. Record them.
12183 (AT_loc_list_ptr): Handle loc and view lists.
12184 (view_list_to_loc_list_val_node): New.
12185 (print_dw_val): Handle dw_val_class_view_list.
12186 (size_of_die): Likewise.
12187 (value_format): Likewise.
12188 (loc_list_has_views): New.
12189 (gen_llsym): Set vl_symbol too.
12190 (maybe_gen_llsym, skip_loc_list_entry): New.
12191 (dwarf2out_maybe_output_loclist_view_pair): New.
12192 (output_loc_list): Output view list or entries too.
12193 (output_view_list_offset): New.
12194 (output_die): Handle dw_val_class_view_list.
12195 (output_dwarf_version): New.
12196 (output_compilation_unit_header): Use it.
12197 (output_skeleton_debug_sections): Likewise.
12198 (output_rnglists, output_line_info): Likewise.
12199 (output_pubnames, output_aranges): Update version comments.
12200 (output_one_line_info_table): Output view numbers in asm comments.
12201 (dw_loc_list): Determine current endview, pass it to new_loc_list.
12202 Call maybe_gen_llsym.
12203 (loc_list_from_tree_1): Adjust.
12204 (add_AT_location_description): Create view list attribute if
12205 needed, check it's absent otherwise.
12206 (convert_cfa_to_fb_loc_list): Adjust.
12207 (maybe_emit_file): Call output_asm_line_debug_info for test.
12208 (dwarf2out_var_location): Reset views as needed. Precompute
12209 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
12210 attribute. Set view.
12211 (new_line_info_table): Reset next view.
12212 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
12213 (dwarf2out_source_line): Likewise. Output view resets and labels to
12214 the assembler, or select appropriate line info opcodes.
12215 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
12216 (optimize_string_length): Catch it. Adjust.
12217 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
12218 dw_val_class_view_list, and remove it if no longer needed.
12219 (hash_loc_list): Hash view numbers.
12220 (loc_list_hasher::equal): Compare them.
12221 (optimize_location_lists): Check whether a view list symbol is
12222 needed, and whether the locview attribute is present, and
12223 whether they match. Remove the locview attribute if no longer
12225 (index_location_lists): Call skip_loc_list_entry for test.
12226 (dwarf2out_finish): Call output_asm_line_debug_info for test.
12227 Use output_dwarf_version.
12228 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
12229 (struct dw_val_node): Add val_view_list.
12230 * final.c (SEEN_NEXT_VIEW): New.
12231 (set_next_view_needed): New.
12232 (clear_next_view_needed): New.
12233 (maybe_output_next_view): New.
12234 (final_start_function): Rename to...
12235 (final_start_function_1): ... this. Take pointer to FIRST,
12236 add SEEN parameter. Emit param bindings in the initial view.
12237 (final_start_function): Reintroduce SEEN-less interface.
12238 (final): Rename to...
12239 (final_1): ... this. Take SEEN parameter. Output final pending
12240 next view at the end.
12241 (final): Reintroduce seen-less interface.
12242 (final_scan_insn): Output pending next view before switching
12243 sections or ending a block. Mark the next view as needed when
12244 outputting variable locations. Notify debug backend of section
12245 changes, and of location view changes.
12246 (rest_of_handle_final): Adjust.
12247 * toplev.c (process_options): Autodetect value for debug variable
12248 location views option. Warn on incompat5 without -gdwarf-5.
12249 * doc/invoke.texi (gvariable-location-views): New.
12250 (gvariable-location-views=incompat5): New.
12251 (gno-variable-location-views): New.
12253 2018-02-08 David Malcolm <dmalcolm@redhat.com>
12255 PR tree-optimization/84136
12256 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
12257 that the result of find_edge is non-NULL.
12259 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
12262 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
12263 storing integer register in SImode. Fix cost of 256 and 512
12264 byte aligned SSE register store.
12266 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
12268 * config/i386/i386.c (ix86_multiplication_cost): Fix
12269 multiplication cost for TARGET_AVX512DQ.
12271 2018-02-08 Marek Polacek <polacek@redhat.com>
12273 PR tree-optimization/84238
12274 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
12277 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
12279 PR tree-optimization/84265
12280 * tree-vect-stmts.c (vectorizable_store): Don't treat
12281 VMAT_CONTIGUOUS accesses as grouped.
12282 (vectorizable_load): Likewise.
12284 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
12286 PR tree-optimization/81635
12287 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
12288 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
12289 (test_round_for_mask): New functions.
12290 (wide_int_cc_tests): Call test_round_for_mask.
12291 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
12292 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
12293 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
12294 range returned by get_range_info.
12296 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
12299 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
12300 * symtab.c: Include builtins.h
12301 (symtab_node::output_to_lto_symbol_table_p): Move here
12302 from lto-streamer-out.c:output_symbol_p.
12303 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
12304 (output_symbol_p): Move all logic to symtab.c
12305 (produce_symtab): Update.
12307 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12309 * config/s390/s390-opts.h (enum indirect_branch): Define.
12310 * config/s390/s390-protos.h (s390_return_addr_from_memory)
12311 (s390_indirect_branch_via_thunk)
12312 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
12313 (enum s390_indirect_branch_type): Define.
12314 * config/s390/s390.c (struct s390_frame_layout, struct
12315 machine_function): Remove.
12316 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
12317 (indirect_branch_table_label_no, indirect_branch_table_name):
12319 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
12320 (enum s390_indirect_branch_option): Define.
12321 (s390_return_addr_from_memory): New function.
12322 (s390_handle_string_attribute): New function.
12323 (s390_attribute_table): Add new attribute handler.
12324 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
12325 (s390_indirect_branch_via_thunk): New function.
12326 (s390_indirect_branch_via_inline_thunk): New function.
12327 (s390_function_ok_for_sibcall): When jumping via thunk disallow
12328 sibling call optimization for non z10 compiles.
12329 (s390_emit_call): Force indirect branch target to be a single
12330 register. Add r1 clobber for non-z10 compiles.
12331 (s390_emit_epilogue): Emit return jump via return_use expander.
12332 (s390_reorg): Handle JUMP_INSNs as execute targets.
12333 (s390_option_override_internal): Perform validity checks for the
12334 new command line options.
12335 (s390_indirect_branch_attrvalue): New function.
12336 (s390_indirect_branch_settings): New function.
12337 (s390_set_current_function): Invoke s390_indirect_branch_settings.
12338 (s390_output_indirect_thunk_function): New function.
12339 (s390_code_end): Implement target hook.
12340 (s390_case_values_threshold): Implement target hook.
12341 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
12343 * config/s390/s390.h (struct s390_frame_layout)
12344 (struct machine_function): Move here from s390.c.
12345 (TARGET_INDIRECT_BRANCH_NOBP_RET)
12346 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
12347 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
12348 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
12349 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
12350 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
12351 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
12352 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
12353 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
12354 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
12355 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
12356 (mnemonic attribute): Add values which aren't recognized
12358 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
12359 pattern for branch conversion. Fix mnemonic attribute.
12360 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
12361 indirect branch via thunk if requested.
12362 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
12363 ("*indirect_jump"): Disable for branch conversion using out of
12365 ("indirect_jump_via_thunk<mode>_z10")
12366 ("indirect_jump_via_thunk<mode>")
12367 ("indirect_jump_via_inlinethunk<mode>_z10")
12368 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
12369 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
12370 ("casesi_jump_via_inlinethunk<mode>_z10")
12371 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
12372 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
12373 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
12374 ("*indirect2_jump"): Disable for branch conversion.
12375 ("casesi_jump"): Turn into expander and expand patterns for branch
12377 ("return_use"): New expander.
12378 ("*return"): Emit return via thunk and rename it to ...
12379 ("*return<mode>"): ... this one.
12380 * config/s390/s390.opt: Add new options and and enum for the
12383 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
12385 * lra-constraints.c (match_reload): Unconditionally use
12386 gen_lowpart_SUBREG, rather than selecting between that
12387 and equivalent gen_rtx_SUBREG code.
12389 2018-02-08 Richard Biener <rguenther@suse.de>
12391 PR tree-optimization/84233
12392 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
12393 changed flag instead of boguously re-using phi_inserted.
12395 2018-02-08 Martin Jambor <mjambor@suse.cz>
12397 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
12398 static local variables.
12400 2018-02-08 Richard Biener <rguenther@suse.de>
12402 PR tree-optimization/84278
12403 * tree-vect-stmts.c (vectorizable_store): When looking for
12404 smaller vector types to perform grouped strided loads/stores
12405 make sure the mode is supported by the target.
12406 (vectorizable_load): Likewise.
12408 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
12410 * config/aarch64/aarch64.c (aarch64_components_for_bb):
12411 Increase LDP/STP opportunities by adding adjacent callee-saves.
12413 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
12415 PR rtl-optimization/84068
12416 PR rtl-optimization/83459
12417 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
12419 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
12421 PR tree-optimization/84224
12422 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
12423 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
12424 non-zero arguments.
12426 2018-02-07 Iain Sandoe <iain@codesourcery.com>
12429 * config/rs6000/altivec.md (*restore_world): Remove LR use.
12430 * config/rs6000/predicates.md (restore_world_operation): Adjust op
12431 count, remove one USE.
12433 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
12435 * doc/install.texi (Configuration): Document the
12436 --with-long-double-format={ibm,ieee} PowerPC configuration
12440 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
12441 Convert from define_expand to be define_insn_and_split. Rework
12442 float/double/_Float128 conversions to QI/HI/SImode to work with
12443 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
12444 conversions to QI/HImode types did a store and then a load to
12445 truncate the value. For conversions to VSX registers, don't split
12446 the insn, instead emit the code directly. Use the code iterator
12447 any_fix to combine signed and unsigned conversions.
12448 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
12449 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
12450 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
12451 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
12452 (fix_<mode>di2_hw): Likewise.
12453 (fixuns_<mode>di2_hw): Likewise.
12454 (fix_<mode>si2_hw): Likewise.
12455 (fixuns_<mode>si2_hw): Likewise.
12456 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
12457 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
12458 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
12459 fix<uns>_trunc<SFDF:mode>si2_p8.
12460 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
12461 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
12462 (fix<uns>_<mode>_mem): Likewise.
12463 (fctiw<u>z_<mode>_mem): Likewise.
12464 (fix<uns>_<mode>_mem): Likewise.
12465 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
12466 the register allocator from doing a direct move to the GPRs to do
12467 a store, and instead use the ISA 3.0 store byte/half-word from
12468 vector register instruction. For IEEE 128-bit floating point,
12469 also optimize stores of 32-bit ints.
12470 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
12472 2018-02-07 Alan Hayward <alan.hayward@arm.com>
12474 * genextract.c (push_pathstr_operand): New function to support
12476 (walk_rtx): Call push_pathstr_operand.
12477 (print_path): Support [a-zA-Z].
12479 2018-02-07 Richard Biener <rguenther@suse.de>
12481 PR tree-optimization/84037
12482 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
12483 (cse_and_gimplify_to_preheader): Declare.
12484 (vect_get_place_in_interleaving_chain): Likewise.
12485 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
12487 (_loop_vec_info::~_loop_vec_info): Delete it.
12488 (cse_and_gimplify_to_preheader): New function.
12489 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
12490 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
12491 (vectorizable_load): Likewise. For grouped stores always base
12492 the IV on the first element.
12493 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
12494 condition before gimplifying.
12496 2018-02-07 Jakub Jelinek <jakub@redhat.com>
12498 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
12499 *DIV_EXPR and *MOD_EXPR.
12501 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
12504 * config/i386/i386.c (ix86_option_override_internal): Mask out
12505 the CF_SET bit when checking -fcf-protection.
12507 2018-02-07 Tom de Vries <tom@codesourcery.com>
12510 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
12513 2018-02-07 Richard Biener <rguenther@suse.de>
12515 PR tree-optimization/84204
12516 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
12519 PR tree-optimization/84205
12520 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
12521 special-case isl_ast_op_zdiv_r.
12523 PR tree-optimization/84223
12524 * graphite-scop-detection.c (gather_bbs::before_dom_children):
12525 Only add conditions from within the region.
12526 (gather_bbs::after_dom_children): Adjust.
12528 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
12531 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
12532 * config/avr/avr.md: Only post-reload split REG-REG moves if
12533 either register is GENERAL_REG_P.
12535 2018-02-07 Jakub Jelinek <jakub@redhat.com>
12537 PR tree-optimization/84235
12538 * tree-ssa-scopedtables.c
12539 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
12540 if the subtraction is performed in floating point type where NaNs are
12541 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
12542 build 1. Formatting fix.
12544 2018-02-06 Jakub Jelinek <jakub@redhat.com>
12547 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
12548 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
12549 and skip it regardless of bb boundaries. Use CALL_P macro,
12550 don't test INSN_P (insn) together with CALL_P or JUMP_P check
12551 unnecessarily, formatting fix.
12553 2018-02-06 Michael Collison <michael.collison@arm.com>
12555 * config/arm/thumb2.md:
12556 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
12557 (*thumb_mov_notscc): Ditto.
12559 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
12562 * config/rs6000/rs6000.md (su code attribute): Use "u" for
12563 unsigned_fix, not "s".
12565 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12567 * configure.ac (gcc_fn_eh_frame_ro): New function.
12568 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
12569 correct .eh_frame permissions.
12570 * configure: Regenerate.
12572 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
12574 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
12575 irrelevant options.
12577 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12579 * config/rs6000/rs6000.c (rs6000_option_override_internal):
12580 Display warning message for -mno-speculate-indirect-jumps.
12582 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
12584 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
12586 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
12588 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
12590 PR tree-optimization/84225
12591 * tree-eh.c (find_trapping_overflow): Only call
12592 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
12594 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
12597 * config/i386/i386.c: Reimplement the check of possible options
12598 -mibt/-mshstk conbination. Change error messages.
12599 * doc/invoke.texi: Fix a typo: remove extra '='.
12601 2018-02-06 Marek Polacek <polacek@redhat.com>
12603 PR tree-optimization/84228
12604 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
12606 2018-02-06 Tamar Christina <tamar.christina@arm.com>
12609 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
12610 emitted arch directives.
12611 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
12612 __ARM_FEATURE_COPROC before changing architectures.
12614 2018-02-06 Richard Biener <rguenther@suse.de>
12616 * config/i386/i386.c (print_reg): Fix typo.
12617 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
12619 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
12621 * configure: Regenerate.
12623 2018-02-05 Martin Sebor <msebor@redhat.com>
12625 PR tree-optimization/83369
12626 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
12629 2018-02-05 Martin Liska <mliska@suse.cz>
12631 * doc/invoke.texi: Cherry-pick upstream r323995.
12633 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
12635 * ira.c (ira_init_register_move_cost): Adjust comment.
12637 2018-02-05 Martin Liska <mliska@suse.cz>
12639 PR gcov-profile/84137
12640 * doc/gcov.texi: Fix typo in documentation.
12642 2018-02-05 Martin Liska <mliska@suse.cz>
12644 PR gcov-profile/83879
12645 * doc/gcov.texi: Document necessity of --dynamic-list-data when
12646 using dlopen functionality.
12648 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
12650 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
12651 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
12652 _mm_maskz_range_ss, _mm_mask_range_round_ss,
12653 _mm_maskz_range_round_ss): New intrinsics.
12654 (__builtin_ia32_rangesd128_round)
12655 (__builtin_ia32_rangess128_round): Remove.
12656 (__builtin_ia32_rangesd128_mask_round,
12657 __builtin_ia32_rangess128_mask_round): New builtins.
12658 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
12659 __builtin_ia32_rangess128_round): Remove.
12660 (__builtin_ia32_rangesd128_mask_round,
12661 __builtin_ia32_rangess128_mask_round): New builtins.
12662 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
12663 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
12664 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
12665 "<round_saeonly_constraint>")): Changed to ...
12666 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
12667 "<round_saeonly_scalar_constraint>")): ... this.
12668 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
12669 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
12670 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
12671 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
12672 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
12674 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
12676 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
12678 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
12679 Remove all values except native, 8540 and 8548.
12681 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
12683 * config/i386/i386.c (ix86_output_function_return): Pass
12684 INVALID_REGNUM, instead of -1, as invalid register number to
12685 indirect_thunk_name and output_indirect_thunk.
12687 2018-02-02 Julia Koval <julia.koval@intel.com>
12689 * config.gcc: Add -march=icelake.
12690 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
12691 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
12692 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
12693 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
12694 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
12695 (processor_target_table): Add icelake.
12696 (ix86_option_override_internal): Handle new PTAs.
12697 (get_builtin_code_for_version): Handle icelake.
12698 (M_INTEL_COREI7_ICELAKE): New.
12699 (fold_builtin_cpu): Handle icelake.
12700 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
12701 * doc/invoke.texi: Add -march=icelake.
12703 2018-02-02 Julia Koval <julia.koval@intel.com>
12705 * config/i386/i386.c (ix86_option_override_internal): Change flags type
12706 to wide_int_bitmask.
12707 * wide-int-bitmask.h: New.
12709 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
12712 * config/i386/i386.md: Replace Pmode with word_mode in
12713 builtin_setjmp_setup and builtin_longjmp to support x32.
12715 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
12719 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
12721 (rs6000_supported_cpu_names): New static variable.
12722 (linux_cpu_translation_table): Likewise.
12723 (elf_platform) <cpu>: Define new static variable and use it.
12724 Translate kernel AT_PLATFORM name to canonical name if needed.
12725 Error if platform name is unknown.
12727 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
12730 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
12732 2018-02-01 Jeff Law <law@redhat.com>
12735 * config/i386/i386.c (release_scratch_register_on_entry): Add new
12736 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
12737 the scratch if RELEASE_VIA_POP is false.
12738 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
12739 If we have to save a temporary register, decrement SIZE appropriately.
12740 Pass new arguments to release_scratch_register_on_entry.
12741 (ix86_adjust_stack_and_probe): Likewise.
12742 (ix86_emit_probe_stack_range): Pass new arguments to
12743 release_scratch_register_on_entry.
12745 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
12747 PR rtl-optimization/84157
12748 * combine.c (change_zero_ext): Use REG_P predicate in
12749 front of HARD_REGISTER_P predicate.
12751 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
12753 * config/avr/avr.c (avr_option_override): Move disabling of
12754 -fdelete-null-pointer-checks to...
12755 * common/config/avr/avr-common.c (avr_option_optimization_table):
12758 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
12760 PR tree-optimization/81635
12761 * tree-data-ref.c (split_constant_offset_1): For types that
12762 wrap on overflow, try to use range info to prove that wrapping
12765 2018-02-01 Renlin Li <renlin.li@arm.com>
12768 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
12769 TAILCALL_ADDR_REGS.
12770 (aarch64_register_move_cost): Likewise.
12771 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
12772 TAILCALL_ADDR_REGS.
12773 (REG_CLASS_NAMES): Likewise.
12774 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
12775 TAILCALL_ADDR_REGS. Remove IP registers.
12776 * config/aarch64/aarch64.md (Ucs): Update register constraint.
12778 2018-02-01 Richard Biener <rguenther@suse.de>
12780 * domwalk.h (dom_walker::dom_walker): Add additional constructor
12781 for specifying RPO order and allow NULL for that.
12782 * domwalk.c (dom_walker::dom_walker): Likewise.
12783 (dom_walker::walk): Handle NULL RPO order.
12784 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
12786 (rewrite_update_dom_walker): Likewise.
12787 (mark_def_dom_walker): Likewise.
12789 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
12791 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
12792 (aarch64_maybe_expand_sve_subreg_move): Declare.
12793 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
12794 * config/aarch64/predicates.md (aarch64_any_register_operand): New
12796 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
12797 that are semantically a reverse operation.
12798 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
12799 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
12800 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
12802 (aarch64_can_change_mode_class): For big-endian, forbid changes
12803 between two SVE modes if they have different element sizes.
12805 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
12807 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
12808 the TImode handling for big-endian targets.
12810 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
12812 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
12813 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
12815 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
12816 Remove BSWAP handing for big-endian targets and use the form of
12817 LD1RQ appropariate for the mode.
12819 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
12821 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
12822 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
12823 duplicated element.
12825 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
12828 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
12829 check for operands that need to go through aarch64_sve_reload_be.
12831 2018-02-01 Jakub Jelinek <jakub@redhat.com>
12833 PR tree-optimization/81661
12834 PR tree-optimization/84117
12835 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
12836 * tree-eh.c: Include gimplify.h.
12837 (find_trapping_overflow, replace_trapping_overflow,
12838 rewrite_to_non_trapping_overflow): New functions.
12839 * tree-vect-loop.c: Include tree-eh.h.
12840 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
12841 * tree-data-ref.c: Include tree-eh.h.
12842 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
12844 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
12846 PR rtl-optimization/84123
12847 * combine.c (change_zero_ext): Check if hard register satisfies
12848 can_change_dest_mode before calling gen_lowpart_SUBREG.
12850 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
12853 * ira.c (ira_init_register_move_cost): Remove assert.
12855 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
12857 PR rtl-optimization/84071
12858 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
12859 * doc/tm.texi: Regenerate.
12861 2018-01-31 Richard Biener <rguenther@suse.de>
12863 PR tree-optimization/84132
12864 * tree-data-ref.c (analyze_miv_subscript): Properly
12865 check whether evolution_function_is_affine_multivariate_p
12866 before calling gcd_of_steps_may_divide_p.
12868 2018-01-31 Julia Koval <julia.koval@intel.com>
12871 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
12872 * config/i386/i386.md (rdpid_rex64) New.
12873 (rdpid): Make 32bit only.
12875 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
12878 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
12879 an IDENTIFIER_NODE for FUNCTION_TYPE's.
12881 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
12884 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
12886 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
12888 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
12890 PR rtl-optimization/84071
12891 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
12892 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
12894 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
12896 * config/arc/arc.c (arc_handle_aux_attribute): New function.
12897 (arc_attribute_table): Add 'aux' attribute.
12898 (arc_in_small_data_p): Consider aux like variables.
12899 (arc_is_aux_reg_p): New function.
12900 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
12901 (arc_get_aux_arg): New function.
12902 (prepare_move_operands): Handle aux-register access.
12903 (arc_handle_aux_attribute): New function.
12904 * doc/extend.texi (ARC Variable attributes): Add subsection.
12906 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
12908 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
12909 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
12910 (arc_attribute_table): Add 'uncached' attribute.
12911 (arc_print_operand): Print '.di' flag for uncached memory
12913 (arc_in_small_data_p): Do not consider for small data the uncached
12915 (arc_is_uncached_mem_p): New function.
12916 * config/arc/predicates.md (compact_store_memory_operand): Check
12917 for uncached memory accesses.
12918 (nonvol_nonimm_operand): Likewise.
12919 * gcc/doc/extend.texi (ARC Type Attribute): New subsection.
12921 2018-01-31 Jakub Jelinek <jakub@redhat.com>
12924 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
12925 falign-loops=): Add Optimization flag.
12927 2018-01-30 Jeff Law <law@redhat.com>
12930 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
12931 INT_REGISTERS_SAVED. Check it prior to calling
12932 get_scratch_register_on_entry.
12933 (ix86_adjust_stack_and_probe): Similarly.
12934 (ix86_emit_probe_stack_range): Similarly.
12935 (ix86_expand_prologue): Corresponding changes.
12937 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12940 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
12941 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
12943 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
12946 * lra-constraints.c (curr_insn_transform): Process AND in the
12949 2018-01-30 Jakub Jelinek <jakub@redhat.com>
12951 PR rtl-optimization/83986
12952 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
12953 dependence against last_pending_memory_flush in addition to
12954 pending_jump_insns.
12956 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
12958 PR tree-optimization/81611
12959 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
12962 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12965 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
12968 2018-01-30 Richard Biener <rguenther@suse.de>
12969 Jakub Jelinek <jakub@redhat.com>
12971 PR tree-optimization/84111
12972 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
12973 inner loops added during recursion, as they don't have up-to-date
12976 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
12979 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
12980 (can_inline_edge_by_limits_p): ... here.
12981 (can_early_inline_edge_p, check_callers,
12982 update_caller_keys, update_callee_keys, recursive_inlining,
12983 add_new_edges_to_heap, speculation_useful_p,
12984 inline_small_functions,
12985 inline_small_functions, flatten_function,
12986 inline_to_all_callers_1): Update.
12988 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
12990 * profile-count.c (profile_count::combine_with_ipa_count): Handle
12993 2018-01-30 Richard Biener <rguenther@suse.de>
12995 PR tree-optimization/83008
12996 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
12997 invariant and constant vector uses in stmts when they need
12998 more than one stmt.
13000 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13003 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
13004 * configure: Regenerate.
13006 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
13008 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
13010 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
13011 Use gen_rtx_REG rather than gen_lowpart.
13013 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
13015 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
13016 rather than 0 when creating partial subregs.
13018 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
13020 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
13023 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
13026 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
13027 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
13028 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
13029 flags. This restores the settings used before the 2017-07-24.
13030 Turning off pre increment/decrement/modify allows IVOPTS to
13031 optimize DF/SF loops where the index is an int.
13033 2018-01-29 Richard Biener <rguenther@suse.de>
13034 Kelvin Nilsen <kelvin@gcc.gnu.org>
13037 * tree-vect-stmts.c (vectorizable_call): Don't call
13038 targetm.vectorize_builtin_md_vectorized_function if callee is
13041 2018-01-22 Carl Love <cel@us.ibm.com>
13043 * doc/extend.tex: Fix typo in second arg in
13044 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
13046 2018-01-29 Richard Biener <rguenther@suse.de>
13048 PR tree-optimization/84086
13049 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
13050 (flush_ssaname_freelist): When SSA names were released reset
13051 the SCEV hash table.
13053 2018-01-29 Richard Biener <rguenther@suse.de>
13055 PR tree-optimization/84057
13056 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
13057 removed paths when removing edges.
13059 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
13061 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
13062 -mfunction-return=@var{choice}.
13064 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
13066 PR diagnostic/84034
13067 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
13068 Handle CR like TAB.
13069 (layout::print_source_line): Likewise.
13070 (test_get_line_width_without_trailing_whitespace): Add test cases.
13072 2018-01-27 Jakub Jelinek <jakub@redhat.com>
13074 PR middle-end/84040
13075 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
13078 2018-01-26 Jim Wilson <jimw@sifive.com>
13080 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
13082 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
13085 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13087 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
13088 and CMP + SUB-immediate -> SUBS.
13090 2018-01-26 Martin Sebor <msebor@redhat.com>
13092 PR tree-optimization/83896
13093 * tree-ssa-strlen.c (get_string_len): Rename...
13094 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
13095 Avoid assuming length is constant.
13096 (handle_char_store): Use HOST_WIDE_INT for string length.
13098 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
13101 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
13102 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
13104 2018-01-26 Richard Biener <rguenther@suse.de>
13106 PR rtl-optimization/84003
13107 * dse.c (record_store): Only record redundant stores when
13108 the earlier store aliases at least all accesses the later one does.
13110 2018-01-26 Jakub Jelinek <jakub@redhat.com>
13112 PR rtl-optimization/83985
13113 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
13114 REG_CFA_RESTORE insns.
13115 (delete_unmarked_insns): Don't ignore separate shrink wrapping
13116 REG_CFA_RESTORE insns here.
13119 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
13120 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
13122 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
13124 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
13125 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
13126 (arc_init): Likewise.
13127 (arc_override_options): Likewise.
13128 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
13130 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
13132 * config/arc/arc.h (TARGET_DBNZ): Define.
13133 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
13134 properly set the tune attribute.
13135 (dbnz): Use TARGET_DBNZ guard.
13136 * config/arc/arc.opt (mtune): Add core3 option.
13138 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
13140 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
13141 recognize new pic like addresses.
13142 (arc_delegitimize_address): Clean up.
13144 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
13146 * config/arc/arc-arches.def: Option mrf16 valid for all
13148 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
13149 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
13150 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
13151 * config/arc/arc-tables.opt: Regenerate.
13152 * config/arc/arc.c (arc_conditional_register_usage): Handle
13153 reduced register file case.
13154 (arc_file_start): Set must have build attributes.
13155 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
13156 mrf16 option value.
13157 * config/arc/arc.opt (mrf16): Add new option.
13158 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
13159 * config/arc/genmultilib.awk: Handle new mrf16 option.
13160 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
13161 * config/arc/t-multilib: Regenerate.
13162 * doc/invoke.texi (ARC Options): Document mrf16 option.
13164 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
13166 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
13167 * config/arc/arc.c (arc_handle_secure_attribute): New function.
13168 (arc_attribute_table): Add 'secure_call' attribute.
13169 (arc_print_operand): Print secure call operand.
13170 (arc_function_ok_for_sibcall): Don't optimize tail calls when
13172 (arc_is_secure_call_p): New function. * config/arc/arc.md
13173 (call_i): Add support for sjli instruction.
13174 (call_value_i): Likewise.
13175 * config/arc/constraints.md (Csc): New constraint.
13177 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
13178 John Eric Martin <John.Martin@emmicro-us.com>
13180 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
13181 * config/arc/arc.c (_arc_jli_section): New struct.
13182 (arc_jli_section): New type.
13183 (rc_jli_sections): New static variable.
13184 (arc_handle_jli_attribute): New function.
13185 (arc_attribute_table): Add jli_always and jli_fixed attribute.
13186 (arc_file_end): New function.
13187 (TARGET_ASM_FILE_END): Define.
13188 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
13189 (arc_add_jli_section): New function.
13190 (jli_call_scan): Likewise.
13191 (arc_reorg): Call jli_call_scan.
13192 (arc_output_addsi): Remove 'S' from printing asm operand.
13193 (arc_is_jli_call_p): New function.
13194 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
13196 (movhi_insn): Likewise.
13197 (movsi_insn): Likewise.
13198 (movsi_set_cc_insn): Likewise.
13199 (loadqi_update): Likewise.
13200 (load_zeroextendqisi_update): Likewise.
13201 (load_signextendqisi_update): Likewise.
13202 (loadhi_update): Likewise.
13203 (load_zeroextendhisi_update): Likewise.
13204 (load_signextendhisi_update): Likewise.
13205 (loadsi_update): Likewise.
13206 (loadsf_update): Likewise.
13207 (movsicc_insn): Likewise.
13208 (bset_insn): Likewise.
13209 (bxor_insn): Likewise.
13210 (bclr_insn): Likewise.
13211 (bmsk_insn): Likewise.
13212 (bicsi3_insn): Likewise.
13213 (cmpsi_cc_c_insn): Likewise.
13214 (movsi_ne): Likewise.
13215 (movsi_cond_exec): Likewise.
13216 (clrsbsi2): Likewise.
13217 (norm_f): Likewise.
13227 (call_i): Remove 'S' asm letter, add jli instruction.
13228 (call_value_i): Likewise.
13229 * config/arc/arc.op (mjli-always): New option.
13230 * config/arc/constraints.md (Cji): New constraint.
13231 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
13233 (subsf3_fpx): Likewise.
13234 (mulsf3_fpx): Likewise.
13235 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
13237 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
13238 function attrbutes.
13239 * doc/invoke.texi (ARC): Document mjli-always option.
13241 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
13243 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
13244 avoid addition with 0 and use incw and decw where possible.
13246 2018-01-26 Richard Biener <rguenther@suse.de>
13248 PR tree-optimization/81082
13249 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
13250 association if it requires casting to unsigned.
13251 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
13252 from fold_plusminus_mult_expr to catch important cases late when
13253 range info is available.
13255 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13257 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
13258 * configure.ac (hidden_linkonce): New test.
13259 * configure: Regenerate.
13260 * config.in: Regenerate.
13262 2018-01-26 Julia Koval <julia.koval@intel.com>
13264 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
13265 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
13266 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
13267 _mm_mask_bitshuffle_epi64_mask): Fix type.
13268 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
13269 USI_FTYPE_V4DI_V4DI_USI): Remove.
13270 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
13271 __builtin_ia32_vpshufbitqmb256_mask,
13272 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
13273 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
13274 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
13276 2018-01-26 Alan Modra <amodra@gmail.com>
13279 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
13280 UNSPEC_VBPERMQ. Sort other unspecs.
13282 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
13284 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
13286 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
13288 PR middle-end/83055
13289 * predict.c (drop_profile): Do not push/pop cfun; update also
13291 (handle_missing_profiles): Fix logic looking for zero profiles.
13293 2018-01-25 Jakub Jelinek <jakub@redhat.com>
13295 PR middle-end/83977
13296 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
13297 on functions with #pragma omp declare simd or functions with simd
13299 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
13300 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
13301 Remove trailing \n from warning_at calls.
13303 2018-01-25 Tom de Vries <tom@codesourcery.com>
13306 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
13307 for neutered workers.
13309 2018-01-24 Joseph Myers <joseph@codesourcery.com>
13312 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
13313 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
13315 2018-01-24 Jeff Law <law@redhat.com>
13318 * i386.c (get_probe_interval): Move to earlier point.
13319 (ix86_compute_frame_layout): If -fstack-clash-protection and
13320 the frame is larger than the probe interval, then use pushes
13321 to save registers rather than reg->mem moves.
13322 (ix86_expand_prologue): Remove conditional for int_registers_saved
13325 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
13328 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
13329 min/max for never referenced object.
13331 2018-01-24 Jakub Jelinek <jakub@redhat.com>
13333 PR middle-end/83977
13334 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
13336 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
13337 attributes from DECL_ATTRIBUTES (decl) without affecting
13338 DECL_ATTRIBUTES (current_function_decl).
13339 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
13340 functions with non-NULL DECL_ABSTRACT_ORIGIN.
13342 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
13344 PR tree-optimization/83979
13345 * fold-const.c (fold_comparison): Use constant_boolean_node
13346 instead of boolean_{true,false}_node.
13348 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
13350 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
13353 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13355 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
13356 Simplify the clause that sets the length attribute.
13357 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
13358 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
13359 clause that sets the length attribute.
13360 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
13362 2018-01-24 Tom de Vries <tom@codesourcery.com>
13365 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
13366 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
13367 Add strict parameter.
13368 (prevent_branch_around_nothing): Insert dummy insn between branch to
13369 label and label with no ptx insn inbetween.
13370 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
13372 2018-01-24 Tom de Vries <tom@codesourcery.com>
13375 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
13376 for neutered threads in warp.
13377 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
13379 2018-01-24 Richard Biener <rguenther@suse.de>
13381 PR tree-optimization/83176
13382 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
13385 2018-01-24 Richard Biener <rguenther@suse.de>
13387 PR tree-optimization/82819
13388 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
13389 code generating pluses that are no-ops in the target precision.
13391 2018-01-24 Richard Biener <rguenther@suse.de>
13393 PR middle-end/84000
13394 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
13396 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
13398 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
13399 to merge probabilities.
13400 * predict.c (probably_never_executed): Also mark as cold functions
13401 with global 0 profile and guessed local profile.
13402 * profile-count.c (profile_probability::combine_with_count): New
13404 * profile-count.h (profile_probability::operator*,
13405 profile_probability::operator*=, profile_probability::operator/,
13406 profile_probability::operator/=): Reduce precision to adjusted
13407 and set value to guessed on contradictory divisions.
13408 (profile_probability::combine_with_freq): Remove.
13409 (profile_probability::combine_wiht_count): Declare.
13410 (profile_count::force_nonzero):: Set to adjusted.
13411 (profile_count::probability_in):: Set quality to adjusted.
13412 * tree-ssa-tail-merge.c (replace_block_by): Use
13413 combine_with_count.
13415 2018-01-23 Andrew Waterman <andrew@sifive.com>
13416 Jim Wilson <jimw@sifive.com>
13418 * config/riscv/riscv.c (riscv_stack_boundary): New.
13419 (riscv_option_override): Set riscv_stack_boundary. Handle
13420 riscv_preferred_stack_boundary_arg.
13421 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
13422 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
13423 (STACK_BOUNDARY): Set to riscv_stack_boundary.
13424 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
13425 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
13426 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
13428 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
13431 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
13432 of struct ix86_frame.
13433 (ix86_expand_epilogue): Likewise. Add a local variable for
13434 the reg_save_offset field in struct ix86_frame.
13436 2018-01-23 Bin Cheng <bin.cheng@arm.com>
13438 PR tree-optimization/82604
13439 * tree-loop-distribution.c (enum partition_kind): New enum item
13440 PKIND_PARTIAL_MEMSET.
13441 (partition_builtin_p): Support above new enum item.
13442 (generate_code_for_partition): Ditto.
13443 (compute_access_range): Differentiate cases that equality can be
13444 proven at all loops, the innermost loops or no loops.
13445 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
13446 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
13447 (finalize_partitions, distribute_loop): Don't fuse partition of
13448 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
13449 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
13450 parloop is enabled.
13452 2018-01-23 Martin Liska <mliska@suse.cz>
13454 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
13455 order to ignore the predictor.
13456 (PRED_POLYMORPHIC_CALL): Likewise.
13457 (PRED_RECURSIVE_CALL): Likewise.
13459 2018-01-23 Martin Liska <mliska@suse.cz>
13461 * tree-profile.c (tree_profiling): Print function header to
13462 aware reader which function we are working on.
13463 * value-prof.c (gimple_find_values_to_profile): Do not print
13464 not interesting value histograms.
13466 2018-01-23 Martin Liska <mliska@suse.cz>
13468 * profile-count.h (enum profile_quality): Add
13469 profile_uninitialized as the first value. Do not number values
13470 as they are zero based.
13471 (profile_count::verify): Update sanity check.
13472 (profile_probability::verify): Likewise.
13474 2018-01-23 Nathan Sidwell <nathan@acm.org>
13476 * doc/invoke.texi (ffor-scope): Deprecate.
13478 2018-01-23 David Malcolm <dmalcolm@redhat.com>
13480 PR tree-optimization/83510
13481 * domwalk.c (set_all_edges_as_executable): New function.
13482 (dom_walker::dom_walker): Convert bool param
13483 "skip_unreachable_blocks" to enum reachability. Move setup of
13484 edge flags to set_all_edges_as_executable and only do it when
13485 reachability is REACHABLE_BLOCKS.
13486 * domwalk.h (enum dom_walker::reachability): New enum.
13487 (dom_walker::dom_walker): Convert bool param
13488 "skip_unreachable_blocks" to enum reachability.
13489 (set_all_edges_as_executable): New decl.
13490 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
13491 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
13493 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
13494 but converting true to REACHABLE_BLOCKS.
13495 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
13497 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
13498 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
13499 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
13501 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
13502 if check_all_array_refs will be called.
13504 2018-01-23 David Malcolm <dmalcolm@redhat.com>
13506 * tree.c (selftest::test_location_wrappers): Add more test
13509 2018-01-23 David Malcolm <dmalcolm@redhat.com>
13511 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
13512 (selftest::test_bit_in_range): Likewise.
13514 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
13517 * doc/sourcebuild.texi (vect_float): Say that the selector
13518 only describes the situation when -funsafe-math-optimizations is on.
13519 (vect_float_strict): Document.
13521 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
13523 PR tree-optimization/83965
13524 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
13525 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
13526 instead of checking only for a reduction.
13527 (vect_recog_widen_sum_pattern): Likewise.
13529 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
13531 * predict.c (probably_never_executed): Only use precise profile info.
13532 (compute_function_frequency): Skip after inlining hack since we now
13533 have quality checking.
13535 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
13537 * profile-count.h (profile_probability::very_unlikely,
13538 profile_probability::unlikely, profile_probability::even): Set
13539 precision to guessed.
13541 2018-01-23 Richard Biener <rguenther@suse.de>
13543 PR tree-optimization/83963
13544 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
13545 Properly terminate dominator walk when crossing the exit edge not
13546 when visiting its source block.
13548 2018-01-23 Jakub Jelinek <jakub@redhat.com>
13551 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
13552 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
13554 2018-01-22 Jakub Jelinek <jakub@redhat.com>
13556 PR tree-optimization/83957
13557 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
13558 semicolon after for body surrounded by braces.
13560 PR tree-optimization/83081
13561 * profile-count.h (profile_probability::split): New method.
13562 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
13563 Use profile_probability::split.
13564 (do_compare_rtx_and_jump): Fix adjustment of probabilities
13565 when splitting a single conditional jump into 2.
13567 2018-01-22 David Malcolm <dmalcolm@redhat.com>
13569 PR tree-optimization/69452
13570 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
13573 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
13575 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
13576 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
13577 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
13579 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
13581 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
13583 * config/rl78/rl78.md (movdi): New define_expand.
13584 * config/rl78/rl78.c (rl78_split_movdi): New function.
13586 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
13589 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
13591 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
13592 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
13593 128-bit to produce an UNSPEC move to get the double word with the
13594 signbit and then a shift directly to do signbit.
13595 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
13596 implementation with a new version that just does either a direct
13597 move or a regular move. Move memory interface to separate insns.
13598 Move insns so they are next to the expander.
13599 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
13600 with signbit move. Split big and little endian case.
13601 (signbit<mode>2_dm_mem_le): Likewise.
13602 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
13603 (signbit<mode>2_dm2): Likewise.
13605 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
13607 * config/rl78/rl78.md (anddi3): New define_expand.
13609 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
13611 * config/rl78/rl78.md (umindi3): New define_expand.
13613 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
13615 * config/rl78/rl78.md (smindi3): New define_expand.
13617 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
13619 * config/rl78/rl78.md (smaxdi3): New define_expand.
13621 2018-01-22 Carl Love <cel@us.ibm.com>
13623 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
13624 LVX_V1TI): Add macro expansion.
13625 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
13626 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
13627 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
13628 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
13629 Change check to determine if the instruction is a byte reversing
13630 entry. Fix typo in comment.
13631 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
13632 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
13633 Add def_builtin calls for new builtins.
13634 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
13635 Add define_insn expansion.
13637 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
13639 * config/rl78/rl78.md (umaxdi3): New define_expand.
13641 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
13643 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
13644 for non-QImode registers.
13646 2018-01-22 Richard Biener <rguenther@suse.de>
13648 PR tree-optimization/83963
13649 * graphite-scop-detection.c (scop_detection::get_sese): Delay
13650 including the loop exit block.
13651 (scop_detection::merge_sese): Likewise.
13652 (scop_detection::add_scop): Do it here instead.
13654 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13656 * doc/sourcebuild.texi (arm_softfloat): Document.
13658 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
13661 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
13662 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
13663 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
13665 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13666 David Edelsohn <dje.gcc@gmail.com>
13669 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
13670 Change "crset eq" to "crset 2".
13671 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
13672 (*call_indirect_aix<mode>_nospec): Likewise.
13673 (*call_value_indirect_aix<mode>_nospec): Likewise.
13674 (*call_indirect_elfv2<mode>_nospec): Likewise.
13675 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
13676 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
13677 change assembly output from . to $.
13678 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
13679 (indirect_jump<mode>_nospec): Change assembly output from . to $.
13680 (*tablejump<mode>_internal1_nospec): Likewise.
13682 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
13685 * config/sh/sh_optimize_sett_clrt.cc:
13686 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
13688 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
13690 PR tree-optimization/83940
13691 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
13692 offset_dt to vect_constant_def rather than vect_unknown_def_type.
13693 (vect_check_load_store_mask): Add a mask_dt_out parameter and
13694 use it to pass back the definition type.
13695 (vect_check_store_rhs): Likewise rhs_dt_out.
13696 (vect_build_gather_load_calls): Add a mask_dt argument and use
13697 it instead of a call to vect_is_simple_use.
13698 (vectorizable_store): Update calls to vect_check_load_store_mask
13699 and vect_check_store_rhs. Use the dt returned by the latter instead
13700 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
13701 instead of calls to vect_is_simple_use. Pass the scalar rather
13702 than the vector operand to vect_is_simple_use when handling
13703 second and subsequent copies of an rhs value.
13704 (vectorizable_load): Update calls to vect_check_load_store_mask
13705 and vect_build_gather_load_calls. Use the cached mask_dt and
13706 gs_info.offset_dt instead of calls to vect_is_simple_use.
13708 2018-01-20 Jakub Jelinek <jakub@redhat.com>
13710 PR middle-end/83945
13711 * tree-emutls.c: Include gimplify.h.
13712 (lower_emutls_2): New function.
13713 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
13714 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
13715 it before further processing.
13718 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
13719 UINTVAL (trueop1) instead of INTVAL (op1).
13721 2018-01-19 Jakub Jelinek <jakub@redhat.com>
13725 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
13726 INCOMING_FRAME_SP_OFFSET if not defined.
13727 (scan_trace): Add ENTRY argument. If true and
13728 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
13729 emit a note to adjust the CFA offset.
13730 (create_cfi_notes): Adjust scan_trace callers.
13731 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
13732 INCOMING_FRAME_SP_OFFSET in the CIE.
13733 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
13734 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
13736 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
13737 * doc/tm.texi: Regenerated.
13739 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13741 PR rtl-optimization/83147
13742 * lra-constraints.c (remove_inheritance_pseudos): Use
13743 lra_substitute_pseudo_within_insn.
13745 2018-01-19 Tom de Vries <tom@codesourcery.com>
13746 Cesar Philippidis <cesar@codesourcery.com>
13749 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
13751 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
13754 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
13755 spaces for function labels.
13757 2018-01-19 Martin Liska <mliska@suse.cz>
13759 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
13760 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
13761 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
13762 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
13763 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
13764 (PRED_CONST_RETURN): Change from 69 to 65.
13765 (PRED_NULL_RETURN): Change from 91 to 71.
13766 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
13767 (PRED_LOOP_GUARD): Change from 66 to 73.
13769 2018-01-19 Martin Liska <mliska@suse.cz>
13771 * predict.c (predict_insn_def): Add new assert.
13772 (struct branch_predictor): Change type to signed integer.
13773 (test_prediction_value_range): Amend test to cover
13774 PROB_UNINITIALIZED.
13775 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
13776 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
13777 (PRED_LOOP_ITERATIONS_MAX): Likewise.
13778 (PRED_LOOP_IV_COMPARE): Likewise.
13779 * predict.h (PROB_UNINITIALIZED): Define new constant.
13781 2018-01-19 Martin Liska <mliska@suse.cz>
13783 * predict.c (dump_prediction): Add new format for
13784 analyze_brprob.py script which is enabled with -details
13786 * profile-count.h (precise_p): New function.
13788 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
13790 PR tree-optimization/83922
13791 * tree-vect-loop.c (vect_verify_full_masking): Return false if
13792 there are no statements that need masking.
13793 (vect_active_double_reduction_p): New function.
13794 (vect_analyze_loop_operations): Use it when handling phis that
13795 are not in the loop header.
13797 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
13799 PR tree-optimization/83914
13800 * tree-vect-loop.c (vectorizable_induction): Don't convert
13801 init_expr or apply the peeling adjustment for inductions
13802 that are nested within the vectorized loop.
13804 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13806 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
13809 2018-01-18 Jakub Jelinek <jakub@redhat.com>
13813 * function.h (gimplify_parameters): Add gimple_seq * argument.
13814 * function.c: Include gimple.h and options.h.
13815 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
13816 for the added local temporaries if needed.
13817 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
13818 if there are any parameter cleanups, wrap whole body into a
13819 try/finally with the cleanups.
13821 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
13824 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
13825 Use GET_MODE_CLASS for scalar floating point.
13827 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
13831 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
13832 Fix call of call_cgraph_insertion_hooks.
13834 2018-01-18 Martin Sebor <msebor@redhat.com>
13836 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
13838 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
13841 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
13844 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
13847 * common.opt: (-ffile-prefix-map): New option.
13848 * opts.c (common_handle_option): Defer it.
13849 * opts-global.c (handle_common_deferred_options): Handle it.
13850 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
13851 * file-prefix-map.h: New file.
13852 (remap_debug_filename, add_debug_prefix_map): ...here.
13853 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
13854 * final.c (debug_prefix_map, add_debug_prefix_map
13855 remap_debug_filename): Move to...
13856 * file-prefix-map.c: New file.
13857 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
13858 generalize, get rid of alloca(), use strrchr() instead of strchr().
13859 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
13860 Implement in terms of add_prefix_map().
13861 (remap_macro_filename, remap_debug_filename): Implement in term of
13863 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
13864 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
13865 * dbxout.c: Include file-prefix-map.h.
13866 * varasm.c: Likewise.
13867 * vmsdbgout.c: Likewise.
13868 * xcoffout.c: Likewise.
13869 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
13870 * doc/cppopts.texi (-fmacro-prefix-map): Document.
13871 * doc/invoke.texi (-ffile-prefix-map): Document.
13872 (-fdebug-prefix-map): Update description.
13874 2018-01-18 Martin Liska <mliska@suse.cz>
13876 * config/i386/i386.c (indirect_thunk_name): Document that also
13878 (output_indirect_thunk): Document why both instructions
13879 (pause and lfence) are generated.
13881 2018-01-18 Richard Biener <rguenther@suse.de>
13883 PR tree-optimization/83887
13884 * graphite-scop-detection.c
13885 (scop_detection::get_nearest_dom_with_single_entry): Remove.
13886 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
13887 (scop_detection::merge_sese): Re-implement with a flood-fill
13888 algorithm that properly finds a SESE region if it exists.
13890 2018-01-18 Jakub Jelinek <jakub@redhat.com>
13893 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
13894 pointer_diff optimizations use view_convert instead of convert.
13896 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13898 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
13899 Generate different code for -mno-speculate-indirect-jumps.
13900 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
13901 (*call_indirect_aix<mode>): Disable for
13902 -mno-speculate-indirect-jumps.
13903 (*call_indirect_aix<mode>_nospec): New define_insn.
13904 (*call_value_indirect_aix<mode>): Disable for
13905 -mno-speculate-indirect-jumps.
13906 (*call_value_indirect_aix<mode>_nospec): New define_insn.
13907 (*sibcall_nonlocal_sysv<mode>): Generate different code for
13908 -mno-speculate-indirect-jumps.
13909 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
13911 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
13913 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
13914 long double type, set the flags for noting the default long double
13915 type, even if we don't pass or return a long double type.
13917 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
13920 * ipa-inline.c (flatten_function): Do not overwrite final inlining
13923 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
13925 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
13926 support for merge[hl].
13927 (fold_mergehl_helper): New helper function.
13928 (tree-vector-builder.h): New #include for tree_vector_builder usage.
13929 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
13930 (altivec_vmrglw_direct): Add xxmrglw insn.
13932 2018-01-17 Andrew Waterman <andrew@sifive.com>
13934 * config/riscv/riscv.c (riscv_conditional_register_usage): If
13935 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
13937 2018-01-17 David Malcolm <dmalcolm@redhat.com>
13940 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
13941 call the lto_location_cache before reading the
13942 DECL_SOURCE_LOCATION of the types.
13944 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
13945 Richard Sandiford <richard.sandiford@linaro.org>
13947 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
13948 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
13949 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
13950 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
13951 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
13953 * config/aarch64/constraints.md (aarch64_movti_operand):
13955 * config/aarch64/predicates.md (Uti): Add new constraint.
13957 2018-01-17 Carl Love <cel@us.ibm.com>
13959 * config/rs6000/vsx.md (define_expand xl_len_r,
13960 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
13961 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
13963 (define_expand, define_insn): Move the shift left from the
13964 define_insn to the define_expand for lxvl and stxvl instructions.
13965 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
13966 and XL_LEN_R definitions to PURE.
13968 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
13970 * config/i386/i386.c (indirect_thunk_name): Declare regno
13971 as unsigned int. Compare regno with INVALID_REGNUM.
13972 (output_indirect_thunk): Ditto.
13973 (output_indirect_thunk_function): Ditto.
13974 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
13975 in the call to output_indirect_thunk_function.
13977 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
13979 PR middle-end/83884
13980 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
13981 rather than the size of inner_type to determine the stack slot size
13982 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
13984 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
13987 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
13990 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
13992 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
13993 endian Linux systems to optionally enable multilibs for selecting
13994 the long double type if the user configured an explicit type.
13995 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
13996 have no long double multilibs if not defined.
13997 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
13998 warn if the user used -mabi={ieee,ibm}longdouble and we built
13999 multilibs for long double.
14000 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
14001 appropriate multilib option.
14002 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
14004 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
14005 for building long double multilibs.
14006 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
14008 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
14010 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
14013 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
14015 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
14018 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
14021 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
14024 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
14026 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
14027 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
14029 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
14031 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
14032 different rtl trees depending on TARGET_64BIT.
14033 (rs6000_gen_lvx): Likewise.
14035 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
14037 * config/visium/visium.md (nop): Tweak comment.
14038 (hazard_nop): Likewise.
14040 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14042 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
14043 -mspeculate-indirect-jumps.
14044 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
14045 for -mno-speculate-indirect-jumps.
14046 (*call_indirect_elfv2<mode>_nospec): New define_insn.
14047 (*call_value_indirect_elfv2<mode>): Disable for
14048 -mno-speculate-indirect-jumps.
14049 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
14050 (indirect_jump): Emit different RTL for
14051 -mno-speculate-indirect-jumps.
14052 (*indirect_jump<mode>): Disable for
14053 -mno-speculate-indirect-jumps.
14054 (*indirect_jump<mode>_nospec): New define_insn.
14055 (tablejump): Emit different RTL for
14056 -mno-speculate-indirect-jumps.
14057 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
14058 (tablejumpsi_nospec): New define_expand.
14059 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
14060 (tablejumpdi_nospec): New define_expand.
14061 (*tablejump<mode>_internal1): Disable for
14062 -mno-speculate-indirect-jumps.
14063 (*tablejump<mode>_internal1_nospec): New define_insn.
14064 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
14067 2018-01-16 Artyom Skrobov tyomitch@gmail.com
14069 * caller-save.c (insert_save): Drop unnecessary parameter. All
14072 2018-01-16 Jakub Jelinek <jakub@redhat.com>
14073 Richard Biener <rguenth@suse.de>
14076 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
14077 return early, inline manually is_gimple_sizepos. Make sure if we
14078 call gimplify_expr we don't end up with a gimple constant.
14079 * tree.c (variably_modified_type_p): Don't return true for
14080 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
14081 * gimplify.h (is_gimple_sizepos): Remove.
14083 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
14085 PR tree-optimization/83857
14086 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
14087 vectorizable_live_operation for pure SLP statements.
14088 (vectorizable_live_operation): Handle PHIs.
14090 2018-01-16 Richard Biener <rguenther@suse.de>
14092 PR tree-optimization/83867
14093 * tree-vect-stmts.c (vect_transform_stmt): Precompute
14094 nested_in_vect_loop_p since the scalar stmt may get invalidated.
14096 2018-01-16 Jakub Jelinek <jakub@redhat.com>
14099 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
14100 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
14101 If off is not INTEGER_CST, issue a may not be aligned warning
14102 rather than isn't aligned. Use isn%'t rather than isn't.
14103 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
14105 <case MULT_EXPR>: Improve the case when bottom and one of the
14106 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
14107 operand, in that case check if the other operand is multiple of
14108 bottom divided by the INTEGER_CST operand.
14110 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
14113 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
14114 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
14115 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
14116 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
14117 * config/pa/pa.c (pa_function_arg_advance): Likewise.
14118 (pa_function_arg, pa_arg_partial_bytes): Likewise.
14119 (pa_function_arg_size): New function.
14121 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
14123 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
14124 in a separate statement.
14126 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
14128 PR tree-optimization/83847
14129 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
14130 group gathers and scatters.
14132 2018-01-16 Jakub Jelinek <jakub@redhat.com>
14134 PR rtl-optimization/86620
14135 * params.def (max-sched-ready-insns): Bump minimum value to 1.
14137 PR rtl-optimization/83213
14138 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
14139 to last if both are JUMP_INSNs.
14141 PR tree-optimization/83843
14142 * gimple-ssa-store-merging.c
14143 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
14144 store_immediate_info for bswap/nop orig_stores.
14146 2018-01-15 Andrew Waterman <andrew@sifive.com>
14148 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
14150 <UDIV>: Increase cost if !TARGET_DIV.
14152 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
14154 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
14155 (define_attr "cr_logical_3op"): New.
14156 (cceq_ior_compare): Adjust.
14157 (cceq_ior_compare_complement): Adjust.
14158 (*cceq_rev_compare): Adjust.
14159 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
14160 (is_cracked_insn): Adjust.
14161 (insn_must_be_first_in_group): Adjust.
14162 * config/rs6000/40x.md: Adjust.
14163 * config/rs6000/440.md: Adjust.
14164 * config/rs6000/476.md: Adjust.
14165 * config/rs6000/601.md: Adjust.
14166 * config/rs6000/603.md: Adjust.
14167 * config/rs6000/6xx.md: Adjust.
14168 * config/rs6000/7450.md: Adjust.
14169 * config/rs6000/7xx.md: Adjust.
14170 * config/rs6000/8540.md: Adjust.
14171 * config/rs6000/cell.md: Adjust.
14172 * config/rs6000/e300c2c3.md: Adjust.
14173 * config/rs6000/e500mc.md: Adjust.
14174 * config/rs6000/e500mc64.md: Adjust.
14175 * config/rs6000/e5500.md: Adjust.
14176 * config/rs6000/e6500.md: Adjust.
14177 * config/rs6000/mpc.md: Adjust.
14178 * config/rs6000/power4.md: Adjust.
14179 * config/rs6000/power5.md: Adjust.
14180 * config/rs6000/power6.md: Adjust.
14181 * config/rs6000/power7.md: Adjust.
14182 * config/rs6000/power8.md: Adjust.
14183 * config/rs6000/power9.md: Adjust.
14184 * config/rs6000/rs64.md: Adjust.
14185 * config/rs6000/titan.md: Adjust.
14187 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
14189 * config/i386/predicates.md (indirect_branch_operand): Rewrite
14190 ix86_indirect_branch_register logic.
14192 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
14194 * config/i386/constraints.md (Bs): Update
14195 ix86_indirect_branch_register check. Don't check
14196 ix86_indirect_branch_register with GOT_memory_operand.
14198 * config/i386/predicates.md (GOT_memory_operand): Don't check
14199 ix86_indirect_branch_register here.
14200 (GOT32_symbol_operand): Likewise.
14202 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
14204 * config/i386/predicates.md (constant_call_address_operand):
14205 Rewrite ix86_indirect_branch_register logic.
14206 (sibcall_insn_operand): Likewise.
14208 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
14210 * config/i386/constraints.md (Bs): Replace
14211 ix86_indirect_branch_thunk_register with
14212 ix86_indirect_branch_register.
14214 * config/i386/i386.md (indirect_jump): Likewise.
14215 (tablejump): Likewise.
14216 (*sibcall_memory): Likewise.
14217 (*sibcall_value_memory): Likewise.
14218 Peepholes of indirect call and jump via memory: Likewise.
14219 * config/i386/i386.opt: Likewise.
14220 * config/i386/predicates.md (indirect_branch_operand): Likewise.
14221 (GOT_memory_operand): Likewise.
14222 (call_insn_operand): Likewise.
14223 (sibcall_insn_operand): Likewise.
14224 (GOT32_symbol_operand): Likewise.
14226 2018-01-15 Jakub Jelinek <jakub@redhat.com>
14228 PR middle-end/83837
14229 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
14230 type rather than type addr's type points to.
14231 (expand_omp_atomic_mutex): Likewise.
14232 (expand_omp_atomic): Likewise.
14234 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
14237 * config/i386/i386.c (output_indirect_thunk_function): Use
14238 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
14239 for __x86_return_thunk.
14241 2018-01-15 Richard Biener <rguenther@suse.de>
14243 PR middle-end/83850
14244 * expmed.c (extract_bit_field_1): Fix typo.
14246 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14249 * config/arm/iterators.md (VF): New mode iterator.
14250 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
14251 Remove integer-related logic from pattern.
14252 (neon_vabd<mode>_3): Likewise.
14254 2018-01-15 Jakub Jelinek <jakub@redhat.com>
14256 PR middle-end/82694
14257 * common.opt (fstrict-overflow): No longer an alias.
14258 (fwrapv-pointer): New option.
14259 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
14260 also for pointer types based on flag_wrapv_pointer.
14261 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
14262 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
14263 opts->x_flag_wrapv got set.
14264 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
14265 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
14266 POINTER_TYPE_OVERFLOW_UNDEFINED.
14267 * match.pd: Likewise in address comparison pattern.
14268 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
14270 2018-01-15 Richard Biener <rguenther@suse.de>
14273 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
14274 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
14275 Reset type names to their identifier if their TYPE_DECL doesn't
14276 have linkage (and thus is used for ODR and devirt).
14277 (save_debug_info_for_decl): Remove.
14278 (save_debug_info_for_type): Likewise.
14279 (add_tree_to_fld_list): Adjust.
14280 * tree-pretty-print.c (dump_generic_node): Make dumping of
14281 type names more robust.
14283 2018-01-15 Richard Biener <rguenther@suse.de>
14285 * BASE-VER: Bump to 8.0.1.
14287 2018-01-14 Martin Sebor <msebor@redhat.com>
14290 * builtins.c (check_access): Avoid warning when the no-warning bit
14293 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
14295 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
14296 * ira-color (allocno_hard_regs_compare): Likewise.
14298 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
14301 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
14302 Use .pushsection/.popsection.
14304 2018-01-14 Martin Sebor <msebor@redhat.com>
14307 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
14309 2018-01-14 Jakub Jelinek <jakub@redhat.com>
14311 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
14312 entry from extra_headers.
14313 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
14314 extra_headers, make the list bitwise identical to the i?86-*-* one.
14316 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
14318 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
14319 -mcmodel=large with -mindirect-branch=thunk,
14320 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
14321 -mfunction-return=thunk-extern.
14322 * doc/invoke.texi: Document -mcmodel=large is incompatible with
14323 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
14324 -mfunction-return=thunk and -mfunction-return=thunk-extern.
14326 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
14328 * config/i386/i386.c (print_reg): Print the name of the full
14329 integer register without '%'.
14330 (ix86_print_operand): Handle 'V'.
14331 * doc/extend.texi: Document 'V' modifier.
14333 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
14335 * config/i386/constraints.md (Bs): Disallow memory operand for
14336 -mindirect-branch-register.
14338 * config/i386/predicates.md (indirect_branch_operand): Likewise.
14339 (GOT_memory_operand): Likewise.
14340 (call_insn_operand): Likewise.
14341 (sibcall_insn_operand): Likewise.
14342 (GOT32_symbol_operand): Likewise.
14343 * config/i386/i386.md (indirect_jump): Call convert_memory_address
14344 for -mindirect-branch-register.
14345 (tablejump): Likewise.
14346 (*sibcall_memory): Likewise.
14347 (*sibcall_value_memory): Likewise.
14348 Disallow peepholes of indirect call and jump via memory for
14349 -mindirect-branch-register.
14350 (*call_pop): Replace m with Bw.
14351 (*call_value_pop): Likewise.
14352 (*sibcall_pop_memory): Replace m with Bs.
14353 * config/i386/i386.opt (mindirect-branch-register): New option.
14354 * doc/invoke.texi: Document -mindirect-branch-register option.
14356 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
14358 * config/i386/i386-protos.h (ix86_output_function_return): New.
14359 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
14360 set function_return_type.
14361 (indirect_thunk_name): Add ret_p to indicate thunk for function
14363 (output_indirect_thunk_function): Pass false to
14364 indirect_thunk_name.
14365 (ix86_output_indirect_branch_via_reg): Likewise.
14366 (ix86_output_indirect_branch_via_push): Likewise.
14367 (output_indirect_thunk_function): Create alias for function
14368 return thunk if regno < 0.
14369 (ix86_output_function_return): New function.
14370 (ix86_handle_fndecl_attribute): Handle function_return.
14371 (ix86_attribute_table): Add function_return.
14372 * config/i386/i386.h (machine_function): Add
14373 function_return_type.
14374 * config/i386/i386.md (simple_return_internal): Use
14375 ix86_output_function_return.
14376 (simple_return_internal_long): Likewise.
14377 * config/i386/i386.opt (mfunction-return=): New option.
14378 (indirect_branch): Mention -mfunction-return=.
14379 * doc/extend.texi: Document function_return function attribute.
14380 * doc/invoke.texi: Document -mfunction-return= option.
14382 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
14384 * config/i386/i386-opts.h (indirect_branch): New.
14385 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
14386 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
14387 with local indirect jump when converting indirect call and jump.
14388 (ix86_set_indirect_branch_type): New.
14389 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
14390 (indirectlabelno): New.
14391 (indirect_thunk_needed): Likewise.
14392 (indirect_thunk_bnd_needed): Likewise.
14393 (indirect_thunks_used): Likewise.
14394 (indirect_thunks_bnd_used): Likewise.
14395 (INDIRECT_LABEL): Likewise.
14396 (indirect_thunk_name): Likewise.
14397 (output_indirect_thunk): Likewise.
14398 (output_indirect_thunk_function): Likewise.
14399 (ix86_output_indirect_branch_via_reg): Likewise.
14400 (ix86_output_indirect_branch_via_push): Likewise.
14401 (ix86_output_indirect_branch): Likewise.
14402 (ix86_output_indirect_jmp): Likewise.
14403 (ix86_code_end): Call output_indirect_thunk_function if needed.
14404 (ix86_output_call_insn): Call ix86_output_indirect_branch if
14406 (ix86_handle_fndecl_attribute): Handle indirect_branch.
14407 (ix86_attribute_table): Add indirect_branch.
14408 * config/i386/i386.h (machine_function): Add indirect_branch_type
14409 and has_local_indirect_jump.
14410 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
14412 (tablejump): Likewise.
14413 (*indirect_jump): Use ix86_output_indirect_jmp.
14414 (*tablejump_1): Likewise.
14415 (simple_return_indirect_internal): Likewise.
14416 * config/i386/i386.opt (mindirect-branch=): New option.
14417 (indirect_branch): New.
14420 (thunk-inline): Likewise.
14421 (thunk-extern): Likewise.
14422 * doc/extend.texi: Document indirect_branch function attribute.
14423 * doc/invoke.texi: Document -mindirect-branch= option.
14425 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
14428 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
14430 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
14432 * ipa-inline.c (want_inline_small_function_p): Return false if
14433 inlining has already failed with CIF_FINAL_ERROR.
14434 (update_caller_keys): Call want_inline_small_function_p before
14436 (update_callee_keys): Likewise.
14438 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
14440 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
14442 (rs6000_quadword_masked_address_p): Likewise.
14443 (quad_aligned_load_p): Likewise.
14444 (quad_aligned_store_p): Likewise.
14445 (const_load_sequence_p): Add comment to describe the outer-most loop.
14446 (mimic_memory_attributes_and_flags): New function.
14447 (rs6000_gen_stvx): Likewise.
14448 (replace_swapped_aligned_store): Likewise.
14449 (rs6000_gen_lvx): Likewise.
14450 (replace_swapped_aligned_load): Likewise.
14451 (replace_swapped_load_constant): Capitalize argument name in
14452 comment describing this function.
14453 (rs6000_analyze_swaps): Add a third pass to search for vector loads
14454 and stores that access quad-word aligned addresses and replace
14455 with stvx or lvx instructions when appropriate.
14456 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
14457 New function prototype.
14458 (rs6000_quadword_masked_address_p): Likewise.
14459 (rs6000_gen_lvx): Likewise.
14460 (rs6000_gen_stvx): Likewise.
14461 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
14462 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
14463 when memory address is aligned.
14464 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
14465 this split to select lvx instruction when memory address is aligned.
14466 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
14467 instruction when memory address is aligned.
14468 (*vsx_le_perm_load_v16qi): Likewise.
14469 (four unnamed splitters): Modify to select the stvx instruction
14470 when memory is aligned.
14472 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
14474 * predict.c (determine_unlikely_bbs): Handle correctly BBs
14475 which appears in the queue multiple times.
14477 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
14478 Alan Hayward <alan.hayward@arm.com>
14479 David Sherwood <david.sherwood@arm.com>
14481 * tree-vectorizer.h (vec_lower_bound): New structure.
14482 (_loop_vec_info): Add check_nonzero and lower_bounds.
14483 (LOOP_VINFO_CHECK_NONZERO): New macro.
14484 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
14485 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
14486 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
14487 fields. Make seg_len the distance travelled, not including the
14489 (dr_direction_indicator): Declare.
14490 (dr_zero_step_indicator): Likewise.
14491 (dr_known_forward_stride_p): Likewise.
14492 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
14494 (runtime_alias_check_p): Allow runtime alias checks with
14496 (operator ==): Compare access_size and align.
14497 (prune_runtime_alias_test_list): Rework for new distinction between
14498 the access_size and seg_len.
14499 (create_intersect_range_checks_index): Likewise. Cope with polynomial
14501 (get_segment_min_max): New function.
14502 (create_intersect_range_checks): Use it.
14503 (dr_step_indicator): New function.
14504 (dr_direction_indicator): Likewise.
14505 (dr_zero_step_indicator): Likewise.
14506 (dr_known_forward_stride_p): Likewise.
14507 * tree-loop-distribution.c (data_ref_segment_size): Return
14508 DR_STEP * (niters - 1).
14509 (compute_alias_check_pairs): Update call to the dr_with_seg_len
14511 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
14512 (vect_preserves_scalar_order_p): New function, split out from...
14513 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
14514 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
14515 (vect_vfa_access_size): New function.
14516 (vect_vfa_align): Likewise.
14517 (vect_compile_time_alias): Take access_size_a and access_b arguments.
14518 (dump_lower_bound): New function.
14519 (vect_check_lower_bound): Likewise.
14520 (vect_small_gap_p): Likewise.
14521 (vectorizable_with_step_bound_p): Likewise.
14522 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
14523 depencies if the vectorization factor is 1. Convert the checks
14524 for nonzero steps into checks on the bounds of DR_STEP. Try using
14525 a bunds check for variable steps if the minimum required step is
14526 relatively small. Update calls to the dr_with_seg_len
14527 constructor and to vect_compile_time_alias.
14528 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
14530 (vect_loop_versioning): Call it.
14531 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
14533 (vect_estimate_min_profitable_iters): Account for any bounds checks.
14535 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
14536 Alan Hayward <alan.hayward@arm.com>
14537 David Sherwood <david.sherwood@arm.com>
14539 * doc/sourcebuild.texi (vect_scatter_store): Document.
14540 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
14542 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
14544 * genopinit.c (main): Add supports_vec_scatter_store and
14545 supports_vec_scatter_store_cached to target_optabs.
14546 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
14547 IFN_MASK_SCATTER_STORE.
14548 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
14550 * internal-fn.h (internal_store_fn_p): Declare.
14551 (internal_fn_stored_value_index): Likewise.
14552 * internal-fn.c (scatter_store_direct): New macro.
14553 (expand_scatter_store_optab_fn): New function.
14554 (direct_scatter_store_optab_supported_p): New macro.
14555 (internal_store_fn_p): New function.
14556 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
14557 IFN_MASK_SCATTER_STORE.
14558 (internal_fn_mask_index): Likewise.
14559 (internal_fn_stored_value_index): New function.
14560 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
14561 for scatter stores.
14562 * optabs-query.h (supports_vec_scatter_store_p): Declare.
14563 * optabs-query.c (supports_vec_scatter_store_p): New function.
14564 * tree-vectorizer.h (vect_get_store_rhs): Declare.
14565 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
14566 true for scatter stores.
14567 (vect_gather_scatter_fn_p): Handle scatter stores too.
14568 (vect_check_gather_scatter): Consider using scatter stores if
14569 supports_vec_scatter_store_p.
14570 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
14571 scatter stores too.
14572 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
14573 internal_fn_stored_value_index.
14574 (check_load_store_masking): Handle scatter stores too.
14575 (vect_get_store_rhs): Make public.
14576 (vectorizable_call): Use internal_store_fn_p.
14577 (vectorizable_store): Handle scatter store internal functions.
14578 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
14579 when deciding whether the end of the group has been reached.
14580 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
14581 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
14582 (mask_scatter_store<mode>): New insns.
14584 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
14585 Alan Hayward <alan.hayward@arm.com>
14586 David Sherwood <david.sherwood@arm.com>
14588 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
14589 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
14590 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
14592 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
14593 Use vect_truncate_gather_scatter_offset if we can't treat the
14594 operation as a normal gather load or scatter store.
14595 (get_group_load_store_type): Take the gather_scatter_info
14596 as argument. Try using a gather load or scatter store for
14597 single-element groups.
14598 (get_load_store_type): Update calls to get_group_load_store_type
14599 and vect_use_strided_gather_scatters_p.
14601 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
14602 Alan Hayward <alan.hayward@arm.com>
14603 David Sherwood <david.sherwood@arm.com>
14605 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
14606 optional tree argument.
14607 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
14609 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
14610 but continue to use the current value as a fallback.
14611 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
14612 to compare the updates.
14613 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
14614 (get_load_store_type): Use it when handling a strided access.
14615 (vect_get_strided_load_store_ops): New function.
14616 (vect_get_data_ptr_increment): Likewise.
14617 (vectorizable_load): Handle strided gather loads. Always pass
14618 a step to vect_create_data_ref_ptr and bump_vector_ptr.
14620 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
14621 Alan Hayward <alan.hayward@arm.com>
14622 David Sherwood <david.sherwood@arm.com>
14624 * doc/md.texi (gather_load@var{m}): Document.
14625 (mask_gather_load@var{m}): Likewise.
14626 * genopinit.c (main): Add supports_vec_gather_load and
14627 supports_vec_gather_load_cached to target_optabs.
14628 * optabs-tree.c (init_tree_optimization_optabs): Use
14629 ggc_cleared_alloc to allocate target_optabs.
14630 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
14631 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
14633 * internal-fn.h (internal_load_fn_p): Declare.
14634 (internal_gather_scatter_fn_p): Likewise.
14635 (internal_fn_mask_index): Likewise.
14636 (internal_gather_scatter_fn_supported_p): Likewise.
14637 * internal-fn.c (gather_load_direct): New macro.
14638 (expand_gather_load_optab_fn): New function.
14639 (direct_gather_load_optab_supported_p): New macro.
14640 (direct_internal_fn_optab): New function.
14641 (internal_load_fn_p): Likewise.
14642 (internal_gather_scatter_fn_p): Likewise.
14643 (internal_fn_mask_index): Likewise.
14644 (internal_gather_scatter_fn_supported_p): Likewise.
14645 * optabs-query.c (supports_at_least_one_mode_p): New function.
14646 (supports_vec_gather_load_p): Likewise.
14647 * optabs-query.h (supports_vec_gather_load_p): Declare.
14648 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
14649 and memory_type field.
14650 (NUM_PATTERNS): Bump to 15.
14651 * tree-vect-data-refs.c: Include internal-fn.h.
14652 (vect_gather_scatter_fn_p): New function.
14653 (vect_describe_gather_scatter_call): Likewise.
14654 (vect_check_gather_scatter): Try using internal functions for
14655 gather loads. Recognize existing calls to a gather load function.
14656 (vect_analyze_data_refs): Consider using gather loads if
14657 supports_vec_gather_load_p.
14658 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
14659 (vect_get_gather_scatter_offset_type): Likewise.
14660 (vect_convert_mask_for_vectype): Likewise.
14661 (vect_add_conversion_to_patterm): Likewise.
14662 (vect_try_gather_scatter_pattern): Likewise.
14663 (vect_recog_gather_scatter_pattern): New pattern recognizer.
14664 (vect_vect_recog_func_ptrs): Add it.
14665 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
14666 internal_fn_mask_index and internal_gather_scatter_fn_p.
14667 (check_load_store_masking): Take the gather_scatter_info as an
14668 argument and handle gather loads.
14669 (vect_get_gather_scatter_ops): New function.
14670 (vectorizable_call): Check internal_load_fn_p.
14671 (vectorizable_load): Likewise. Handle gather load internal
14673 (vectorizable_store): Update call to check_load_store_masking.
14674 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
14675 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
14676 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
14677 (aarch64_gather_scale_operand_d): New predicates.
14678 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
14679 (mask_gather_load<mode>): New insns.
14681 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
14682 Alan Hayward <alan.hayward@arm.com>
14683 David Sherwood <david.sherwood@arm.com>
14685 * optabs.def (fold_left_plus_optab): New optab.
14686 * doc/md.texi (fold_left_plus_@var{m}): Document.
14687 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
14688 * internal-fn.c (fold_left_direct): Define.
14689 (expand_fold_left_optab_fn): Likewise.
14690 (direct_fold_left_optab_supported_p): Likewise.
14691 * fold-const-call.c (fold_const_fold_left): New function.
14692 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
14693 * tree-parloops.c (valid_reduction_p): New function.
14694 (gather_scalar_reductions): Use it.
14695 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
14696 (vect_finish_replace_stmt): Declare.
14697 * tree-vect-loop.c (fold_left_reduction_fn): New function.
14698 (needs_fold_left_reduction_p): New function, split out from...
14699 (vect_is_simple_reduction): ...here. Accept reductions that
14700 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
14701 (vect_force_simple_reduction): Also store the reduction type in
14702 the assignment's STMT_VINFO_REDUC_TYPE.
14703 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
14704 (merge_with_identity): New function.
14705 (vect_expand_fold_left): Likewise.
14706 (vectorize_fold_left_reduction): Likewise.
14707 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
14708 scalar phi in place for it. Check for target support and reject
14709 cases that would reassociate the operation. Defer the transform
14710 phase to vectorize_fold_left_reduction.
14711 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
14712 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
14713 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
14715 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
14717 * tree-if-conv.c (predicate_mem_writes): Remove redundant
14718 call to ifc_temp_var.
14720 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
14721 Alan Hayward <alan.hayward@arm.com>
14722 David Sherwood <david.sherwood@arm.com>
14724 * target.def (legitimize_address_displacement): Take the original
14725 offset as a poly_int.
14726 * targhooks.h (default_legitimize_address_displacement): Update
14728 * targhooks.c (default_legitimize_address_displacement): Likewise.
14729 * doc/tm.texi: Regenerate.
14730 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
14731 as an argument, moving assert of ad->disp == ad->disp_term to...
14732 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
14733 Try calling targetm.legitimize_address_displacement before expanding
14734 the address rather than afterwards, and adjust for the new interface.
14735 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
14736 Match the new hook interface. Handle SVE addresses.
14737 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
14738 new hook interface.
14740 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
14742 * Makefile.in (OBJS): Add early-remat.o.
14743 * target.def (select_early_remat_modes): New hook.
14744 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
14745 * doc/tm.texi: Regenerate.
14746 * targhooks.h (default_select_early_remat_modes): Declare.
14747 * targhooks.c (default_select_early_remat_modes): New function.
14748 * timevar.def (TV_EARLY_REMAT): New timevar.
14749 * passes.def (pass_early_remat): New pass.
14750 * tree-pass.h (make_pass_early_remat): Declare.
14751 * early-remat.c: New file.
14752 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
14754 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
14756 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
14757 Alan Hayward <alan.hayward@arm.com>
14758 David Sherwood <david.sherwood@arm.com>
14760 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
14761 vfm1 with a bound_epilog parameter.
14762 (vect_do_peeling): Update calls accordingly, and move the prologue
14763 call earlier in the function. Treat the base bound_epilog as 0 for
14764 fully-masked loops and retain vf - 1 for other loops. Add 1 to
14765 this base when peeling for gaps.
14766 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
14767 with fully-masked loops.
14768 (vect_estimate_min_profitable_iters): Handle the single peeled
14769 iteration in that case.
14771 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
14772 Alan Hayward <alan.hayward@arm.com>
14773 David Sherwood <david.sherwood@arm.com>
14775 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
14776 single-element interleaving even if the size is not a power of 2.
14777 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
14778 accesses for single-element interleaving if the group size is
14781 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
14782 Alan Hayward <alan.hayward@arm.com>
14783 David Sherwood <david.sherwood@arm.com>
14785 * doc/md.texi (fold_extract_last_@var{m}): Document.
14786 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
14787 * optabs.def (fold_extract_last_optab): New optab.
14788 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
14789 * internal-fn.c (fold_extract_direct): New macro.
14790 (expand_fold_extract_optab_fn): Likewise.
14791 (direct_fold_extract_optab_supported_p): Likewise.
14792 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
14793 * tree-vect-loop.c (vect_model_reduction_cost): Handle
14794 EXTRACT_LAST_REDUCTION.
14795 (get_initial_def_for_reduction): Do not create an initial vector
14796 for EXTRACT_LAST_REDUCTION reductions.
14797 (vectorizable_reduction): Leave the scalar phi in place for
14798 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
14799 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
14800 epilogue code for EXTRACT_LAST_REDUCTION and defer the
14801 transform phase to vectorizable_condition.
14802 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
14804 (vect_finish_stmt_generation): ...here.
14805 (vect_finish_replace_stmt): New function.
14806 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
14807 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
14809 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
14811 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
14812 Alan Hayward <alan.hayward@arm.com>
14813 David Sherwood <david.sherwood@arm.com>
14815 * doc/md.texi (extract_last_@var{m}): Document.
14816 * optabs.def (extract_last_optab): New optab.
14817 * internal-fn.def (EXTRACT_LAST): New internal function.
14818 * internal-fn.c (cond_unary_direct): New macro.
14819 (expand_cond_unary_optab_fn): Likewise.
14820 (direct_cond_unary_optab_supported_p): Likewise.
14821 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
14822 loops using EXTRACT_LAST.
14823 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
14824 (extract_last_<mode>): ...this optab.
14825 (vec_extract<mode><Vel>): Update accordingly.
14827 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
14828 Alan Hayward <alan.hayward@arm.com>
14829 David Sherwood <david.sherwood@arm.com>
14831 * target.def (empty_mask_is_expensive): New hook.
14832 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
14833 * doc/tm.texi: Regenerate.
14834 * targhooks.h (default_empty_mask_is_expensive): Declare.
14835 * targhooks.c (default_empty_mask_is_expensive): New function.
14836 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
14837 if the target says that empty masks are expensive.
14838 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
14840 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
14842 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
14843 Alan Hayward <alan.hayward@arm.com>
14844 David Sherwood <david.sherwood@arm.com>
14846 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
14847 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
14848 (vect_use_loop_mask_for_alignment_p): New function.
14849 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
14850 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
14851 niters_skip argument. Make sure that the first niters_skip elements
14852 of the first iteration are inactive.
14853 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
14854 Update call to vect_set_loop_masks_directly.
14855 (get_misalign_in_elems): New function, split out from...
14856 (vect_gen_prolog_loop_niters): ...here.
14857 (vect_update_init_of_dr): Take a code argument that specifies whether
14858 the adjustment should be added or subtracted.
14859 (vect_update_init_of_drs): Likewise.
14860 (vect_prepare_for_masked_peels): New function.
14861 (vect_do_peeling): Skip prologue peeling if we're using a mask
14862 instead. Update call to vect_update_inits_of_drs.
14863 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
14865 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
14866 alignment. Do not include the number of peeled iterations in
14867 the minimum threshold in that case.
14868 (vectorizable_induction): Adjust the start value down by
14869 LOOP_VINFO_MASK_SKIP_NITERS iterations.
14870 (vect_transform_loop): Call vect_prepare_for_masked_peels.
14871 Take the number of skipped iterations into account when calculating
14873 * tree-vect-stmts.c (vect_gen_while_not): New function.
14875 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
14876 Alan Hayward <alan.hayward@arm.com>
14877 David Sherwood <david.sherwood@arm.com>
14879 * doc/sourcebuild.texi (vect_fully_masked): Document.
14880 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
14881 default value to 0.
14882 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
14884 (vect_analyze_loop_2): ...here. Don't check the vectorization
14885 factor against the number of loop iterations if the loop is
14888 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
14889 Alan Hayward <alan.hayward@arm.com>
14890 David Sherwood <david.sherwood@arm.com>
14892 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
14893 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
14894 (dump_groups): Update accordingly.
14895 (iv_use::mem_type): New member variable.
14896 (address_p): New function.
14897 (record_use): Add a mem_type argument and initialize the new
14899 (record_group_use): Add a mem_type argument. Use address_p.
14900 Remove obsolete null checks of base_object. Update call to record_use.
14901 (find_interesting_uses_op): Update call to record_group_use.
14902 (find_interesting_uses_cond): Likewise.
14903 (find_interesting_uses_address): Likewise.
14904 (get_mem_type_for_internal_fn): New function.
14905 (find_address_like_use): Likewise.
14906 (find_interesting_uses_stmt): Try find_address_like_use before
14907 calling find_interesting_uses_op.
14908 (addr_offset_valid_p): Use the iv mem_type field as the type
14909 of the addressed memory.
14910 (add_autoinc_candidates): Likewise.
14911 (get_address_cost): Likewise.
14912 (split_small_address_groups_p): Use address_p.
14913 (split_address_groups): Likewise.
14914 (add_iv_candidate_for_use): Likewise.
14915 (autoinc_possible_for_pair): Likewise.
14916 (rewrite_groups): Likewise.
14917 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
14918 (determine_group_iv_cost): Update after split of USE_ADDRESS.
14919 (get_alias_ptr_type_for_ptr_address): New function.
14920 (rewrite_use_address): Rewrite address uses in calls that were
14921 identified by find_address_like_use.
14923 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
14924 Alan Hayward <alan.hayward@arm.com>
14925 David Sherwood <david.sherwood@arm.com>
14927 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
14929 * gimple-expr.h (is_gimple_addressable: Likewise.
14930 * gimple-expr.c (is_gimple_address): Likewise.
14931 * internal-fn.c (expand_call_mem_ref): New function.
14932 (expand_mask_load_optab_fn): Use it.
14933 (expand_mask_store_optab_fn): Likewise.
14935 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
14936 Alan Hayward <alan.hayward@arm.com>
14937 David Sherwood <david.sherwood@arm.com>
14939 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
14940 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
14941 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
14942 (cond_umax@var{mode}): Document.
14943 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
14944 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
14945 (cond_umin_optab, cond_umax_optab): New optabs.
14946 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
14947 (COND_IOR, COND_XOR): New internal functions.
14948 * internal-fn.h (get_conditional_internal_fn): Declare.
14949 * internal-fn.c (cond_binary_direct): New macro.
14950 (expand_cond_binary_optab_fn): Likewise.
14951 (direct_cond_binary_optab_supported_p): Likewise.
14952 (get_conditional_internal_fn): New function.
14953 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
14954 Cope with reduction statements that are vectorized as calls rather
14956 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
14957 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
14958 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
14959 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
14960 (UNSPEC_COND_EOR): New unspecs.
14961 (optab): Add mappings for them.
14962 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
14963 (sve_int_op, sve_fp_op): New int attributes.
14965 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
14966 Alan Hayward <alan.hayward@arm.com>
14967 David Sherwood <david.sherwood@arm.com>
14969 * optabs.def (while_ult_optab): New optab.
14970 * doc/md.texi (while_ult@var{m}@var{n}): Document.
14971 * internal-fn.def (WHILE_ULT): New internal function.
14972 * internal-fn.h (direct_internal_fn_supported_p): New override
14973 that takes two types as argument.
14974 * internal-fn.c (while_direct): New macro.
14975 (expand_while_optab_fn): New function.
14976 (convert_optab_supported_p): Likewise.
14977 (direct_while_optab_supported_p): New macro.
14978 * wide-int.h (wi::udiv_ceil): New function.
14979 * tree-vectorizer.h (rgroup_masks): New structure.
14980 (vec_loop_masks): New typedef.
14981 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
14982 and fully_masked_p.
14983 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
14984 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
14985 (vect_max_vf): New function.
14986 (slpeel_make_loop_iterate_ntimes): Delete.
14987 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
14988 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
14989 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
14990 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
14991 internal-fn.h, stor-layout.h and optabs-query.h.
14992 (vect_set_loop_mask): New function.
14993 (add_preheader_seq): Likewise.
14994 (add_header_seq): Likewise.
14995 (interleave_supported_p): Likewise.
14996 (vect_maybe_permute_loop_masks): Likewise.
14997 (vect_set_loop_masks_directly): Likewise.
14998 (vect_set_loop_condition_masked): Likewise.
14999 (vect_set_loop_condition_unmasked): New function, split out from
15000 slpeel_make_loop_iterate_ntimes.
15001 (slpeel_make_loop_iterate_ntimes): Rename to..
15002 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
15003 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
15004 (vect_do_peeling): Update call accordingly.
15005 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
15007 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
15008 mask_compare_type, can_fully_mask_p and fully_masked_p.
15009 (release_vec_loop_masks): New function.
15010 (_loop_vec_info): Use it to free the loop masks.
15011 (can_produce_all_loop_masks_p): New function.
15012 (vect_get_max_nscalars_per_iter): Likewise.
15013 (vect_verify_full_masking): Likewise.
15014 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
15015 retries, and free the mask rgroups before retrying. Check loop-wide
15016 reasons for disallowing fully-masked loops. Make the final decision
15017 about whether use a fully-masked loop or not.
15018 (vect_estimate_min_profitable_iters): Do not assume that peeling
15019 for the number of iterations will be needed for fully-masked loops.
15020 (vectorizable_reduction): Disable fully-masked loops.
15021 (vectorizable_live_operation): Likewise.
15022 (vect_halve_mask_nunits): New function.
15023 (vect_double_mask_nunits): Likewise.
15024 (vect_record_loop_mask): Likewise.
15025 (vect_get_loop_mask): Likewise.
15026 (vect_transform_loop): Handle the case in which the final loop
15027 iteration might handle a partial vector. Call vect_set_loop_condition
15028 instead of slpeel_make_loop_iterate_ntimes.
15029 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
15030 (check_load_store_masking): New function.
15031 (prepare_load_store_mask): Likewise.
15032 (vectorizable_store): Handle fully-masked loops.
15033 (vectorizable_load): Likewise.
15034 (supportable_widening_operation): Use vect_halve_mask_nunits for
15036 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
15037 (vect_gen_while): New function.
15038 * config/aarch64/aarch64.md (umax<mode>3): New expander.
15039 (aarch64_uqdec<mode>): New insn.
15041 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
15042 Alan Hayward <alan.hayward@arm.com>
15043 David Sherwood <david.sherwood@arm.com>
15045 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
15046 (reduc_xor_scal_optab): New optabs.
15047 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
15048 (reduc_xor_scal_@var{m}): Document.
15049 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
15050 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
15051 internal functions.
15052 * fold-const-call.c (fold_const_call): Handle them.
15053 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
15054 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
15055 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
15056 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
15057 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
15058 (UNSPEC_XORV): New unspecs.
15059 (optab): Add entries for them.
15060 (BITWISEV): New int iterator.
15061 (bit_reduc_op): New int attributes.
15063 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
15064 Alan Hayward <alan.hayward@arm.com>
15065 David Sherwood <david.sherwood@arm.com>
15067 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
15068 * internal-fn.def (VEC_SHL_INSERT): New internal function.
15069 * optabs.def (vec_shl_insert_optab): New optab.
15070 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
15071 (duplicate_and_interleave): Likewise.
15072 * tree-vect-loop.c: Include internal-fn.h.
15073 (neutral_op_for_slp_reduction): New function, split out from
15074 get_initial_defs_for_reduction.
15075 (get_initial_def_for_reduction): Handle option 2 for variable-length
15076 vectors by loading the neutral value into a vector and then shifting
15077 the initial value into element 0.
15078 (get_initial_defs_for_reduction): Replace the code argument with
15079 the neutral value calculated by neutral_op_for_slp_reduction.
15080 Use gimple_build_vector for constant-length vectors.
15081 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
15082 but the first group_size elements have a neutral value.
15083 Use duplicate_and_interleave otherwise.
15084 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
15085 Update call to get_initial_defs_for_reduction. Handle SLP
15086 reductions for variable-length vectors by creating one vector
15087 result for each scalar result, with the elements associated
15088 with other scalar results stubbed out with the neutral value.
15089 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
15090 Require IFN_VEC_SHL_INSERT for double reductions on
15091 variable-length vectors, or SLP reductions that have
15092 a neutral value. Require can_duplicate_and_interleave_p
15093 support for variable-length unchained SLP reductions if there
15094 is no neutral value, such as for MIN/MAX reductions. Also require
15095 the number of vector elements to be a multiple of the number of
15096 SLP statements when doing variable-length unchained SLP reductions.
15097 Update call to vect_create_epilog_for_reduction.
15098 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
15099 and remove initial values.
15100 (duplicate_and_interleave): Make public.
15101 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
15102 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
15104 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
15105 Alan Hayward <alan.hayward@arm.com>
15106 David Sherwood <david.sherwood@arm.com>
15108 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
15109 (can_duplicate_and_interleave_p): New function.
15110 (vect_get_and_check_slp_defs): Take the vector of statements
15111 rather than just the current one. Remove excess parentheses.
15112 Restriction rejectinon of vect_constant_def and vect_external_def
15113 for variable-length vectors to boolean types, or types for which
15114 can_duplicate_and_interleave_p is false.
15115 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
15116 (duplicate_and_interleave): New function.
15117 (vect_get_constant_vectors): Use gimple_build_vector for
15118 constant-length vectors and suitable variable-length constant
15119 vectors. Use duplicate_and_interleave for other variable-length
15120 vectors. Don't defer the update when inserting new statements.
15122 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
15123 Alan Hayward <alan.hayward@arm.com>
15124 David Sherwood <david.sherwood@arm.com>
15126 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
15127 min_profitable_iters doesn't go negative.
15129 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
15130 Alan Hayward <alan.hayward@arm.com>
15131 David Sherwood <david.sherwood@arm.com>
15133 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
15134 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
15135 * optabs.def (vec_mask_load_lanes_optab): New optab.
15136 (vec_mask_store_lanes_optab): Likewise.
15137 * internal-fn.def (MASK_LOAD_LANES): New internal function.
15138 (MASK_STORE_LANES): Likewise.
15139 * internal-fn.c (mask_load_lanes_direct): New macro.
15140 (mask_store_lanes_direct): Likewise.
15141 (expand_mask_load_optab_fn): Handle masked operations.
15142 (expand_mask_load_lanes_optab_fn): New macro.
15143 (expand_mask_store_optab_fn): Handle masked operations.
15144 (expand_mask_store_lanes_optab_fn): New macro.
15145 (direct_mask_load_lanes_optab_supported_p): Likewise.
15146 (direct_mask_store_lanes_optab_supported_p): Likewise.
15147 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
15149 (vect_load_lanes_supported): Likewise.
15150 * tree-vect-data-refs.c (strip_conversion): New function.
15151 (can_group_stmts_p): Likewise.
15152 (vect_analyze_data_ref_accesses): Use it instead of checking
15153 for a pair of assignments.
15154 (vect_store_lanes_supported): Take a masked_p parameter.
15155 (vect_load_lanes_supported): Likewise.
15156 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
15157 vect_store_lanes_supported and vect_load_lanes_supported.
15158 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
15159 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
15160 parameter. Don't allow gaps for masked accesses.
15161 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
15162 and vect_load_lanes_supported.
15163 (get_load_store_type): Take a masked_p parameter and update
15164 call to get_group_load_store_type.
15165 (vectorizable_store): Update call to get_load_store_type.
15166 Handle IFN_MASK_STORE_LANES.
15167 (vectorizable_load): Update call to get_load_store_type.
15168 Handle IFN_MASK_LOAD_LANES.
15170 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
15171 Alan Hayward <alan.hayward@arm.com>
15172 David Sherwood <david.sherwood@arm.com>
15174 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
15176 * config/aarch64/aarch64-protos.h
15177 (aarch64_sve_struct_memory_operand_p): Declare.
15178 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
15179 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
15180 (VPRED, vpred): Handle SVE structure modes.
15181 * config/aarch64/constraints.md (Utx): New constraint.
15182 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
15183 (aarch64_sve_struct_nonimmediate_operand): New predicates.
15184 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
15185 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
15186 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
15187 structure modes. Split into pieces after RA.
15188 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
15189 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
15191 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
15192 SVE structure modes.
15193 (aarch64_classify_address): Likewise.
15194 (sizetochar): Move earlier in file.
15195 (aarch64_print_operand): Handle SVE register lists.
15196 (aarch64_array_mode): New function.
15197 (aarch64_sve_struct_memory_operand_p): Likewise.
15198 (TARGET_ARRAY_MODE): Redefine.
15200 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
15201 Alan Hayward <alan.hayward@arm.com>
15202 David Sherwood <david.sherwood@arm.com>
15204 * target.def (array_mode): New target hook.
15205 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
15206 * doc/tm.texi: Regenerate.
15207 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
15208 * hooks.c (hook_optmode_mode_uhwi_none): New function.
15209 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
15210 targetm.array_mode.
15211 * stor-layout.c (mode_for_array): Likewise. Support polynomial
15214 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
15215 Alan Hayward <alan.hayward@arm.com>
15216 David Sherwood <david.sherwood@arm.com>
15218 * fold-const.c (fold_binary_loc): Check the argument types
15219 rather than the result type when testing for a vector operation.
15221 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
15223 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
15224 * doc/tm.texi: Regenerate.
15226 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
15227 Alan Hayward <alan.hayward@arm.com>
15228 David Sherwood <david.sherwood@arm.com>
15230 * doc/invoke.texi (-msve-vector-bits=): Document new option.
15231 (sve): Document new AArch64 extension.
15232 * doc/md.texi (w): Extend the description of the AArch64
15233 constraint to include SVE vectors.
15234 (Upl, Upa): Document new AArch64 predicate constraints.
15235 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
15237 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
15238 (msve-vector-bits=): New option.
15239 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
15240 SVE when these are disabled.
15241 (sve): New extension.
15242 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
15243 modes. Adjust their number of units based on aarch64_sve_vg.
15244 (MAX_BITSIZE_MODE_ANY_MODE): Define.
15245 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
15246 aarch64_addr_query_type.
15247 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
15248 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
15249 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
15250 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
15251 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
15252 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
15253 (aarch64_simd_imm_zero_p): Delete.
15254 (aarch64_check_zero_based_sve_index_immediate): Declare.
15255 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
15256 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
15257 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
15258 (aarch64_sve_float_mul_immediate_p): Likewise.
15259 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
15260 rather than an rtx.
15261 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
15262 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
15263 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
15264 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
15265 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
15266 (aarch64_regmode_natural_size): Likewise.
15267 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
15268 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
15270 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
15271 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
15272 for VG and the SVE predicate registers.
15273 (V_ALIASES): Add a "z"-prefixed alias.
15274 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
15275 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
15276 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
15277 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
15278 (REG_CLASS_NAMES): Add entries for them.
15279 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
15280 and the predicate registers.
15281 (aarch64_sve_vg): Declare.
15282 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
15283 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
15284 (REGMODE_NATURAL_SIZE): Define.
15285 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
15287 * config/aarch64/aarch64.c: Include cfgrtl.h.
15288 (simd_immediate_info): Add a constructor for series vectors,
15289 and an associated step field.
15290 (aarch64_sve_vg): New variable.
15291 (aarch64_dbx_register_number): Handle VG and the predicate registers.
15292 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
15293 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
15294 (VEC_ANY_DATA, VEC_STRUCT): New constants.
15295 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
15296 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
15297 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
15298 (aarch64_get_mask_mode): New functions.
15299 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
15300 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
15301 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
15302 predicate modes and predicate registers. Explicitly restrict
15303 GPRs to modes of 16 bytes or smaller. Only allow FP registers
15304 to store a vector mode if it is recognized by
15305 aarch64_classify_vector_mode.
15306 (aarch64_regmode_natural_size): New function.
15307 (aarch64_hard_regno_caller_save_mode): Return the original mode
15309 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
15310 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
15311 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
15312 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
15314 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
15315 does not overlap dest if the function is frame-related. Handle
15317 (aarch64_split_add_offset): New function.
15318 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
15319 them aarch64_add_offset.
15320 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
15321 and update call to aarch64_sub_sp.
15322 (aarch64_add_cfa_expression): New function.
15323 (aarch64_expand_prologue): Pass extra temporary registers to the
15324 functions above. Handle the case in which we need to emit new
15325 DW_CFA_expressions for registers that were originally saved
15326 relative to the stack pointer, but now have to be expressed
15327 relative to the frame pointer.
15328 (aarch64_output_mi_thunk): Pass extra temporary registers to the
15330 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
15331 IP0 and IP1 values for SVE frames.
15332 (aarch64_expand_vec_series): New function.
15333 (aarch64_expand_sve_widened_duplicate): Likewise.
15334 (aarch64_expand_sve_const_vector): Likewise.
15335 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
15336 Handle SVE constants. Use emit_move_insn to move a force_const_mem
15337 into the register, rather than emitting a SET directly.
15338 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
15339 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
15340 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
15341 (offset_9bit_signed_scaled_p): New functions.
15342 (aarch64_replicate_bitmask_imm): New function.
15343 (aarch64_bitmask_imm): Use it.
15344 (aarch64_cannot_force_const_mem): Reject expressions involving
15345 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
15346 (aarch64_classify_index): Handle SVE indices, by requiring
15347 a plain register index with a scale that matches the element size.
15348 (aarch64_classify_address): Handle SVE addresses. Assert that
15349 the mode of the address is VOIDmode or an integer mode.
15350 Update call to aarch64_classify_symbol.
15351 (aarch64_classify_symbolic_expression): Update call to
15352 aarch64_classify_symbol.
15353 (aarch64_const_vec_all_in_range_p): New function.
15354 (aarch64_print_vector_float_operand): Likewise.
15355 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
15356 "vN" for FP registers with SVE modes. Handle (const ...) vectors
15357 and the FP immediates 1.0 and 0.5.
15358 (aarch64_print_address_internal): Handle SVE addresses.
15359 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
15360 (aarch64_regno_regclass): Handle predicate registers.
15361 (aarch64_secondary_reload): Handle big-endian reloads of SVE
15363 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
15364 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
15365 (aarch64_convert_sve_vector_bits): New function.
15366 (aarch64_override_options): Use it to handle -msve-vector-bits=.
15367 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
15368 rather than an rtx.
15369 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
15370 Handle SVE vector and predicate modes. Accept VL-based constants
15371 that need only one temporary register, and VL offsets that require
15372 no temporary registers.
15373 (aarch64_conditional_register_usage): Mark the predicate registers
15374 as fixed if SVE isn't available.
15375 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
15376 Return true for SVE vector and predicate modes.
15377 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
15378 rather than an unsigned int. Handle SVE modes.
15379 (aarch64_preferred_simd_mode): Update call accordingly. Handle
15381 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
15383 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
15384 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
15385 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
15386 (aarch64_sve_float_mul_immediate_p): New functions.
15387 (aarch64_sve_valid_immediate): New function.
15388 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
15389 Explicitly reject structure modes. Check for INDEX constants.
15390 Handle PTRUE and PFALSE constants.
15391 (aarch64_check_zero_based_sve_index_immediate): New function.
15392 (aarch64_simd_imm_zero_p): Delete.
15393 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
15394 vector modes. Accept constants in the range of CNT[BHWD].
15395 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
15396 ask for an Advanced SIMD mode.
15397 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
15398 (aarch64_simd_vector_alignment): Handle SVE predicates.
15399 (aarch64_vectorize_preferred_vector_alignment): New function.
15400 (aarch64_simd_vector_alignment_reachable): Use it instead of
15402 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
15403 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
15405 (MAX_VECT_LEN): Delete.
15406 (expand_vec_perm_d): Add a vec_flags field.
15407 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
15408 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
15409 (aarch64_evpc_ext): Don't apply a big-endian lane correction
15411 (aarch64_evpc_rev): Rename to...
15412 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
15413 (aarch64_evpc_rev_global): New function.
15414 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
15415 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
15417 (aarch64_evpc_sve_tbl): New function.
15418 (aarch64_expand_vec_perm_const_1): Update after rename of
15419 aarch64_evpc_rev. Handle SVE permutes too, trying
15420 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
15421 than aarch64_evpc_tbl.
15422 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
15423 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
15424 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
15425 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
15426 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
15427 (aarch64_expand_sve_vcond): New functions.
15428 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
15429 of aarch64_vector_mode_p.
15430 (aarch64_dwarf_poly_indeterminate_value): New function.
15431 (aarch64_compute_pressure_classes): Likewise.
15432 (aarch64_can_change_mode_class): Likewise.
15433 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
15434 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
15435 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
15436 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
15437 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
15438 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
15439 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
15440 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
15442 (Dn, Dl, Dr): Accept const as well as const_vector.
15443 (Dz): Likewise. Compare against CONST0_RTX.
15444 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
15445 of "vector" where appropriate.
15446 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
15447 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
15448 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
15449 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
15450 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
15451 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
15452 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
15453 (v_int_equiv): Extend to SVE modes.
15454 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
15456 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
15457 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
15458 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
15459 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
15460 (SVE_COND_FP_CMP): New int iterators.
15461 (perm_hilo): Handle the new unpack unspecs.
15462 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
15464 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
15465 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
15466 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
15467 (aarch64_equality_operator, aarch64_constant_vector_operand)
15468 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
15469 (aarch64_sve_nonimmediate_operand): Likewise.
15470 (aarch64_sve_general_operand): Likewise.
15471 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
15472 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
15473 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
15474 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
15475 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
15476 (aarch64_sve_float_arith_immediate): Likewise.
15477 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
15478 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
15479 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
15480 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
15481 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
15482 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
15483 (aarch64_sve_float_arith_operand): Likewise.
15484 (aarch64_sve_float_arith_with_sub_operand): Likewise.
15485 (aarch64_sve_float_mul_operand): Likewise.
15486 (aarch64_sve_vec_perm_operand): Likewise.
15487 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
15488 (aarch64_mov_operand): Accept const_poly_int and const_vector.
15489 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
15490 as well as const_vector.
15491 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
15492 in file. Use CONST0_RTX and CONSTM1_RTX.
15493 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
15494 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
15495 Use aarch64_simd_imm_zero.
15496 * config/aarch64/aarch64-sve.md: New file.
15497 * config/aarch64/aarch64.md: Include it.
15498 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
15499 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
15500 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
15501 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
15502 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
15503 (sve): New attribute.
15504 (enabled): Disable instructions with the sve attribute unless
15506 (movqi, movhi): Pass CONST_POLY_INT operaneds through
15507 aarch64_expand_mov_immediate.
15508 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
15509 CNT[BHSD] immediates.
15510 (movti): Split CONST_POLY_INT moves into two halves.
15511 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
15512 Split additions that need a temporary here if the destination
15513 is the stack pointer.
15514 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
15515 (*add<mode>3_poly_1): New instruction.
15516 (set_clobber_cc): New expander.
15518 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
15520 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
15521 parameter and use it instead of GET_MODE_SIZE (innermode). Use
15522 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
15523 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
15524 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
15525 Change innermode from fixed_mode_size to machine_mode.
15526 (simplify_subreg): Update call accordingly. Handle a constant-sized
15527 subreg of a variable-length CONST_VECTOR.
15529 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
15530 Alan Hayward <alan.hayward@arm.com>
15531 David Sherwood <david.sherwood@arm.com>
15533 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
15534 (add_offset_to_base): New function, split out from...
15535 (create_mem_ref): ...here. When handling a scale other than 1,
15536 check first whether the address is valid without the offset.
15537 Add it into the base if so, leaving the index and scale as-is.
15539 2018-01-12 Jakub Jelinek <jakub@redhat.com>
15542 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
15543 fold_for_warn before checking if arg2 is INTEGER_CST.
15545 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
15547 * config/rs6000/predicates.md (load_multiple_operation): Delete.
15548 (store_multiple_operation): Delete.
15549 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
15550 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
15551 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
15552 guarded by TARGET_STRING.
15553 (rs6000_output_load_multiple): Delete.
15554 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
15555 OPTION_MASK_STRING / TARGET_STRING handling.
15556 (print_operand) <'N', 'O'>: Add comment that these are unused now.
15557 (const rs6000_opt_masks) <"string">: Change mask to 0.
15558 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
15559 (MASK_STRING): Delete.
15560 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
15562 (load_multiple): Delete.
15569 (store_multiple): Delete.
15576 (movmemsi_8reg): Delete.
15577 (corresponding unnamed define_insn): Delete.
15578 (movmemsi_6reg): Delete.
15579 (corresponding unnamed define_insn): Delete.
15580 (movmemsi_4reg): Delete.
15581 (corresponding unnamed define_insn): Delete.
15582 (movmemsi_2reg): Delete.
15583 (corresponding unnamed define_insn): Delete.
15584 (movmemsi_1reg): Delete.
15585 (corresponding unnamed define_insn): Delete.
15586 * config/rs6000/rs6000.opt (mno-string): New.
15587 (mstring): Replace by deprecation warning stub.
15588 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
15590 2018-01-12 Jakub Jelinek <jakub@redhat.com>
15592 * regrename.c (regrename_do_replace): If replacing the same
15593 reg multiple times, try to reuse last created gen_raw_REG.
15596 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
15597 main to workaround a bug in GDB.
15599 2018-01-12 Tom de Vries <tom@codesourcery.com>
15602 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
15604 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
15606 PR rtl-optimization/80481
15607 * ira-color.c (get_cap_member): New function.
15608 (allocnos_conflict_by_live_ranges_p): Use it.
15609 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
15610 (setup_slot_coalesced_allocno_live_ranges): Ditto.
15612 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
15615 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
15616 (*saddl_se_1): Ditto.
15617 (*ssubsi_1): Ditto.
15618 (*ssubl_se_1): Ditto.
15620 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
15622 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
15623 rather than wi::to_widest for DR_INITs.
15624 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
15625 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
15626 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
15628 (vect_analyze_group_access_1): Note that here.
15630 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
15632 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
15633 polynomial type sizes.
15635 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
15637 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
15638 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
15639 (gimple_add_tmp_var): Likewise.
15641 2018-01-12 Martin Liska <mliska@suse.cz>
15643 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
15644 (gimple_alloc_sizes): Likewise.
15645 (dump_gimple_statistics): Use PRIu64 in printf format.
15646 * gimple.h: Change uint64_t to int.
15648 2018-01-12 Martin Liska <mliska@suse.cz>
15650 * tree-core.h: Use uint64_t instead of int.
15651 * tree.c (tree_node_counts): Likewise.
15652 (tree_node_sizes): Likewise.
15653 (dump_tree_statistics): Use PRIu64 in printf format.
15655 2018-01-12 Martin Liska <mliska@suse.cz>
15657 * Makefile.in: As qsort_chk is implemented in vec.c, add
15658 vec.o to linkage of gencfn-macros.
15659 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
15660 passing the info to record_node_allocation_statistics.
15661 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
15663 * ggc-common.c (struct ggc_usage): Add operator== and use
15664 it in operator< and compare function.
15665 * mem-stats.h (struct mem_usage): Likewise.
15666 * vec.c (struct vec_usage): Remove operator< and compare
15667 function. Can be simply inherited.
15669 2018-01-12 Martin Jambor <mjambor@suse.cz>
15672 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
15673 * tree-ssa-math-opts.c: Include domwalk.h.
15674 (convert_mult_to_fma_1): New function.
15675 (fma_transformation_info): New type.
15676 (fma_deferring_state): Likewise.
15677 (cancel_fma_deferring): New function.
15678 (result_of_phi): Likewise.
15679 (last_fma_candidate_feeds_initial_phi): Likewise.
15680 (convert_mult_to_fma): Added deferring logic, split actual
15681 transformation to convert_mult_to_fma_1.
15682 (math_opts_dom_walker): New type.
15683 (math_opts_dom_walker::after_dom_children): New method, body moved
15684 here from pass_optimize_widening_mul::execute, added deferring logic
15686 (pass_optimize_widening_mul::execute): Moved most of code to
15687 math_opts_dom_walker::after_dom_children.
15688 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
15689 * config/i386/i386.c (ix86_option_override_internal): Added
15690 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
15692 2018-01-12 Richard Biener <rguenther@suse.de>
15695 * dwarf2out.c (gen_variable_die): Do not reset old_die for
15696 inline instance vars.
15698 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
15701 * config/rx/rx.c (rx_is_restricted_memory_address):
15702 Handle SUBREG case.
15704 2018-01-12 Richard Biener <rguenther@suse.de>
15706 PR tree-optimization/80846
15707 * target.def (split_reduction): New target hook.
15708 * targhooks.c (default_split_reduction): New function.
15709 * targhooks.h (default_split_reduction): Declare.
15710 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
15711 target requests first reduce vectors by combining low and high
15713 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
15714 (get_vectype_for_scalar_type_and_size): Export.
15715 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
15716 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
15717 * doc/tm.texi: Regenerate.
15718 * config/i386/i386.c (ix86_split_reduction): Implement
15719 TARGET_VECTORIZE_SPLIT_REDUCTION.
15721 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
15724 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
15725 in PIC mode except for TARGET_VXWORKS_RTP.
15726 * config/sparc/sparc.c: Include cfgrtl.h.
15727 (TARGET_INIT_PIC_REG): Define.
15728 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
15729 (sparc_pic_register_p): New predicate.
15730 (sparc_legitimate_address_p): Use it.
15731 (sparc_legitimize_pic_address): Likewise.
15732 (sparc_delegitimize_address): Likewise.
15733 (sparc_mode_dependent_address_p): Likewise.
15734 (gen_load_pcrel_sym): Remove 4th parameter.
15735 (load_got_register): Adjust call to above. Remove obsolete stuff.
15736 (sparc_expand_prologue): Do not call load_got_register here.
15737 (sparc_flat_expand_prologue): Likewise.
15738 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
15739 (sparc_use_pseudo_pic_reg): New function.
15740 (sparc_init_pic_reg): Likewise.
15741 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
15742 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
15744 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
15746 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
15747 Add item for branch_cost.
15749 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
15751 PR rtl-optimization/83565
15752 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
15753 not extend the result to a larger mode for rotate operations.
15754 (num_sign_bit_copies1): Likewise.
15756 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15759 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
15761 Use values-Xc.o for -pedantic.
15762 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
15764 2018-01-12 Martin Liska <mliska@suse.cz>
15767 * ipa-devirt.c (final_warning_record::grow_type_warnings):
15769 (possible_polymorphic_call_targets): Use it.
15770 (ipa_devirt): Likewise.
15772 2018-01-12 Martin Liska <mliska@suse.cz>
15774 * profile-count.h (enum profile_quality): Use 0 as invalid
15775 enum value of profile_quality.
15777 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
15779 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
15780 -mext-string options.
15782 2018-01-12 Richard Biener <rguenther@suse.de>
15784 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
15785 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
15786 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
15788 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
15790 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
15792 * configure.ac (--with-long-double-format): Add support for the
15793 configuration option to change the default long double format on
15795 * config.gcc (powerpc*-linux*-*): Likewise.
15796 * configure: Regenerate.
15797 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
15798 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
15799 used without modification.
15801 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15803 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
15804 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
15805 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
15806 MISC_BUILTIN_SPEC_BARRIER.
15807 (rs6000_init_builtins): Likewise.
15808 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
15810 (speculation_barrier): New define_insn.
15811 * doc/extend.texi: Document __builtin_speculation_barrier.
15813 2018-01-11 Jakub Jelinek <jakub@redhat.com>
15816 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
15817 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
15818 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
15820 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
15821 integral modes instead of "ss" and "sd".
15822 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
15823 vectors with 32-bit and 64-bit elements.
15824 (vecdupssescalarmodesuffix): New mode attribute.
15825 (vec_dup<mode>): Use it.
15827 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
15830 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
15831 frame if argument is passed on stack.
15833 2018-01-11 Jakub Jelinek <jakub@redhat.com>
15836 * ree.c (combine_reaching_defs): Optimize also
15837 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
15838 reg2=any_extend(exp); reg1=reg2;, formatting fix.
15840 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
15842 PR middle-end/83189
15843 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
15845 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
15847 PR middle-end/83718
15848 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
15849 after they are computed.
15851 2018-01-11 Bin Cheng <bin.cheng@arm.com>
15853 PR tree-optimization/83695
15854 * gimple-loop-linterchange.cc
15855 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
15856 reset cached scev information after interchange.
15857 (pass_linterchange::execute): Remove call to scev_reset_htab.
15859 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15861 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
15862 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
15863 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
15864 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
15865 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
15866 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
15867 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
15868 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
15869 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
15870 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
15871 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
15872 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
15873 (V_lane_reg): Likewise.
15874 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
15876 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
15877 (vfmal_lane_low<mode>_intrinsic,
15878 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
15879 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
15880 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
15881 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
15882 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
15883 vfmsl_lane_high<mode>_intrinsic): New define_insns.
15885 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15887 * config/arm/arm-cpus.in (fp16fml): New feature.
15888 (ALL_SIMD): Add fp16fml.
15889 (armv8.2-a): Add fp16fml as an option.
15890 (armv8.3-a): Likewise.
15891 (armv8.4-a): Add fp16fml as part of fp16.
15892 * config/arm/arm.h (TARGET_FP16FML): Define.
15893 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
15895 * config/arm/arm-modes.def (V2HF): Define.
15896 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
15897 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
15898 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
15899 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
15900 vfmsl_low, vfmsl_high): New set of builtins.
15901 * config/arm/iterators.md (PLUSMINUS): New code iterator.
15902 (vfml_op): New code attribute.
15903 (VFMLHALVES): New int iterator.
15904 (VFML, VFMLSEL): New mode attributes.
15905 (V_reg): Define mapping for V2HF.
15906 (V_hi, V_lo): New mode attributes.
15907 (VF_constraint): Likewise.
15908 (vfml_half, vfml_half_selector): New int attributes.
15909 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
15911 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
15912 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
15914 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
15915 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
15916 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
15917 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
15919 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
15920 Document new effective target and option set.
15922 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15924 * config/arm/arm-cpus.in (armv8_4): New feature.
15925 (ARMv8_4a): New fgroup.
15926 (armv8.4-a): New arch.
15927 * config/arm/arm-tables.opt: Regenerate.
15928 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
15929 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
15930 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
15931 Add matching rules for -march=armv8.4-a and extensions.
15932 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
15934 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
15937 * config/rx/rx.md (BW): New mode attribute.
15938 (sync_lock_test_and_setsi): Add mode suffix to insn output.
15940 2018-01-11 Richard Biener <rguenther@suse.de>
15942 PR tree-optimization/83435
15943 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
15944 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
15945 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
15947 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
15948 Alan Hayward <alan.hayward@arm.com>
15949 David Sherwood <david.sherwood@arm.com>
15951 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
15953 (aarch64_classify_address): Initialize it. Track polynomial offsets.
15954 (aarch64_print_address_internal): Use it to check for a zero offset.
15956 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
15957 Alan Hayward <alan.hayward@arm.com>
15958 David Sherwood <david.sherwood@arm.com>
15960 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
15961 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
15962 Return a poly_int64 rather than a HOST_WIDE_INT.
15963 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
15964 rather than a HOST_WIDE_INT.
15965 * config/aarch64/aarch64.h (aarch64_frame): Protect with
15966 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
15967 hard_fp_offset, frame_size, initial_adjust, callee_offset and
15968 final_offset from HOST_WIDE_INT to poly_int64.
15969 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
15970 to_constant when getting the number of units in an Advanced SIMD
15972 (aarch64_builtin_vectorized_function): Check for a constant number
15974 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
15976 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
15977 attribute instead of GET_MODE_NUNITS.
15978 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
15979 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
15980 GET_MODE_SIZE for fixed-size registers.
15981 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
15982 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
15983 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
15984 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
15985 (aarch64_print_operand, aarch64_print_address_internal)
15986 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
15987 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
15988 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
15989 Handle polynomial GET_MODE_SIZE.
15990 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
15991 wider than SImode without modification.
15992 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
15993 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
15994 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
15995 passing and returning SVE modes.
15996 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
15997 rather than GEN_INT.
15998 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
15999 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
16000 (aarch64_allocate_and_probe_stack_space): Likewise.
16001 (aarch64_layout_frame): Cope with polynomial offsets.
16002 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
16003 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
16004 polynomial offsets.
16005 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
16006 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
16007 poly_int64 rather than a HOST_WIDE_INT.
16008 (aarch64_get_separate_components, aarch64_process_components)
16009 (aarch64_expand_prologue, aarch64_expand_epilogue)
16010 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
16011 (aarch64_anchor_offset): New function, split out from...
16012 (aarch64_legitimize_address): ...here.
16013 (aarch64_builtin_vectorization_cost): Handle polynomial
16014 TYPE_VECTOR_SUBPARTS.
16015 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
16017 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
16018 number of elements from the PARALLEL rather than the mode.
16019 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
16020 rather than GET_MODE_BITSIZE.
16021 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
16022 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
16023 (aarch64_expand_vec_perm_const_1): Handle polynomial
16024 d->perm.length () and d->perm elements.
16025 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
16026 Apply to_constant to d->perm elements.
16027 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
16028 polynomial CONST_VECTOR_NUNITS.
16029 (aarch64_move_pointer): Take amount as a poly_int64 rather
16031 (aarch64_progress_pointer): Avoid temporary variable.
16032 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
16033 the mode attribute instead of GET_MODE.
16035 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
16036 Alan Hayward <alan.hayward@arm.com>
16037 David Sherwood <david.sherwood@arm.com>
16039 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
16040 x exists before using it.
16041 (aarch64_add_constant_internal): Rename to...
16042 (aarch64_add_offset_1): ...this. Replace regnum with separate
16043 src and dest rtxes. Handle the case in which they're different,
16044 including when the offset is zero. Replace scratchreg with an rtx.
16045 Use 2 additions if there is no spare register into which we can
16046 move a 16-bit constant.
16047 (aarch64_add_constant): Delete.
16048 (aarch64_add_offset): Replace reg with separate src and dest
16049 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
16050 Use aarch64_add_offset_1.
16051 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
16052 an rtx rather than an int. Take the delta as a poly_int64
16053 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
16054 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
16055 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
16056 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
16057 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
16058 and aarch64_add_sp.
16059 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
16060 aarch64_add_constant.
16062 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
16064 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
16065 Use scalar_float_mode.
16067 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
16069 * config/aarch64/aarch64-simd.md
16070 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
16071 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
16072 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
16073 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
16074 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
16075 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
16076 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
16077 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
16078 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
16079 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
16081 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16084 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
16085 targ_options->x_arm_arch_string is non NULL.
16087 2018-01-11 Tamar Christina <tamar.christina@arm.com>
16089 * config/aarch64/aarch64.h
16090 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
16092 2018-01-11 Sudakshina Das <sudi.das@arm.com>
16095 * expmed.c (emit_store_flag_force): Swap if const op0
16096 and change VOIDmode to mode of op0.
16098 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
16100 PR rtl-optimization/83761
16101 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
16102 than bytes to mode_for_size.
16104 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
16106 PR middle-end/83189
16107 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
16108 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
16111 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
16113 PR middle-end/83575
16114 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
16115 when in layout mode.
16116 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
16117 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
16120 2018-01-10 Michael Collison <michael.collison@arm.com>
16122 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
16123 * config/aarch64/aarch64-option-extension.def: Add
16124 AARCH64_OPT_EXTENSION of 'fp16fml'.
16125 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
16126 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
16127 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
16128 * config/aarch64/constraints.md (Ui7): New constraint.
16129 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
16130 (VFMLA_SEL_W): Ditto.
16133 (VFMLA16_LOW): New int iterator.
16134 (VFMLA16_HIGH): Ditto.
16135 (UNSPEC_FMLAL): New unspec.
16136 (UNSPEC_FMLSL): Ditto.
16137 (UNSPEC_FMLAL2): Ditto.
16138 (UNSPEC_FMLSL2): Ditto.
16139 (f16mac): New code attribute.
16140 * config/aarch64/aarch64-simd-builtins.def
16141 (aarch64_fmlal_lowv2sf): Ditto.
16142 (aarch64_fmlsl_lowv2sf): Ditto.
16143 (aarch64_fmlalq_lowv4sf): Ditto.
16144 (aarch64_fmlslq_lowv4sf): Ditto.
16145 (aarch64_fmlal_highv2sf): Ditto.
16146 (aarch64_fmlsl_highv2sf): Ditto.
16147 (aarch64_fmlalq_highv4sf): Ditto.
16148 (aarch64_fmlslq_highv4sf): Ditto.
16149 (aarch64_fmlal_lane_lowv2sf): Ditto.
16150 (aarch64_fmlsl_lane_lowv2sf): Ditto.
16151 (aarch64_fmlal_laneq_lowv2sf): Ditto.
16152 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
16153 (aarch64_fmlalq_lane_lowv4sf): Ditto.
16154 (aarch64_fmlsl_lane_lowv4sf): Ditto.
16155 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
16156 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
16157 (aarch64_fmlal_lane_highv2sf): Ditto.
16158 (aarch64_fmlsl_lane_highv2sf): Ditto.
16159 (aarch64_fmlal_laneq_highv2sf): Ditto.
16160 (aarch64_fmlsl_laneq_highv2sf): Ditto.
16161 (aarch64_fmlalq_lane_highv4sf): Ditto.
16162 (aarch64_fmlsl_lane_highv4sf): Ditto.
16163 (aarch64_fmlalq_laneq_highv4sf): Ditto.
16164 (aarch64_fmlsl_laneq_highv4sf): Ditto.
16165 * config/aarch64/aarch64-simd.md:
16166 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
16167 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
16168 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
16169 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
16170 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
16171 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
16172 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
16173 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
16174 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
16175 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
16176 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
16177 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
16178 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
16179 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
16180 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
16181 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
16182 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
16183 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
16184 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
16185 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
16186 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
16187 (vfmlsl_low_u32): Ditto.
16188 (vfmlalq_low_u32): Ditto.
16189 (vfmlslq_low_u32): Ditto.
16190 (vfmlal_high_u32): Ditto.
16191 (vfmlsl_high_u32): Ditto.
16192 (vfmlalq_high_u32): Ditto.
16193 (vfmlslq_high_u32): Ditto.
16194 (vfmlal_lane_low_u32): Ditto.
16195 (vfmlsl_lane_low_u32): Ditto.
16196 (vfmlal_laneq_low_u32): Ditto.
16197 (vfmlsl_laneq_low_u32): Ditto.
16198 (vfmlalq_lane_low_u32): Ditto.
16199 (vfmlslq_lane_low_u32): Ditto.
16200 (vfmlalq_laneq_low_u32): Ditto.
16201 (vfmlslq_laneq_low_u32): Ditto.
16202 (vfmlal_lane_high_u32): Ditto.
16203 (vfmlsl_lane_high_u32): Ditto.
16204 (vfmlal_laneq_high_u32): Ditto.
16205 (vfmlsl_laneq_high_u32): Ditto.
16206 (vfmlalq_lane_high_u32): Ditto.
16207 (vfmlslq_lane_high_u32): Ditto.
16208 (vfmlalq_laneq_high_u32): Ditto.
16209 (vfmlslq_laneq_high_u32): Ditto.
16210 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
16211 (AARCH64_FL_FOR_ARCH8_4): New.
16212 (AARCH64_ISA_F16FML): New ISA flag.
16213 (TARGET_F16FML): New feature flag for fp16fml.
16214 (doc/invoke.texi): Document new fp16fml option.
16216 2018-01-10 Michael Collison <michael.collison@arm.com>
16218 * config/aarch64/aarch64-builtins.c:
16219 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
16220 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
16221 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
16222 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
16223 (AARCH64_ISA_SHA3): New ISA flag.
16224 (TARGET_SHA3): New feature flag for sha3.
16225 * config/aarch64/iterators.md (sha512_op): New int attribute.
16226 (CRYPTO_SHA512): New int iterator.
16227 (UNSPEC_SHA512H): New unspec.
16228 (UNSPEC_SHA512H2): Ditto.
16229 (UNSPEC_SHA512SU0): Ditto.
16230 (UNSPEC_SHA512SU1): Ditto.
16231 * config/aarch64/aarch64-simd-builtins.def
16232 (aarch64_crypto_sha512hqv2di): New builtin.
16233 (aarch64_crypto_sha512h2qv2di): Ditto.
16234 (aarch64_crypto_sha512su0qv2di): Ditto.
16235 (aarch64_crypto_sha512su1qv2di): Ditto.
16236 (aarch64_eor3qv8hi): Ditto.
16237 (aarch64_rax1qv2di): Ditto.
16238 (aarch64_xarqv2di): Ditto.
16239 (aarch64_bcaxqv8hi): Ditto.
16240 * config/aarch64/aarch64-simd.md:
16241 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
16242 (aarch64_crypto_sha512su0qv2di): Ditto.
16243 (aarch64_crypto_sha512su1qv2di): Ditto.
16244 (aarch64_eor3qv8hi): Ditto.
16245 (aarch64_rax1qv2di): Ditto.
16246 (aarch64_xarqv2di): Ditto.
16247 (aarch64_bcaxqv8hi): Ditto.
16248 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
16249 (vsha512h2q_u64): Ditto.
16250 (vsha512su0q_u64): Ditto.
16251 (vsha512su1q_u64): Ditto.
16252 (veor3q_u16): Ditto.
16253 (vrax1q_u64): Ditto.
16254 (vxarq_u64): Ditto.
16255 (vbcaxq_u16): Ditto.
16256 * config/arm/types.md (crypto_sha512): New type attribute.
16257 (crypto_sha3): Ditto.
16258 (doc/invoke.texi): Document new sha3 option.
16260 2018-01-10 Michael Collison <michael.collison@arm.com>
16262 * config/aarch64/aarch64-builtins.c:
16263 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
16264 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
16265 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
16266 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
16267 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
16268 (AARCH64_ISA_SM4): New ISA flag.
16269 (TARGET_SM4): New feature flag for sm4.
16270 * config/aarch64/aarch64-simd-builtins.def
16271 (aarch64_sm3ss1qv4si): Ditto.
16272 (aarch64_sm3tt1aq4si): Ditto.
16273 (aarch64_sm3tt1bq4si): Ditto.
16274 (aarch64_sm3tt2aq4si): Ditto.
16275 (aarch64_sm3tt2bq4si): Ditto.
16276 (aarch64_sm3partw1qv4si): Ditto.
16277 (aarch64_sm3partw2qv4si): Ditto.
16278 (aarch64_sm4eqv4si): Ditto.
16279 (aarch64_sm4ekeyqv4si): Ditto.
16280 * config/aarch64/aarch64-simd.md:
16281 (aarch64_sm3ss1qv4si): Ditto.
16282 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
16283 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
16284 (aarch64_sm4eqv4si): Ditto.
16285 (aarch64_sm4ekeyqv4si): Ditto.
16286 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
16287 (sm3part_op): Ditto.
16288 (CRYPTO_SM3TT): Ditto.
16289 (CRYPTO_SM3PART): Ditto.
16290 (UNSPEC_SM3SS1): New unspec.
16291 (UNSPEC_SM3TT1A): Ditto.
16292 (UNSPEC_SM3TT1B): Ditto.
16293 (UNSPEC_SM3TT2A): Ditto.
16294 (UNSPEC_SM3TT2B): Ditto.
16295 (UNSPEC_SM3PARTW1): Ditto.
16296 (UNSPEC_SM3PARTW2): Ditto.
16297 (UNSPEC_SM4E): Ditto.
16298 (UNSPEC_SM4EKEY): Ditto.
16299 * config/aarch64/constraints.md (Ui2): New constraint.
16300 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
16301 * config/arm/types.md (crypto_sm3): New type attribute.
16302 (crypto_sm4): Ditto.
16303 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
16304 (vsm3tt1aq_u32): Ditto.
16305 (vsm3tt1bq_u32): Ditto.
16306 (vsm3tt2aq_u32): Ditto.
16307 (vsm3tt2bq_u32): Ditto.
16308 (vsm3partw1q_u32): Ditto.
16309 (vsm3partw2q_u32): Ditto.
16310 (vsm4eq_u32): Ditto.
16311 (vsm4ekeyq_u32): Ditto.
16312 (doc/invoke.texi): Document new sm4 option.
16314 2018-01-10 Michael Collison <michael.collison@arm.com>
16316 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
16317 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
16318 (AARCH64_FL_FOR_ARCH8_4): New.
16319 (AARCH64_FL_V8_4): New flag.
16320 (doc/invoke.texi): Document new armv8.4-a option.
16322 2018-01-10 Michael Collison <michael.collison@arm.com>
16324 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
16325 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
16326 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
16327 * config/aarch64/aarch64-option-extension.def: Add
16328 AARCH64_OPT_EXTENSION of 'sha2'.
16329 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
16330 (crypto): Disable sha2 and aes if crypto disabled.
16331 (crypto): Enable aes and sha2 if enabled.
16332 (simd): Disable sha2 and aes if simd disabled.
16333 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
16335 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
16336 (TARGET_SHA2): New feature flag for sha2.
16337 (TARGET_AES): New feature flag for aes.
16338 * config/aarch64/aarch64-simd.md:
16339 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
16340 conditional on TARGET_AES.
16341 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
16342 (aarch64_crypto_sha1hsi): Make pattern conditional
16344 (aarch64_crypto_sha1hv4si): Ditto.
16345 (aarch64_be_crypto_sha1hv4si): Ditto.
16346 (aarch64_crypto_sha1su1v4si): Ditto.
16347 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
16348 (aarch64_crypto_sha1su0v4si): Ditto.
16349 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
16350 (aarch64_crypto_sha256su0v4si): Ditto.
16351 (aarch64_crypto_sha256su1v4si): Ditto.
16352 (doc/invoke.texi): Document new aes and sha2 options.
16354 2018-01-10 Martin Sebor <msebor@redhat.com>
16356 PR tree-optimization/83781
16357 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
16360 2018-01-11 Martin Sebor <msebor@gmail.com>
16361 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16363 PR tree-optimization/83501
16364 PR tree-optimization/81703
16366 * tree-ssa-strlen.c (get_string_cst): Rename...
16367 (get_string_len): ...to this. Handle global constants.
16368 (handle_char_store): Adjust.
16370 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
16371 Jim Wilson <jimw@sifive.com>
16373 * config/riscv/riscv-protos.h (riscv_output_return): New.
16374 * config/riscv/riscv.c (struct machine_function): New naked_p field.
16375 (riscv_attribute_table, riscv_output_return),
16376 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
16377 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
16378 (riscv_compute_frame_info): Only compute frame->mask if not a naked
16380 (riscv_expand_prologue): Add early return for naked function.
16381 (riscv_expand_epilogue): Likewise.
16382 (riscv_function_ok_for_sibcall): Return false for naked function.
16383 (riscv_set_current_function): New.
16384 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
16385 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
16386 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
16387 * doc/extend.texi (RISC-V Function Attributes): New.
16389 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
16391 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
16392 check for 128-bit long double before checking TCmode.
16393 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
16394 128-bit long doubles before checking TFmode or TCmode.
16395 (FLOAT128_IBM_P): Likewise.
16397 2018-01-10 Martin Sebor <msebor@redhat.com>
16399 PR tree-optimization/83671
16400 * builtins.c (c_strlen): Unconditionally return zero for the empty
16402 Use -Warray-bounds for warnings.
16403 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
16404 for non-constant array indices with COMPONENT_REF, arrays of
16405 arrays, and pointers to arrays.
16406 (gimple_fold_builtin_strlen): Determine and set length range for
16407 non-constant character arrays.
16409 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
16411 PR middle-end/81897
16412 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
16415 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
16417 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
16419 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
16422 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
16423 VECTOR_MEM_ALTIVEC_OR_VSX_P.
16424 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
16425 indexed_or_indirect_operand predicate.
16426 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
16427 (*vsx_le_perm_load_v8hi): Likewise.
16428 (*vsx_le_perm_load_v16qi): Likewise.
16429 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
16430 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
16431 (*vsx_le_perm_store_v8hi): Likewise.
16432 (*vsx_le_perm_store_v16qi): Likewise.
16433 (eight unnamed splitters): Likewise.
16435 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
16437 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
16438 * config/rs6000/emmintrin.h: Likewise.
16439 * config/rs6000/mmintrin.h: Likewise.
16440 * config/rs6000/xmmintrin.h: Likewise.
16442 2018-01-10 David Malcolm <dmalcolm@redhat.com>
16445 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
16447 * tree.c (tree_nop_conversion): Return true for location wrapper
16449 (maybe_wrap_with_location): New function.
16450 (selftest::check_strip_nops): New function.
16451 (selftest::test_location_wrappers): New function.
16452 (selftest::tree_c_tests): Call it.
16453 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
16454 (maybe_wrap_with_location): New decl.
16455 (EXPR_LOCATION_WRAPPER_P): New macro.
16456 (location_wrapper_p): New inline function.
16457 (tree_strip_any_location_wrapper): New inline function.
16459 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
16462 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
16463 stack_realign_offset for the largest alignment of stack slot
16465 (ix86_find_max_used_stack_alignment): New function.
16466 (ix86_finalize_stack_frame_flags): Use it. Set
16467 max_used_stack_alignment if we don't realign stack.
16468 * config/i386/i386.h (machine_function): Add
16469 max_used_stack_alignment.
16471 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
16473 * config/arm/arm.opt (-mbranch-cost): New option.
16474 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
16477 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
16480 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
16481 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
16483 2018-01-10 Richard Biener <rguenther@suse.de>
16486 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
16487 early out so it also covers the case where we have a non-NULL
16490 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
16492 PR tree-optimization/83753
16493 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
16494 for non-strided grouped accesses if the number of elements is 1.
16496 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
16499 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
16500 * i386.h (TARGET_USE_GATHER): Define.
16501 * x86-tune.def (X86_TUNE_USE_GATHER): New.
16503 2018-01-10 Martin Liska <mliska@suse.cz>
16506 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
16507 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
16509 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
16510 CLEANUP_NO_PARTITIONING is not set.
16512 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
16514 * doc/rtl.texi: Remove documentation of (const ...) wrappers
16515 for vectors, as a partial revert of r254296.
16516 * rtl.h (const_vec_p): Delete.
16517 (const_vec_duplicate_p): Don't test for vector CONSTs.
16518 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
16519 * expmed.c (make_tree): Likewise.
16522 * common.md (E, F): Use CONSTANT_P instead of checking for
16524 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
16525 checking for CONST_VECTOR.
16527 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
16529 PR middle-end/83575
16530 * predict.c (force_edge_cold): Handle in more sane way edges
16531 with no prediction.
16533 2018-01-09 Carl Love <cel@us.ibm.com>
16535 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
16537 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
16538 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
16539 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
16540 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
16541 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
16542 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
16543 * config/rs6000/rs6000-protos.h: Add extern defition for
16544 rs6000_generate_float2_double_code.
16545 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
16547 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
16548 (float2_v2df): Add define_expand.
16550 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
16553 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
16554 op_mode in the force_to_mode call.
16556 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
16558 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
16559 instead of checking each element individually.
16560 (aarch64_evpc_uzp): Likewise.
16561 (aarch64_evpc_zip): Likewise.
16562 (aarch64_evpc_ext): Likewise.
16563 (aarch64_evpc_rev): Likewise.
16564 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
16565 instead of checking each element individually. Return true without
16567 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
16568 whether all selected elements come from the same input, instead of
16569 checking each element individually. Remove calls to gen_rtx_REG,
16570 start_sequence and end_sequence and instead assert that no rtl is
16573 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
16575 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
16576 order of HIGH and CONST checks.
16578 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
16580 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
16581 if the destination isn't an SSA_NAME.
16583 2018-01-09 Richard Biener <rguenther@suse.de>
16585 PR tree-optimization/83668
16586 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
16588 (canonicalize_loop_form): ... here, renamed from ...
16589 (canonicalize_loop_closed_ssa_form): ... this and amended to
16590 swap successor edges for loop exit blocks to make us use
16591 the RPO order we need for initial schedule generation.
16593 2018-01-09 Joseph Myers <joseph@codesourcery.com>
16595 PR tree-optimization/64811
16596 * match.pd: When optimizing comparisons with Inf, avoid
16597 introducing or losing exceptions from comparisons with NaN.
16599 2018-01-09 Martin Liska <mliska@suse.cz>
16602 * asan.c (shadow_mem_size): Add gcc_assert.
16604 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
16606 Don't save registers in main().
16609 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
16610 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
16611 * config/avr/avr.c (avr_set_current_function): Don't error if
16612 naked, OS_task or OS_main are specified at the same time.
16613 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
16615 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
16617 * common/config/avr/avr-common.c (avr_option_optimization_table):
16618 Switch on -mmain-is-OS_task for optimizing compilations.
16620 2018-01-09 Richard Biener <rguenther@suse.de>
16622 PR tree-optimization/83572
16623 * graphite.c: Include cfganal.h.
16624 (graphite_transform_loops): Connect infinite loops to exit
16625 and remove fake edges at the end.
16627 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
16629 * ipa-inline.c (edge_badness): Revert accidental checkin.
16631 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
16634 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
16635 symbols; not inline clones.
16637 2018-01-09 Jakub Jelinek <jakub@redhat.com>
16640 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
16641 hard registers. Formatting fixes.
16643 PR preprocessor/83722
16644 * gcc.c (try_generate_repro): Pass
16645 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
16646 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
16649 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
16650 Kito Cheng <kito.cheng@gmail.com>
16652 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
16653 (riscv_leaf_function_p): Delete.
16654 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
16656 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16658 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
16660 (do_ifelse): New function.
16661 (do_isel): New function.
16662 (do_sub3): New function.
16663 (do_add3): New function.
16664 (do_load_mask_compare): New function.
16665 (do_overlap_load_compare): New function.
16666 (expand_compare_loop): New function.
16667 (expand_block_compare): Call expand_compare_loop() when appropriate.
16668 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
16669 option description.
16670 (-mblock-compare-inline-loop-limit): New option.
16672 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16675 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
16676 Reverse order of second and third operands in first alternative.
16677 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
16678 of first and second elements in UNSPEC_VPERMR vector.
16679 (altivec_expand_vec_perm_le): Likewise.
16681 2018-01-08 Jeff Law <law@redhat.com>
16683 PR rtl-optimizatin/81308
16684 * tree-switch-conversion.c (cfg_altered): New file scoped static.
16685 (process_switch): If group_case_labels makes a change, then set
16687 (pass_convert_switch::execute): If a switch is converted, then
16688 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
16690 PR rtl-optimization/81308
16691 * recog.c (split_all_insns): Conditionally cleanup the CFG after
16694 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
16696 PR target/83663 - Revert r255946
16697 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
16698 generation for cases where splatting a value is not useful.
16699 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
16700 across a vec_duplicate and a paradoxical subreg forming a vector
16701 mode to a vec_concat.
16703 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16705 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
16706 -march=armv8.3-a variants.
16707 * config/arm/t-multilib: Likewise.
16708 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
16710 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16712 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
16714 (cceq_ior_compare_complement): Give it a name so I can use it, and
16715 change boolean_or_operator predicate to boolean_operator so it can
16716 be used to generate a crand.
16717 (eqne): New code iterator.
16718 (bd/bd_neg): New code_attrs.
16719 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
16720 a single define_insn.
16721 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
16722 decrement (bdnzt/bdnzf/bdzt/bdzf).
16723 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
16724 with the new names of the branch decrement patterns, and added the
16725 names of the branch decrement conditional patterns.
16727 2018-01-08 Richard Biener <rguenther@suse.de>
16729 PR tree-optimization/83563
16730 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
16733 2018-01-08 Richard Biener <rguenther@suse.de>
16735 PR middle-end/83713
16736 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
16738 2018-01-08 Richard Biener <rguenther@suse.de>
16740 PR tree-optimization/83685
16741 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
16742 references to abnormals.
16744 2018-01-08 Richard Biener <rguenther@suse.de>
16747 * dwarf2out.c (output_indirect_strings): Handle empty
16748 skeleton_debug_str_hash.
16749 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
16751 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
16753 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
16754 (emit_store_direct): Likewise.
16755 (arc_trampoline_adjust_address): Likewise.
16756 (arc_asm_trampoline_template): New function.
16757 (arc_initialize_trampoline): Use asm_trampoline_template.
16758 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
16759 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
16760 * config/arc/arc.md (flush_icache): Delete pattern.
16762 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
16764 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
16765 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
16768 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
16771 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
16772 by not USED_FOR_TARGET.
16773 (make_pass_resolve_sw_modes): Likewise.
16775 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
16777 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
16780 2018-01-08 Richard Biener <rguenther@suse.de>
16782 PR middle-end/83580
16783 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
16785 2018-01-08 Richard Biener <rguenther@suse.de>
16787 PR middle-end/83517
16788 * match.pd ((t * 2) / 2) -> t): Add missing :c.
16790 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
16792 PR middle-end/81897
16793 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
16794 basic blocks with a small number of successors.
16795 (convert_control_dep_chain_into_preds): Improve handling of
16797 (dump_predicates): Split apart into...
16798 (dump_pred_chain): ...here...
16799 (dump_pred_info): ...and here.
16800 (can_one_predicate_be_invalidated_p): Add debugging printfs.
16801 (can_chain_union_be_invalidated_p): Improve check for invalidation
16803 (uninit_uses_cannot_happen): Avoid unnecessary if
16804 convert_control_dep_chain_into_preds yielded nothing.
16806 2018-01-06 Martin Sebor <msebor@redhat.com>
16808 PR tree-optimization/83640
16809 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
16810 subtracting negative offset from size.
16811 (builtin_access::overlap): Adjust offset bounds of the access to fall
16812 within the size of the object if possible.
16814 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
16816 PR rtl-optimization/83699
16817 * expmed.c (extract_bit_field_1): Restrict the vector usage of
16818 extract_bit_field_as_subreg to cases in which the extracted
16819 value is also a vector.
16821 * lra-constraints.c (process_alt_operands): Test for the equivalence
16822 substitutions when detecting a possible reload cycle.
16824 2018-01-06 Jakub Jelinek <jakub@redhat.com>
16827 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
16828 by default if flag_selective_schedling{,2}. Formatting fixes.
16830 PR rtl-optimization/83682
16831 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
16832 if it has non-VECTOR_MODE element mode.
16833 (vec_duplicate_p): Likewise.
16835 PR middle-end/83694
16836 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
16837 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
16839 2018-01-05 Jakub Jelinek <jakub@redhat.com>
16842 * config/i386/i386-builtin.def
16843 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
16844 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
16845 Require also OPTION_MASK_ISA_AVX512F in addition to
16846 OPTION_MASK_ISA_GFNI.
16847 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
16848 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
16849 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
16850 to OPTION_MASK_ISA_GFNI.
16851 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
16852 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
16853 OPTION_MASK_ISA_AVX512BW.
16854 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
16855 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
16856 addition to OPTION_MASK_ISA_GFNI.
16857 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
16858 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
16859 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
16860 to OPTION_MASK_ISA_GFNI.
16861 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
16862 a requirement for all ISAs rather than any of them with a few
16864 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
16866 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
16867 bitmasks to be enabled with 3 exceptions, instead of requiring any
16868 enabled ISA with lots of exceptions.
16869 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
16870 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
16871 Change avx512bw in isa attribute to avx512f.
16872 * config/i386/sgxintrin.h: Add license boilerplate.
16873 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
16874 to __AVX512F__ and __AVX512VL to __AVX512VL__.
16875 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
16876 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
16878 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
16879 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
16880 temporarily sse2 rather than sse if not enabled already.
16883 * config/i386/sse.md (VI248_VLBW): Rename to ...
16884 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
16885 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
16886 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
16887 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
16888 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
16889 mode iterator instead of VI248_VLBW.
16891 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
16893 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
16894 (record_modified): Skip clobbers; add debug output.
16895 (param_change_prob): Use sreal frequencies.
16897 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
16899 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
16900 punt for user-aligned variables.
16902 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
16904 * tree-chrec.c (chrec_contains_symbols): Return true for
16907 2018-01-05 Sudakshina Das <sudi.das@arm.com>
16910 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
16911 of (x|y) == x for BICS pattern.
16913 2018-01-05 Jakub Jelinek <jakub@redhat.com>
16915 PR tree-optimization/83605
16916 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
16917 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
16920 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
16922 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
16923 * config/epiphany/rtems.h: New file.
16925 2018-01-04 Jakub Jelinek <jakub@redhat.com>
16926 Uros Bizjak <ubizjak@gmail.com>
16929 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
16930 QIreg_operand instead of register_operand predicate.
16931 * config/i386/i386.c (ix86_rop_should_change_byte_p,
16932 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
16933 comments instead of -fmitigate[-_]rop.
16935 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16938 * cgraphunit.c (symbol_table::compile): Switch to text_section
16939 before calling assembly_start debug hook.
16940 * run-rtl-passes.c (run_rtl_passes): Likewise.
16943 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
16945 * tree-vrp.c (extract_range_from_binary_expr_1): Check
16946 range_int_cst_p rather than !symbolic_range_p before calling
16947 extract_range_from_multiplicative_op_1.
16949 2018-01-04 Jeff Law <law@redhat.com>
16951 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
16952 redundant test in assertion.
16954 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
16956 * doc/rtl.texi: Document machine_mode wrapper classes.
16958 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
16960 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
16961 using tree_to_uhwi.
16963 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
16965 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
16966 the VEC_PERM_EXPR fold to fail.
16968 2018-01-04 Jakub Jelinek <jakub@redhat.com>
16971 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
16972 to switched_sections.
16974 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
16977 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
16978 test for d.testing.
16980 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
16983 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
16984 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
16986 2018-01-04 Jakub Jelinek <jakub@redhat.com>
16989 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
16990 is BLKmode and bitpos not zero or mode change is needed.
16992 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
16995 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
16998 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
17001 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
17002 instead of MULT rtx. Update all corresponding splitters.
17003 (*saddl_se): Ditto.
17004 (*ssub<modesuffix>): Ditto.
17005 (*ssubl_se): Ditto.
17006 (*cmp_sadd_di): Update split patterns.
17007 (*cmp_sadd_si): Ditto.
17008 (*cmp_sadd_sidi): Ditto.
17009 (*cmp_ssub_di): Ditto.
17010 (*cmp_ssub_si): Ditto.
17011 (*cmp_ssub_sidi): Ditto.
17012 * config/alpha/predicates.md (const23_operand): New predicate.
17013 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
17014 Look for ASHIFT, not MULT inner operand.
17015 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
17017 2018-01-04 Martin Liska <mliska@suse.cz>
17019 PR gcov-profile/83669
17020 * gcov.c (output_intermediate_file): Add version to intermediate
17022 * doc/gcov.texi: Document new field 'version' in intermediate
17023 file format. Fix location of '-k' option of gcov command.
17025 2018-01-04 Martin Liska <mliska@suse.cz>
17028 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
17030 2018-01-04 Jakub Jelinek <jakub@redhat.com>
17032 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
17034 2018-01-03 Martin Sebor <msebor@redhat.com>
17036 PR tree-optimization/83655
17037 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
17038 checking calls with invalid arguments.
17040 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17042 * tree-vect-stmts.c (vect_get_store_rhs): New function.
17043 (vectorizable_mask_load_store): Delete.
17044 (vectorizable_call): Return false for masked loads and stores.
17045 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
17046 instead of gimple_assign_rhs1.
17047 (vectorizable_load): Handle IFN_MASK_LOAD.
17048 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
17050 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17052 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
17054 (vectorizable_mask_load_store): ...here.
17055 (vectorizable_load): ...and here.
17057 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17059 * tree-vect-stmts.c (vect_build_all_ones_mask)
17060 (vect_build_zero_merge_argument): New functions, split out from...
17061 (vectorizable_load): ...here.
17063 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17065 * tree-vect-stmts.c (vect_check_store_rhs): New function,
17067 (vectorizable_mask_load_store): ...here.
17068 (vectorizable_store): ...and here.
17070 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17072 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
17074 (vectorizable_mask_load_store): ...here.
17076 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17078 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
17079 (vect_model_store_cost): Take a vec_load_store_type instead of a
17081 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
17082 (vect_model_store_cost): Take a vec_load_store_type instead of a
17084 (vectorizable_mask_load_store): Update accordingly.
17085 (vectorizable_store): Likewise.
17086 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
17088 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17090 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
17091 IFN_MASK_LOAD calls here rather than...
17092 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
17094 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17095 Alan Hayward <alan.hayward@arm.com>
17096 David Sherwood <david.sherwood@arm.com>
17098 * expmed.c (extract_bit_field_1): For vector extracts,
17099 fall back to extract_bit_field_as_subreg if vec_extract
17102 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17103 Alan Hayward <alan.hayward@arm.com>
17104 David Sherwood <david.sherwood@arm.com>
17106 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
17107 they are variable or constant sized.
17108 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
17109 slots for constant-sized data.
17111 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17112 Alan Hayward <alan.hayward@arm.com>
17113 David Sherwood <david.sherwood@arm.com>
17115 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
17116 handling COND_EXPRs with boolean comparisons, try to find a better
17117 basis for the mask type than the boolean itself.
17119 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17121 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
17122 is calculated and how it can be overridden.
17123 * genmodes.c (max_bitsize_mode_any_mode): New variable.
17124 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
17126 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
17129 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17130 Alan Hayward <alan.hayward@arm.com>
17131 David Sherwood <david.sherwood@arm.com>
17133 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
17134 Remove the mode argument.
17135 (aarch64_simd_valid_immediate): Remove the mode and inverse
17137 * config/aarch64/iterators.md (bitsize): New iterator.
17138 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
17139 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
17140 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
17141 aarch64_simd_valid_immediate.
17142 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
17143 (aarch64_reg_or_bic_imm): Likewise.
17144 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
17145 with an insn_type enum and msl with a modifier_type enum.
17146 Replace element_width with a scalar_mode. Change the shift
17147 to unsigned int. Add constructors for scalar_float_mode and
17148 scalar_int_mode elements.
17149 (aarch64_vect_float_const_representable_p): Delete.
17150 (aarch64_can_const_movi_rtx_p)
17151 (aarch64_simd_scalar_immediate_valid_for_move)
17152 (aarch64_simd_make_constant): Update call to
17153 aarch64_simd_valid_immediate.
17154 (aarch64_advsimd_valid_immediate_hs): New function.
17155 (aarch64_advsimd_valid_immediate): Likewise.
17156 (aarch64_simd_valid_immediate): Remove mode and inverse
17157 arguments. Rewrite to use the above. Use const_vec_duplicate_p
17158 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
17159 and aarch64_float_const_representable_p on the result.
17160 (aarch64_output_simd_mov_immediate): Remove mode argument.
17161 Update call to aarch64_simd_valid_immediate and use of
17162 simd_immediate_info.
17163 (aarch64_output_scalar_simd_mov_immediate): Update call
17166 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17167 Alan Hayward <alan.hayward@arm.com>
17168 David Sherwood <david.sherwood@arm.com>
17170 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
17171 (mode_nunits): Likewise CONST_MODE_NUNITS.
17172 * machmode.def (ADJUST_NUNITS): Document.
17173 * genmodes.c (mode_data::need_nunits_adj): New field.
17174 (blank_mode): Update accordingly.
17175 (adj_nunits): New variable.
17176 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
17178 (emit_mode_size_inline): Set need_bytesize_adj for all modes
17179 listed in adj_nunits.
17180 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
17181 listed in adj_nunits. Don't emit case statements for such modes.
17182 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
17183 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
17184 nothing if adj_nunits is nonnull.
17185 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
17186 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
17187 (emit_mode_fbit): Update use of print_maybe_const_decl.
17188 (emit_move_size): Likewise. Treat the array as non-const
17190 (emit_mode_adjustments): Handle adj_nunits.
17192 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17194 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
17195 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
17196 (VECTOR_MODES): Use it.
17197 (make_vector_modes): Take the prefix as an argument.
17199 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17200 Alan Hayward <alan.hayward@arm.com>
17201 David Sherwood <david.sherwood@arm.com>
17203 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
17204 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
17205 for MODE_VECTOR_BOOL.
17206 * machmode.def (VECTOR_BOOL_MODE): Document.
17207 * genmodes.c (VECTOR_BOOL_MODE): New macro.
17208 (make_vector_bool_mode): New function.
17209 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
17211 * lto-streamer-in.c (lto_input_mode_table): Likewise.
17212 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
17214 * stor-layout.c (int_mode_for_mode): Likewise.
17215 * tree.c (build_vector_type_for_mode): Likewise.
17216 * varasm.c (output_constant_pool_2): Likewise.
17217 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
17218 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
17219 for MODE_VECTOR_BOOL.
17220 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
17221 of mode class checks.
17222 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
17223 instead of a list of mode class checks.
17224 (expand_vector_scalar_condition): Likewise.
17225 (type_for_widest_vector_mode): Handle BImode as an inner mode.
17227 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17228 Alan Hayward <alan.hayward@arm.com>
17229 David Sherwood <david.sherwood@arm.com>
17231 * machmode.h (mode_size): Change from unsigned short to
17233 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
17234 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
17235 or if measurement_type is not polynomial.
17236 (fixed_size_mode::includes_p): Check for constant-sized modes.
17237 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
17238 return a poly_uint16 rather than an unsigned short.
17239 (emit_mode_size): Change the type of mode_size from unsigned short
17240 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
17241 (emit_mode_adjustments): Cope with polynomial vector sizes.
17242 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
17244 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
17246 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
17247 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
17248 * caller-save.c (setup_save_areas): Likewise.
17249 (replace_reg_with_saved_mem): Likewise.
17250 * calls.c (emit_library_call_value_1): Likewise.
17251 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
17252 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
17253 (gen_lowpart_for_combine): Likewise.
17254 * convert.c (convert_to_integer_1): Likewise.
17255 * cse.c (equiv_constant, cse_insn): Likewise.
17256 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
17257 (cselib_subst_to_values): Likewise.
17258 * dce.c (word_dce_process_block): Likewise.
17259 * df-problems.c (df_word_lr_mark_ref): Likewise.
17260 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
17261 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
17262 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
17263 (rtl_for_decl_location): Likewise.
17264 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
17265 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
17266 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
17267 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
17268 (expand_expr_real_1): Likewise.
17269 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
17270 (pad_below): Likewise.
17271 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
17272 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
17273 * ira.c (get_subreg_tracking_sizes): Likewise.
17274 * ira-build.c (ira_create_allocno_objects): Likewise.
17275 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
17276 (ira_sort_regnos_for_alter_reg): Likewise.
17277 * ira-costs.c (record_operand_costs): Likewise.
17278 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
17279 (resolve_simple_move): Likewise.
17280 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
17281 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
17282 (lra_constraints): Likewise.
17283 (CONST_POOL_OK_P): Reject variable-sized modes.
17284 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
17285 (add_pseudo_to_slot, lra_spill): Likewise.
17286 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17287 * optabs-query.c (get_best_extraction_insn): Likewise.
17288 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
17289 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
17290 (expand_mult_highpart, valid_multiword_target_p): Likewise.
17291 * recog.c (offsettable_address_addr_space_p): Likewise.
17292 * regcprop.c (maybe_mode_change): Likewise.
17293 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
17294 * regrename.c (build_def_use): Likewise.
17295 * regstat.c (dump_reg_info): Likewise.
17296 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
17297 (find_reloads, find_reloads_subreg_address): Likewise.
17298 * reload1.c (eliminate_regs_1): Likewise.
17299 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
17300 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
17301 (simplify_binary_operation_1, simplify_subreg): Likewise.
17302 * targhooks.c (default_function_arg_padding): Likewise.
17303 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
17304 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
17305 (verify_gimple_assign_ternary): Likewise.
17306 * tree-inline.c (estimate_move_cost): Likewise.
17307 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
17308 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
17309 (get_address_cost_ainc): Likewise.
17310 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
17311 (vect_supportable_dr_alignment): Likewise.
17312 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
17313 (vectorizable_reduction): Likewise.
17314 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
17315 (vectorizable_operation, vectorizable_load): Likewise.
17316 * tree.c (build_same_sized_truth_vector_type): Likewise.
17317 * valtrack.c (cleanup_auto_inc_dec): Likewise.
17318 * var-tracking.c (emit_note_insn_var_location): Likewise.
17319 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
17320 (ADDR_VEC_ALIGN): Likewise.
17322 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17323 Alan Hayward <alan.hayward@arm.com>
17324 David Sherwood <david.sherwood@arm.com>
17326 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
17328 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
17329 or if measurement_type is polynomial.
17330 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
17331 * combine.c (make_extraction): Likewise.
17332 * dse.c (find_shift_sequence): Likewise.
17333 * dwarf2out.c (mem_loc_descriptor): Likewise.
17334 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
17335 (extract_bit_field, extract_low_bits): Likewise.
17336 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
17337 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
17338 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
17339 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
17340 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
17341 * reload.c (find_reloads): Likewise.
17342 * reload1.c (alter_reg): Likewise.
17343 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
17344 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
17345 * tree-if-conv.c (predicate_mem_writes): Likewise.
17346 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
17347 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
17348 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
17349 * valtrack.c (dead_debug_insert_temp): Likewise.
17350 * varasm.c (mergeable_constant_section): Likewise.
17351 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
17353 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17354 Alan Hayward <alan.hayward@arm.com>
17355 David Sherwood <david.sherwood@arm.com>
17357 * expr.c (expand_assignment): Cope with polynomial mode sizes
17358 when assigning to a CONCAT.
17360 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17361 Alan Hayward <alan.hayward@arm.com>
17362 David Sherwood <david.sherwood@arm.com>
17364 * machmode.h (mode_precision): Change from unsigned short to
17366 (mode_to_precision): Return a poly_uint16 rather than an unsigned
17368 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
17369 or if measurement_type is not polynomial.
17370 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
17371 in which the mode is already known to be a scalar_int_mode.
17372 * genmodes.c (emit_mode_precision): Change the type of mode_precision
17373 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
17375 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
17376 for GET_MODE_PRECISION.
17377 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
17378 for GET_MODE_PRECISION.
17379 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
17381 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
17382 (expand_field_assignment, make_extraction): Likewise.
17383 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
17384 (get_last_value): Likewise.
17385 * convert.c (convert_to_integer_1): Likewise.
17386 * cse.c (cse_insn): Likewise.
17387 * expr.c (expand_expr_real_1): Likewise.
17388 * lra-constraints.c (simplify_operand_subreg): Likewise.
17389 * optabs-query.c (can_atomic_load_p): Likewise.
17390 * optabs.c (expand_atomic_load): Likewise.
17391 (expand_atomic_store): Likewise.
17392 * ree.c (combine_reaching_defs): Likewise.
17393 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
17394 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
17395 * tree.h (type_has_mode_precision_p): Likewise.
17396 * ubsan.c (instrument_si_overflow): Likewise.
17398 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17399 Alan Hayward <alan.hayward@arm.com>
17400 David Sherwood <david.sherwood@arm.com>
17402 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
17403 polynomial numbers of units.
17404 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
17405 (valid_vector_subparts_p): New function.
17406 (build_vector_type): Remove temporary shim and take the number
17407 of units as a poly_uint64 rather than an int.
17408 (build_opaque_vector_type): Take the number of units as a
17409 poly_uint64 rather than an int.
17410 * tree.c (build_vector_from_ctor): Handle polynomial
17411 TYPE_VECTOR_SUBPARTS.
17412 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
17413 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
17414 (build_vector_from_val): If the number of units is variable,
17415 use build_vec_duplicate_cst for constant operands and
17416 VEC_DUPLICATE_EXPR otherwise.
17417 (make_vector_type): Remove temporary is_constant ().
17418 (build_vector_type, build_opaque_vector_type): Take the number of
17419 units as a poly_uint64 rather than an int.
17420 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
17422 * cfgexpand.c (expand_debug_expr): Likewise.
17423 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
17424 (store_constructor, expand_expr_real_1): Likewise.
17425 (const_scalar_mask_from_tree): Likewise.
17426 * fold-const-call.c (fold_const_reduction): Likewise.
17427 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
17428 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
17429 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
17430 (fold_relational_const): Likewise.
17431 (native_interpret_vector): Likewise. Change the size from an
17432 int to an unsigned int.
17433 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
17434 TYPE_VECTOR_SUBPARTS.
17435 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
17436 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
17437 duplicating a non-constant operand into a variable-length vector.
17438 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
17439 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
17440 * ipa-icf.c (sem_variable::equals): Likewise.
17441 * match.pd: Likewise.
17442 * omp-simd-clone.c (simd_clone_subparts): Likewise.
17443 * print-tree.c (print_node): Likewise.
17444 * stor-layout.c (layout_type): Likewise.
17445 * targhooks.c (default_builtin_vectorization_cost): Likewise.
17446 * tree-cfg.c (verify_gimple_comparison): Likewise.
17447 (verify_gimple_assign_binary): Likewise.
17448 (verify_gimple_assign_ternary): Likewise.
17449 (verify_gimple_assign_single): Likewise.
17450 * tree-pretty-print.c (dump_generic_node): Likewise.
17451 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
17452 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
17453 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
17454 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
17455 (vect_shift_permute_load_chain): Likewise.
17456 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
17457 (expand_vector_condition, optimize_vector_constructor): Likewise.
17458 (lower_vec_perm, get_compute_type): Likewise.
17459 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
17460 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
17461 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
17462 (vect_recog_mask_conversion_pattern): Likewise.
17463 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
17464 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
17465 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
17466 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
17467 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
17468 (vectorizable_shift, vectorizable_operation, vectorizable_store)
17469 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
17470 (supportable_widening_operation): Likewise.
17471 (supportable_narrowing_operation): Likewise.
17472 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
17474 * varasm.c (output_constant): Likewise.
17476 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17477 Alan Hayward <alan.hayward@arm.com>
17478 David Sherwood <david.sherwood@arm.com>
17480 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
17481 so that both the length == 3 and length != 3 cases set up their
17482 own permute vectors. Add comments explaining why we know the
17483 number of elements is constant.
17484 (vect_permute_load_chain): Likewise.
17486 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17487 Alan Hayward <alan.hayward@arm.com>
17488 David Sherwood <david.sherwood@arm.com>
17490 * machmode.h (mode_nunits): Change from unsigned char to
17492 (ONLY_FIXED_SIZE_MODES): New macro.
17493 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
17494 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
17495 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
17497 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
17498 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
17499 or if measurement_type is not polynomial.
17500 * genmodes.c (ZERO_COEFFS): New macro.
17501 (emit_mode_nunits_inline): Make mode_nunits_inline return a
17503 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
17504 Use ZERO_COEFFS when emitting initializers.
17505 * data-streamer.h (bp_pack_poly_value): New function.
17506 (bp_unpack_poly_value): Likewise.
17507 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
17508 for GET_MODE_NUNITS.
17509 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
17510 for GET_MODE_NUNITS.
17511 * tree.c (make_vector_type): Remove temporary shim and make
17512 the real function take the number of units as a poly_uint64
17513 rather than an int.
17514 (build_vector_type_for_mode): Handle polynomial nunits.
17515 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
17516 * emit-rtl.c (const_vec_series_p_1): Likewise.
17517 (gen_rtx_CONST_VECTOR): Likewise.
17518 * fold-const.c (test_vec_duplicate_folding): Likewise.
17519 * genrecog.c (validate_pattern): Likewise.
17520 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
17521 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
17522 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
17523 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
17524 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
17525 * rtlanal.c (subreg_get_info): Likewise.
17526 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
17527 (vect_grouped_load_supported): Likewise.
17528 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
17529 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
17530 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17531 (simplify_const_unary_operation, simplify_binary_operation_1)
17532 (simplify_const_binary_operation, simplify_ternary_operation)
17533 (test_vector_ops_duplicate, test_vector_ops): Likewise.
17534 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
17535 instead of CONST_VECTOR_NUNITS.
17536 * varasm.c (output_constant_pool_2): Likewise.
17537 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
17538 explicit-encoded elements in the XVEC for variable-length vectors.
17540 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17542 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
17544 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17545 Alan Hayward <alan.hayward@arm.com>
17546 David Sherwood <david.sherwood@arm.com>
17548 * coretypes.h (fixed_size_mode): Declare.
17549 (fixed_size_mode_pod): New typedef.
17550 * builtins.h (target_builtins::x_apply_args_mode)
17551 (target_builtins::x_apply_result_mode): Change type to
17552 fixed_size_mode_pod.
17553 * builtins.c (apply_args_size, apply_result_size, result_vector)
17554 (expand_builtin_apply_args_1, expand_builtin_apply)
17555 (expand_builtin_return): Update accordingly.
17557 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17559 * cse.c (hash_rtx_cb): Hash only the encoded elements.
17560 * cselib.c (cselib_hash_rtx): Likewise.
17561 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
17562 CONST_VECTOR encoding.
17564 2018-01-03 Jakub Jelinek <jakub@redhat.com>
17565 Jeff Law <law@redhat.com>
17568 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
17569 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
17570 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
17571 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
17574 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
17575 explicitly probe *sp in a noreturn function if there were any callee
17576 register saves or frame pointer is needed.
17578 2018-01-03 Jakub Jelinek <jakub@redhat.com>
17581 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
17582 BLKmode for ternary, binary or unary expressions.
17585 * var-tracking.c (delete_vta_debug_insn): New inline function.
17586 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
17587 insns from get_insns () to NULL instead of each bb separately.
17588 Use delete_vta_debug_insn. No longer static.
17589 (vt_debug_insns_local, variable_tracking_main_1): Adjust
17590 delete_vta_debug_insns callers.
17591 * rtl.h (delete_vta_debug_insns): Declare.
17592 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
17593 instead of variable_tracking_main.
17595 2018-01-03 Martin Sebor <msebor@redhat.com>
17597 PR tree-optimization/83603
17598 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
17599 arguments past the endof the argument list in functions declared
17600 without a prototype.
17601 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
17602 Avoid checking when arguments are null.
17604 2018-01-03 Martin Sebor <msebor@redhat.com>
17607 * doc/extend.texi (attribute const): Fix a typo.
17608 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
17609 issuing -Wsuggest-attribute for void functions.
17611 2018-01-03 Martin Sebor <msebor@redhat.com>
17613 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
17614 offset_int::from instead of wide_int::to_shwi.
17615 (maybe_diag_overlap): Remove assertion.
17616 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
17617 * gimple-ssa-sprintf.c (format_directive): Same.
17618 (parse_directive): Same.
17619 (sprintf_dom_walker::compute_format_length): Same.
17620 (try_substitute_return_value): Same.
17622 2018-01-03 Jeff Law <law@redhat.com>
17624 PR middle-end/83654
17625 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
17626 non-constant residual for zero at runtime and avoid probing in
17627 that case. Reorganize code for trailing problem to mirror handling
17630 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17632 PR tree-optimization/83501
17633 * tree-ssa-strlen.c (get_string_cst): New.
17634 (handle_char_store): Call get_string_cst.
17636 2018-01-03 Martin Liska <mliska@suse.cz>
17638 PR tree-optimization/83593
17639 * tree-ssa-strlen.c: Include tree-cfg.h.
17640 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
17641 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
17642 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
17644 (strlen_dom_walker::before_dom_children): Call
17645 gimple_purge_dead_eh_edges. Dump tranformation with details
17647 (strlen_dom_walker::before_dom_children): Update call by adding
17648 new argument cleanup_eh.
17649 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
17651 2018-01-03 Martin Liska <mliska@suse.cz>
17654 * cif-code.def (VARIADIC_THUNK): New enum value.
17655 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
17658 2018-01-03 Jan Beulich <jbeulich@suse.com>
17660 * sse.md (mov<mode>_internal): Tighten condition for when to use
17661 vmovdqu<ssescalarsize> for TI and OI modes.
17663 2018-01-03 Jakub Jelinek <jakub@redhat.com>
17665 Update copyright years.
17667 2018-01-03 Martin Liska <mliska@suse.cz>
17670 * ipa-visibility.c (function_and_variable_visibility): Skip
17671 functions with noipa attribure.
17673 2018-01-03 Jakub Jelinek <jakub@redhat.com>
17675 * gcc.c (process_command): Update copyright notice dates.
17676 * gcov-dump.c (print_version): Ditto.
17677 * gcov.c (print_version): Ditto.
17678 * gcov-tool.c (print_version): Ditto.
17679 * gengtype.c (create_file): Ditto.
17680 * doc/cpp.texi: Bump @copying's copyright year.
17681 * doc/cppinternals.texi: Ditto.
17682 * doc/gcc.texi: Ditto.
17683 * doc/gccint.texi: Ditto.
17684 * doc/gcov.texi: Ditto.
17685 * doc/install.texi: Ditto.
17686 * doc/invoke.texi: Ditto.
17688 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17690 * vector-builder.h (vector_builder::m_full_nelts): Change from
17691 unsigned int to poly_uint64.
17692 (vector_builder::full_nelts): Update prototype accordingly.
17693 (vector_builder::new_vector): Likewise.
17694 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
17695 (vector_builder::operator ==): Likewise.
17696 (vector_builder::finalize): Likewise.
17697 * int-vector-builder.h (int_vector_builder::int_vector_builder):
17698 Take the number of elements as a poly_uint64 rather than an
17700 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
17701 from unsigned int to poly_uint64.
17702 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
17703 (vec_perm_indices::new_vector): Likewise.
17704 (vec_perm_indices::length): Likewise.
17705 (vec_perm_indices::nelts_per_input): Likewise.
17706 (vec_perm_indices::input_nelts): Likewise.
17707 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
17708 number of elements per input as a poly_uint64 rather than an
17709 unsigned int. Use the original encoding for variable-length
17710 vectors, rather than clamping each individual element.
17711 For the second and subsequent elements in each pattern,
17712 clamp the step and base before clamping their sum.
17713 (vec_perm_indices::series_p): Handle polynomial element counts.
17714 (vec_perm_indices::all_in_range_p): Likewise.
17715 (vec_perm_indices_to_tree): Likewise.
17716 (vec_perm_indices_to_rtx): Likewise.
17717 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
17718 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
17719 (tree_vector_builder::new_binary_operation): Handle polynomial
17720 element counts. Return false if we need to know the number
17721 of elements at compile time.
17722 * fold-const.c (fold_vec_perm): Punt if the number of elements
17723 isn't known at compile time.
17725 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17727 * vec-perm-indices.h (vec_perm_builder): Change element type
17728 from HOST_WIDE_INT to poly_int64.
17729 (vec_perm_indices::element_type): Update accordingly.
17730 (vec_perm_indices::clamp): Handle polynomial element_types.
17731 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
17732 (vec_perm_indices::all_in_range_p): Likewise.
17733 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
17735 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
17736 polynomial vec_perm_indices element types.
17737 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
17738 * fold-const.c (fold_vec_perm): Likewise.
17739 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
17740 * tree-vect-generic.c (lower_vec_perm): Likewise.
17741 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
17742 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
17743 element type to HOST_WIDE_INT.
17745 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17746 Alan Hayward <alan.hayward@arm.com>
17747 David Sherwood <david.sherwood@arm.com>
17749 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
17750 rather than an int. Use plus_constant.
17751 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
17752 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
17754 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17755 Alan Hayward <alan.hayward@arm.com>
17756 David Sherwood <david.sherwood@arm.com>
17758 * calls.c (emit_call_1, expand_call): Change struct_value_size from
17759 a HOST_WIDE_INT to a poly_int64.
17761 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17762 Alan Hayward <alan.hayward@arm.com>
17763 David Sherwood <david.sherwood@arm.com>
17765 * calls.c (load_register_parameters): Cope with polynomial
17766 mode sizes. Require a constant size for BLKmode parameters
17767 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
17768 forces a parameter to be padded at the lsb end in order to
17769 fill a complete number of words, require the parameter size
17770 to be ordered wrt UNITS_PER_WORD.
17772 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17773 Alan Hayward <alan.hayward@arm.com>
17774 David Sherwood <david.sherwood@arm.com>
17776 * reload1.c (spill_stack_slot_width): Change element type
17777 from unsigned int to poly_uint64_pod.
17778 (alter_reg): Treat mode sizes as polynomial.
17780 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17781 Alan Hayward <alan.hayward@arm.com>
17782 David Sherwood <david.sherwood@arm.com>
17784 * reload.c (complex_word_subreg_p): New function.
17785 (reload_inner_reg_of_subreg, push_reload): Use it.
17787 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17788 Alan Hayward <alan.hayward@arm.com>
17789 David Sherwood <david.sherwood@arm.com>
17791 * lra-constraints.c (process_alt_operands): Reject matched
17792 operands whose sizes aren't ordered.
17793 (match_reload): Refer to this check here.
17795 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17796 Alan Hayward <alan.hayward@arm.com>
17797 David Sherwood <david.sherwood@arm.com>
17799 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
17800 that the mode size is in the set {1, 2, 4, 8, 16}.
17802 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17803 Alan Hayward <alan.hayward@arm.com>
17804 David Sherwood <david.sherwood@arm.com>
17806 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
17807 Use plus_constant instead of gen_rtx_PLUS.
17809 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17810 Alan Hayward <alan.hayward@arm.com>
17811 David Sherwood <david.sherwood@arm.com>
17813 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
17814 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
17815 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
17816 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
17817 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
17818 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
17819 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
17820 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
17821 * config/i386/i386.c (ix86_push_rounding): ...this new function.
17822 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
17824 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
17825 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
17826 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
17827 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
17828 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
17829 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
17830 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
17831 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
17832 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
17833 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
17835 * expr.c (emit_move_resolve_push): Treat the input and result
17836 of PUSH_ROUNDING as a poly_int64.
17837 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
17838 (emit_push_insn): Likewise.
17839 * lra-eliminations.c (mark_not_eliminable): Likewise.
17840 * recog.c (push_operand): Likewise.
17841 * reload1.c (elimination_effects): Likewise.
17842 * rtlanal.c (nonzero_bits1): Likewise.
17843 * calls.c (store_one_arg): Likewise. Require the padding to be
17844 known at compile time.
17846 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17847 Alan Hayward <alan.hayward@arm.com>
17848 David Sherwood <david.sherwood@arm.com>
17850 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
17851 Use plus_constant instead of gen_rtx_PLUS.
17853 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17854 Alan Hayward <alan.hayward@arm.com>
17855 David Sherwood <david.sherwood@arm.com>
17857 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
17858 rather than an int.
17860 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17861 Alan Hayward <alan.hayward@arm.com>
17862 David Sherwood <david.sherwood@arm.com>
17864 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
17865 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
17866 via stack temporaries. Treat the mode size as polynomial too.
17868 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17869 Alan Hayward <alan.hayward@arm.com>
17870 David Sherwood <david.sherwood@arm.com>
17872 * expr.c (expand_expr_real_2): When handling conversions involving
17873 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
17874 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
17875 as a poly_uint64 too.
17877 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17878 Alan Hayward <alan.hayward@arm.com>
17879 David Sherwood <david.sherwood@arm.com>
17881 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
17883 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17884 Alan Hayward <alan.hayward@arm.com>
17885 David Sherwood <david.sherwood@arm.com>
17887 * combine.c (can_change_dest_mode): Handle polynomial
17888 REGMODE_NATURAL_SIZE.
17889 * expmed.c (store_bit_field_1): Likewise.
17890 * expr.c (store_constructor): Likewise.
17891 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
17892 and polynomial REGMODE_NATURAL_SIZE.
17893 (gen_lowpart_common): Likewise.
17894 * reginfo.c (record_subregs_of_mode): Likewise.
17895 * rtlanal.c (read_modify_subreg_p): Likewise.
17897 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17898 Alan Hayward <alan.hayward@arm.com>
17899 David Sherwood <david.sherwood@arm.com>
17901 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
17902 numbers of elements.
17904 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17905 Alan Hayward <alan.hayward@arm.com>
17906 David Sherwood <david.sherwood@arm.com>
17908 * match.pd: Cope with polynomial numbers of vector elements.
17910 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17911 Alan Hayward <alan.hayward@arm.com>
17912 David Sherwood <david.sherwood@arm.com>
17914 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
17915 in a POINTER_PLUS_EXPR.
17917 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17918 Alan Hayward <alan.hayward@arm.com>
17919 David Sherwood <david.sherwood@arm.com>
17921 * omp-simd-clone.c (simd_clone_subparts): New function.
17922 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
17923 (ipa_simd_modify_function_body): Likewise.
17925 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17926 Alan Hayward <alan.hayward@arm.com>
17927 David Sherwood <david.sherwood@arm.com>
17929 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
17930 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
17931 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
17932 (expand_vector_condition, vector_element): Likewise.
17933 (subparts_gt): New function.
17934 (get_compute_type): Use subparts_gt.
17935 (count_type_subparts): Delete.
17936 (expand_vector_operations_1): Use subparts_gt instead of
17937 count_type_subparts.
17939 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17940 Alan Hayward <alan.hayward@arm.com>
17941 David Sherwood <david.sherwood@arm.com>
17943 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
17944 (vect_compile_time_alias): ...this new function. Do the calculation
17945 on poly_ints rather than trees.
17946 (vect_prune_runtime_alias_test_list): Update call accordingly.
17948 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17949 Alan Hayward <alan.hayward@arm.com>
17950 David Sherwood <david.sherwood@arm.com>
17952 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
17954 (vect_schedule_slp_instance): Likewise.
17956 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17957 Alan Hayward <alan.hayward@arm.com>
17958 David Sherwood <david.sherwood@arm.com>
17960 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
17961 constant and extern definitions for variable-length vectors.
17962 (vect_get_constant_vectors): Note that the number of units
17963 is known to be constant.
17965 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17966 Alan Hayward <alan.hayward@arm.com>
17967 David Sherwood <david.sherwood@arm.com>
17969 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
17970 of units as polynomial. Choose between WIDE and NARROW based
17973 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17974 Alan Hayward <alan.hayward@arm.com>
17975 David Sherwood <david.sherwood@arm.com>
17977 * tree-vect-stmts.c (simd_clone_subparts): New function.
17978 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
17980 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17981 Alan Hayward <alan.hayward@arm.com>
17982 David Sherwood <david.sherwood@arm.com>
17984 * tree-vect-stmts.c (vectorizable_call): Treat the number of
17985 vectors as polynomial. Use build_index_vector for
17986 IFN_GOMP_SIMD_LANE.
17988 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17989 Alan Hayward <alan.hayward@arm.com>
17990 David Sherwood <david.sherwood@arm.com>
17992 * tree-vect-stmts.c (get_load_store_type): Treat the number of
17993 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
17994 for variable-length vectors.
17995 (vectorizable_mask_load_store): Treat the number of units as
17996 polynomial, asserting that it is constant if the condition has
17997 already been enforced.
17998 (vectorizable_store, vectorizable_load): Likewise.
18000 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18001 Alan Hayward <alan.hayward@arm.com>
18002 David Sherwood <david.sherwood@arm.com>
18004 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
18005 of units as polynomial. Punt if we can't tell at compile time
18006 which vector contains the final result.
18008 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18009 Alan Hayward <alan.hayward@arm.com>
18010 David Sherwood <david.sherwood@arm.com>
18012 * tree-vect-loop.c (vectorizable_induction): Treat the number
18013 of units as polynomial. Punt on SLP inductions. Use an integer
18014 VEC_SERIES_EXPR for variable-length integer reductions. Use a
18015 cast of such a series for variable-length floating-point
18018 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18019 Alan Hayward <alan.hayward@arm.com>
18020 David Sherwood <david.sherwood@arm.com>
18022 * tree.h (build_index_vector): Declare.
18023 * tree.c (build_index_vector): New function.
18024 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
18025 of units as polynomial, forcibly converting it to a constant if
18026 vectorizable_reduction has already enforced the condition.
18027 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
18028 to create a {1,2,3,...} vector.
18029 (vectorizable_reduction): Treat the number of units as polynomial.
18030 Choose vectype_in based on the largest scalar element size rather
18031 than the smallest number of units. Enforce the restrictions
18034 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18035 Alan Hayward <alan.hayward@arm.com>
18036 David Sherwood <david.sherwood@arm.com>
18038 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
18039 number of units as polynomial.
18041 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18042 Alan Hayward <alan.hayward@arm.com>
18043 David Sherwood <david.sherwood@arm.com>
18045 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
18046 * target.def (autovectorize_vector_sizes): Return the vector sizes
18047 by pointer, using vector_sizes rather than a bitmask.
18048 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
18049 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
18050 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
18052 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
18053 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
18054 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
18055 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
18056 * omp-general.c (omp_max_vf): Likewise.
18057 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18058 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
18059 * tree-vect-loop.c (vect_analyze_loop): Likewise.
18060 * tree-vect-slp.c (vect_slp_bb): Likewise.
18061 * doc/tm.texi: Regenerate.
18062 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
18064 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
18065 the vector size as a poly_uint64 rather than an unsigned int.
18066 (current_vector_size): Change from an unsigned int to a poly_uint64.
18067 (get_vectype_for_scalar_type): Update accordingly.
18068 * tree.h (build_truth_vector_type): Take the size and number of
18069 units as a poly_uint64 rather than an unsigned int.
18070 (build_vector_type): Add a temporary overload that takes
18071 the number of units as a poly_uint64 rather than an unsigned int.
18072 * tree.c (make_vector_type): Likewise.
18073 (build_truth_vector_type): Take the number of units as a poly_uint64
18074 rather than an unsigned int.
18076 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18077 Alan Hayward <alan.hayward@arm.com>
18078 David Sherwood <david.sherwood@arm.com>
18080 * target.def (get_mask_mode): Take the number of units and length
18081 as poly_uint64s rather than unsigned ints.
18082 * targhooks.h (default_get_mask_mode): Update accordingly.
18083 * targhooks.c (default_get_mask_mode): Likewise.
18084 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
18085 * doc/tm.texi: Regenerate.
18087 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18088 Alan Hayward <alan.hayward@arm.com>
18089 David Sherwood <david.sherwood@arm.com>
18091 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
18092 * omp-general.c (omp_max_vf): Likewise.
18093 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
18094 (expand_omp_simd): Handle polynomial safelen.
18095 * omp-low.c (omplow_simd_context): Add a default constructor.
18096 (omplow_simd_context::max_vf): Change from int to poly_uint64.
18097 (lower_rec_simd_input_clauses): Update accordingly.
18098 (lower_rec_input_clauses): Likewise.
18100 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18101 Alan Hayward <alan.hayward@arm.com>
18102 David Sherwood <david.sherwood@arm.com>
18104 * tree-vectorizer.h (vect_nunits_for_cost): New function.
18105 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
18106 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
18107 (vect_analyze_slp_cost): Likewise.
18108 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
18109 (vect_model_load_cost): Likewise.
18111 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18112 Alan Hayward <alan.hayward@arm.com>
18113 David Sherwood <david.sherwood@arm.com>
18115 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
18116 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
18117 from an unsigned int * to a poly_uint64_pod *.
18118 (calculate_unrolling_factor): New function.
18119 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
18121 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18122 Alan Hayward <alan.hayward@arm.com>
18123 David Sherwood <david.sherwood@arm.com>
18125 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
18126 from an unsigned int to a poly_uint64.
18127 (_loop_vec_info::slp_unrolling_factor): Likewise.
18128 (_loop_vec_info::vectorization_factor): Change from an int
18130 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
18131 (vect_get_num_vectors): New function.
18132 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
18133 (vect_get_num_copies): Use vect_get_num_vectors.
18134 (vect_analyze_data_ref_dependences): Change max_vf from an int *
18135 to an unsigned int *.
18136 (vect_analyze_data_refs): Change min_vf from an int * to a
18138 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
18139 than an unsigned HOST_WIDE_INT.
18140 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
18141 (vect_analyze_data_ref_dependence): Change max_vf from an int *
18142 to an unsigned int *.
18143 (vect_analyze_data_ref_dependences): Likewise.
18144 (vect_compute_data_ref_alignment): Handle polynomial vf.
18145 (vect_enhance_data_refs_alignment): Likewise.
18146 (vect_prune_runtime_alias_test_list): Likewise.
18147 (vect_shift_permute_load_chain): Likewise.
18148 (vect_supportable_dr_alignment): Likewise.
18149 (dependence_distance_ge_vf): Take the vectorization factor as a
18150 poly_uint64 rather than an unsigned HOST_WIDE_INT.
18151 (vect_analyze_data_refs): Change min_vf from an int * to a
18153 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
18154 vfm1 as a poly_uint64 rather than an int. Make the same change
18155 for the returned bound_scalar.
18156 (vect_gen_vector_loop_niters): Handle polynomial vf.
18157 (vect_do_peeling): Likewise. Update call to
18158 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
18159 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
18161 * tree-vect-loop.c (vect_determine_vectorization_factor)
18162 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
18163 (vect_get_known_peeling_cost): Likewise.
18164 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
18165 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
18166 (vect_transform_loop): Likewise. Use the lowest possible VF when
18167 updating the upper bounds of the loop.
18168 (vect_min_worthwhile_factor): Make static. Return an unsigned int
18169 rather than an int.
18170 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
18171 polynomial unroll factors.
18172 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
18173 (vect_make_slp_decision): Likewise.
18174 (vect_supported_load_permutation_p): Likewise, and polynomial
18176 (vect_analyze_slp_cost): Handle polynomial vf.
18177 (vect_slp_analyze_node_operations): Likewise.
18178 (vect_slp_analyze_bb_1): Likewise.
18179 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
18180 than an unsigned HOST_WIDE_INT.
18181 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
18182 (vectorizable_load): Handle polynomial vf.
18183 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
18185 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
18187 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18188 Alan Hayward <alan.hayward@arm.com>
18189 David Sherwood <david.sherwood@arm.com>
18191 * match.pd: Handle bit operations involving three constants
18192 and try to fold one pair.
18194 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18196 * tree-vect-loop-manip.c: Include gimple-fold.h.
18197 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
18198 niters_maybe_zero parameters. Handle other cases besides a step of 1.
18199 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
18200 Add a path that uses a step of VF instead of 1, but disable it
18202 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
18203 and niters_no_overflow parameters. Update calls to
18204 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
18205 Create a new SSA name if the latter choses to use a ste other
18206 than zero, and return it via niters_vector_mult_vf_var.
18207 * tree-vect-loop.c (vect_transform_loop): Update calls to
18208 vect_do_peeling, vect_gen_vector_loop_niters and
18209 slpeel_make_loop_iterate_ntimes.
18210 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
18211 (vect_gen_vector_loop_niters): Update declarations after above changes.
18213 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
18215 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
18216 128-bit round to integer instructions.
18217 (ceil<mode>2): Likewise.
18218 (btrunc<mode>2): Likewise.
18219 (round<mode>2): Likewise.
18221 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18223 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
18224 unaligned VSX load/store on P8/P9.
18225 (expand_block_clear): Allow the use of unaligned VSX
18226 load/store on P8/P9.
18228 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18230 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
18232 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
18233 swap associated with both a load and a store.
18235 2018-01-02 Andrew Waterman <andrew@sifive.com>
18237 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
18238 * config/riscv/riscv.md (clear_cache): Use it.
18240 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
18242 * web.c: Remove out-of-date comment.
18244 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
18246 * expr.c (fixup_args_size_notes): Check that any existing
18247 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
18248 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
18249 (emit_single_push_insn): ...here.
18251 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
18253 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
18254 (const_vector_encoded_nelts): New function.
18255 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
18256 (const_vector_int_elt, const_vector_elt): Declare.
18257 * emit-rtl.c (const_vector_int_elt_1): New function.
18258 (const_vector_elt): Likewise.
18259 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
18260 of CONST_VECTOR_ELT.
18262 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
18264 * expr.c: Include rtx-vector-builder.h.
18265 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
18266 directly on the tree encoding.
18267 (const_vector_from_tree): Likewise.
18268 * optabs.c: Include rtx-vector-builder.h.
18269 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
18270 sequence of "u" values.
18271 * vec-perm-indices.c: Include rtx-vector-builder.h.
18272 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
18273 directly on the vec_perm_indices encoding.
18275 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
18277 * doc/rtl.texi (const_vector): Describe new encoding scheme.
18278 * Makefile.in (OBJS): Add rtx-vector-builder.o.
18279 * rtx-vector-builder.h: New file.
18280 * rtx-vector-builder.c: Likewise.
18281 * rtl.h (rtx_def::u2): Add a const_vector field.
18282 (CONST_VECTOR_NPATTERNS): New macro.
18283 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
18284 (CONST_VECTOR_DUPLICATE_P): Likewise.
18285 (CONST_VECTOR_STEPPED_P): Likewise.
18286 (CONST_VECTOR_ENCODED_ELT): Likewise.
18287 (const_vec_duplicate_p): Check for a duplicated vector encoding.
18288 (unwrap_const_vec_duplicate): Likewise.
18289 (const_vec_series_p): Check for a non-duplicated vector encoding.
18290 Say that the function only returns true for integer vectors.
18291 * emit-rtl.c: Include rtx-vector-builder.h.
18292 (gen_const_vec_duplicate_1): Delete.
18293 (gen_const_vector): Call gen_const_vec_duplicate instead of
18294 gen_const_vec_duplicate_1.
18295 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
18296 (gen_const_vec_duplicate): Use rtx_vector_builder.
18297 (gen_const_vec_series): Likewise.
18298 (gen_rtx_CONST_VECTOR): Likewise.
18299 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
18300 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
18301 Build a new vector rather than modifying a CONST_VECTOR in-place.
18302 (handle_special_swappables): Update call accordingly.
18303 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
18304 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
18305 Build a new vector rather than modifying a CONST_VECTOR in-place.
18306 (handle_special_swappables): Update call accordingly.
18308 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
18310 * simplify-rtx.c (simplify_const_binary_operation): Use
18311 CONST_VECTOR_ELT instead of XVECEXP.
18313 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
18315 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
18316 the selector elements to be different from the data elements
18317 if the selector is a VECTOR_CST.
18318 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
18319 ssizetype for the selector.
18321 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
18323 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
18324 before testing each element individually.
18325 * tree-vect-generic.c (lower_vec_perm): Likewise.
18327 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
18329 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
18330 * selftest-run-tests.c (selftest::run_tests): Call it.
18331 * vector-builder.h (vector_builder::operator ==): New function.
18332 (vector_builder::operator !=): Likewise.
18333 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
18334 (vec_perm_indices::all_from_input_p): New function.
18335 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
18336 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
18337 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
18338 instead of reading the VECTOR_CST directly. Detect whether both
18339 vector inputs are the same before constructing the vec_perm_indices,
18340 and update the number of inputs argument accordingly. Use the
18341 utility functions added above. Only construct sel2 if we need to.
18343 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
18345 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
18346 the broadcast of the low byte.
18347 (expand_mult_highpart): Use an explicit encoding for the permutes.
18348 * optabs-query.c (can_mult_highpart_p): Likewise.
18349 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
18350 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
18351 (vectorizable_bswap): Likewise.
18352 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
18353 explicit encoding for the power-of-2 permutes.
18354 (vect_permute_store_chain): Likewise.
18355 (vect_grouped_load_supported): Likewise.
18356 (vect_permute_load_chain): Likewise.
18358 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
18360 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
18361 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
18362 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
18363 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
18364 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
18365 (vect_gen_perm_mask_any): Likewise.
18367 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
18369 * int-vector-builder.h: New file.
18370 * vec-perm-indices.h: Include int-vector-builder.h.
18371 (vec_perm_indices): Redefine as an int_vector_builder.
18372 (auto_vec_perm_indices): Delete.
18373 (vec_perm_builder): Redefine as a stand-alone class.
18374 (vec_perm_indices::vec_perm_indices): New function.
18375 (vec_perm_indices::clamp): Likewise.
18376 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
18377 (vec_perm_indices::new_vector): New function.
18378 (vec_perm_indices::new_expanded_vector): Update for new
18379 vec_perm_indices class.
18380 (vec_perm_indices::rotate_inputs): New function.
18381 (vec_perm_indices::all_in_range_p): Operate directly on the
18382 encoded form, without computing elided elements.
18383 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
18384 encoding. Update for new vec_perm_indices class.
18385 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
18386 the given vec_perm_builder.
18387 (expand_vec_perm_var): Update vec_perm_builder constructor.
18388 (expand_mult_highpart): Use vec_perm_builder instead of
18389 auto_vec_perm_indices.
18390 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
18391 vec_perm_indices instead of auto_vec_perm_indices. Use a single
18392 or double series encoding as appropriate.
18393 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
18394 vec_perm_indices instead of auto_vec_perm_indices.
18395 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
18396 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
18397 (vect_permute_store_chain): Likewise.
18398 (vect_grouped_load_supported): Likewise.
18399 (vect_permute_load_chain): Likewise.
18400 (vect_shift_permute_load_chain): Likewise.
18401 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
18402 (vect_transform_slp_perm_load): Likewise.
18403 (vect_schedule_slp_instance): Likewise.
18404 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
18405 (vectorizable_mask_load_store): Likewise.
18406 (vectorizable_bswap): Likewise.
18407 (vectorizable_store): Likewise.
18408 (vectorizable_load): Likewise.
18409 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
18410 vec_perm_indices instead of auto_vec_perm_indices. Use
18411 tree_to_vec_perm_builder to read the vector from a tree.
18412 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
18413 vec_perm_builder instead of a vec_perm_indices.
18414 (have_whole_vector_shift): Use vec_perm_builder and
18415 vec_perm_indices instead of auto_vec_perm_indices. Leave the
18416 truncation to calc_vec_perm_mask_for_shift.
18417 (vect_create_epilog_for_reduction): Likewise.
18418 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
18419 from auto_vec_perm_indices to vec_perm_indices.
18420 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
18421 instead of changing individual elements.
18422 (aarch64_vectorize_vec_perm_const): Use new_vector to install
18423 the vector in d.perm.
18424 * config/arm/arm.c (expand_vec_perm_d::perm): Change
18425 from auto_vec_perm_indices to vec_perm_indices.
18426 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
18427 instead of changing individual elements.
18428 (arm_vectorize_vec_perm_const): Use new_vector to install
18429 the vector in d.perm.
18430 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
18431 Update vec_perm_builder constructor.
18432 (rs6000_expand_interleave): Likewise.
18433 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
18434 (rs6000_expand_interleave): Likewise.
18436 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
18438 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
18439 to qimode could truncate the indices.
18440 * optabs.c (expand_vec_perm_var): Likewise.
18442 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
18444 * Makefile.in (OBJS): Add vec-perm-indices.o.
18445 * vec-perm-indices.h: New file.
18446 * vec-perm-indices.c: Likewise.
18447 * target.h (vec_perm_indices): Replace with a forward class
18449 (auto_vec_perm_indices): Move to vec-perm-indices.h.
18450 * optabs.h: Include vec-perm-indices.h.
18451 (expand_vec_perm): Delete.
18452 (selector_fits_mode_p, expand_vec_perm_var): Declare.
18453 (expand_vec_perm_const): Declare.
18454 * target.def (vec_perm_const_ok): Replace with...
18455 (vec_perm_const): ...this new hook.
18456 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
18457 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
18458 * doc/tm.texi: Regenerate.
18459 * optabs.def (vec_perm_const): Delete.
18460 * doc/md.texi (vec_perm_const): Likewise.
18461 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
18462 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
18463 expand_vec_perm for constant permutation vectors. Assert that
18464 the mode of variable permutation vectors is the integer equivalent
18465 of the mode that is being permuted.
18466 * optabs-query.h (selector_fits_mode_p): Declare.
18467 * optabs-query.c: Include vec-perm-indices.h.
18468 (selector_fits_mode_p): New function.
18469 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
18470 is defined, instead of checking whether the vec_perm_const_optab
18471 exists. Use targetm.vectorize.vec_perm_const instead of
18472 targetm.vectorize.vec_perm_const_ok. Check whether the indices
18473 fit in the vector mode before using a variable permute.
18474 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
18475 vec_perm_indices instead of an rtx.
18476 (expand_vec_perm): Replace with...
18477 (expand_vec_perm_const): ...this new function. Take the selector
18478 as a vec_perm_indices rather than an rtx. Also take the mode of
18479 the selector. Update call to shift_amt_for_vec_perm_mask.
18480 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
18481 Use vec_perm_indices::new_expanded_vector to expand the original
18482 selector into bytes. Check whether the indices fit in the vector
18483 mode before using a variable permute.
18484 (expand_vec_perm_var): Make global.
18485 (expand_mult_highpart): Use expand_vec_perm_const.
18486 * fold-const.c: Includes vec-perm-indices.h.
18487 * tree-ssa-forwprop.c: Likewise.
18488 * tree-vect-data-refs.c: Likewise.
18489 * tree-vect-generic.c: Likewise.
18490 * tree-vect-loop.c: Likewise.
18491 * tree-vect-slp.c: Likewise.
18492 * tree-vect-stmts.c: Likewise.
18493 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
18495 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
18496 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
18497 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
18498 (aarch64_vectorize_vec_perm_const): ...this new function.
18499 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
18500 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
18501 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
18502 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
18503 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
18504 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
18505 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
18507 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
18508 check for NEON modes.
18509 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
18510 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
18511 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
18512 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
18514 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
18515 the old VEC_PERM_CONST conditions.
18516 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
18517 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
18518 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
18519 (ia64_vectorize_vec_perm_const_ok): Merge into...
18520 (ia64_vectorize_vec_perm_const): ...this new function.
18521 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
18522 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
18523 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
18524 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
18525 * config/mips/mips.c (mips_expand_vec_perm_const)
18526 (mips_vectorize_vec_perm_const_ok): Merge into...
18527 (mips_vectorize_vec_perm_const): ...this new function.
18528 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
18529 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
18530 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
18531 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
18532 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
18533 (rs6000_expand_vec_perm_const): Delete.
18534 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
18536 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
18537 (altivec_expand_vec_perm_const_le): Take each operand individually.
18538 Operate on constant selectors rather than rtxes.
18539 (altivec_expand_vec_perm_const): Likewise. Update call to
18540 altivec_expand_vec_perm_const_le.
18541 (rs6000_expand_vec_perm_const): Delete.
18542 (rs6000_vectorize_vec_perm_const_ok): Delete.
18543 (rs6000_vectorize_vec_perm_const): New function.
18544 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
18545 an element count and rtx array.
18546 (rs6000_expand_extract_even): Update call accordingly.
18547 (rs6000_expand_interleave): Likewise.
18548 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
18549 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
18550 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
18551 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
18552 (rs6000_expand_vec_perm_const): Delete.
18553 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
18554 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
18555 (altivec_expand_vec_perm_const_le): Take each operand individually.
18556 Operate on constant selectors rather than rtxes.
18557 (altivec_expand_vec_perm_const): Likewise. Update call to
18558 altivec_expand_vec_perm_const_le.
18559 (rs6000_expand_vec_perm_const): Delete.
18560 (rs6000_vectorize_vec_perm_const_ok): Delete.
18561 (rs6000_vectorize_vec_perm_const): New function. Remove stray
18562 reference to the SPE evmerge intructions.
18563 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
18564 an element count and rtx array.
18565 (rs6000_expand_extract_even): Update call accordingly.
18566 (rs6000_expand_interleave): Likewise.
18567 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
18568 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
18570 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
18572 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
18574 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
18575 vector mode and that that mode matches the mode of the data
18577 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
18578 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
18579 directly using expand_vec_perm_1 when forcing selectors into
18581 (expand_vec_perm_var): New function, split out from expand_vec_perm.
18583 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
18585 * optabs-query.h (can_vec_perm_p): Delete.
18586 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
18587 * optabs-query.c (can_vec_perm_p): Split into...
18588 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
18589 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
18590 particular selector is valid.
18591 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
18592 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
18593 (vect_grouped_load_supported): Likewise.
18594 (vect_shift_permute_load_chain): Likewise.
18595 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
18596 (vect_transform_slp_perm_load): Likewise.
18597 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
18598 (vectorizable_bswap): Likewise.
18599 (vect_gen_perm_mask_checked): Likewise.
18600 * fold-const.c (fold_ternary_loc): Likewise. Don't take
18601 implementations of variable permutation vectors into account
18602 when deciding which selector to use.
18603 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
18604 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
18605 with a false third argument.
18606 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
18607 to test whether the constant selector is valid and can_vec_perm_var_p
18608 to test whether a variable selector is valid.
18610 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
18612 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
18613 * optabs-query.c (can_vec_perm_p): Likewise.
18614 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
18615 instead of vec_perm_indices.
18616 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
18617 (vect_gen_perm_mask_checked): Likewise,
18618 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
18619 (vect_gen_perm_mask_checked): Likewise,
18621 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
18623 * optabs-query.h (qimode_for_vec_perm): Declare.
18624 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
18625 (qimode_for_vec_perm): ...this new function.
18626 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
18628 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18630 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
18631 does not have a conditional at the top.
18633 2018-01-02 Richard Biener <rguenther@suse.de>
18635 * ipa-inline.c (big_speedup_p): Fix expression.
18637 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
18640 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
18643 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
18647 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
18648 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
18649 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
18650 cond_taken_branch_cost 3->4.
18652 2018-01-01 Jakub Jelinek <jakub@redhat.com>
18654 PR tree-optimization/83581
18655 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
18656 TODO_cleanup_cfg if any changes have been made.
18658 PR middle-end/83608
18659 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
18660 convert_modes if target mode has the right side, but different mode
18663 PR middle-end/83609
18664 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
18665 last argument when extracting from CONCAT. If either from_real or
18666 from_imag is NULL, use expansion through memory. If result is not
18667 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
18668 the parts directly to inner mode, if even that fails, use expansion
18671 PR middle-end/83623
18672 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
18673 check for bswap in mode rather than HImode and use that in expand_unop
18676 Copyright (C) 2018 Free Software Foundation, Inc.
18678 Copying and distribution of this file, with or without modification,
18679 are permitted in any medium without royalty provided the copyright
18680 notice and this notice are preserved.