1 2018-10-18 David Malcolm <dmalcolm@redhat.com>
3 PR tree-optimization/87562
4 * input.c (get_substring_ranges_for_loc): Use
5 LRK_SPELLING_LOCATION rather than LRK_MACRO_EXPANSION_POINT when
6 getting the linemap for the endpoint. Verify that it's either
7 in the same linemap as the start point's spelling location, or
8 at least in the same file.
10 2018-10-18 Richard Biener <rguenther@suse.de>
12 * config/i386/i386.c (ix86_builtin_vectorization_cost): Do not
13 feed width-specific load/store costs through ix86_vec_cost.
14 * config/i386/x86-tune-costs.h (athlon_cost): Adjust.
16 (bdver_cost): Likewise.
17 (znver1_cost): Likewise.
18 (btver1_cost): Likewise.
19 (btver2_cost): Likewise.
21 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
23 * simplify-rtx.c (simplify_subreg): Call simplify_gen_subreg
24 to simplify subreg of vec_merge.
26 2018-10-18 Richard Biener <rguenther@suse.de>
28 * config/i386/i386.c: Fix costing of vector FMA.
30 2018-10-18 Richard Biener <rguenther@suse.de>
32 * config/i386/i386.c (ix86_vec_cost): Remove !parallel path
34 (ix86_builtin_vectorization_cost): For vec_construct properly
35 cost insertion into SSE regs.
36 (...): Adjust calls to ix86_vec_cost.
38 2018-10-18 Richard Biener <rguenther@suse.de>
42 2018-02-07 Richard Biener <rguenther@suse.de>
44 PR tree-optimization/84204
45 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
48 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
51 * simplify-rtx.c (simplify_subreg): Simplify subreg of vec_merge
53 (test_vector_ops_duplicate): Add test for a scalar subreg of a
54 VEC_MERGE of a VEC_DUPLICATE.
56 2018-10-17 Joseph Myers <joseph@codesourcery.com>
58 * doc/cpp.texi (__STDC_VERSION__): Document C2X handling.
59 * doc/invoke.texi (-std=c2x, -std=gnu2x): Document new options.
60 * doc/standards.texi (C Language): Document C2X.
61 * dwarf2out.c (highest_c_language), config/rl78/rl78.c
62 (rl78_option_override): Handle "GNU C2X" language name.
64 2018-10-17 Joseph Myers <joseph@codesourcery.com>
66 * doc/invoke.texi (-std=c17), doc/standards.texi (C Language):
67 Document C17 as published in 2018.
69 2018-10-17 Eric Botcazou <ebotcazou@adacore.com>
72 * fold-const.c (fold_truth_andor_1): If the right side is not constant,
73 bail out if both sides do not have the same storage order.
75 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
77 * bitmap.c (bitmap_head::dump): New.
78 * bitmap.h (bitmap_head): Add dump().
79 * gimple-ssa-evrp-analyze.c
80 (evrp_range_analyzer::try_find_new_range): Adjust for value_range API.
81 (evrp_range_analyzer::set_ssa_range_info): Same.
82 (evrp_range_analyzer::record_ranges_from_phis): Same.
83 (evrp_range_analyzer::record_ranges_from_stmt): Same.
84 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Same.
85 * gimple-ssa-sprintf.c (get_int_range): Same.
86 (format_integer): Same.
87 (sprintf_dom_walker::handle_gimple_call): Same.
88 * ipa-cp.c (ipcp_vr_lattice::meet_with_1): Same.
89 (ipcp_vr_lattice::top_p): Same.
90 (ipcp_vr_lattice::bottom_p): Same.
91 (ipcp_vr_lattice::set_to_bottom): Same.
92 (ipa_vr_operation_and_type_effects): Same.
93 (propagate_vr_across_jump_function): Same.
94 (ipcp_store_vr_results): Same.
95 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Same.
96 (ipa_print_node_jump_functions_for_edge): Same.
97 (ipa_get_value_range): Same.
98 (ipa_compute_jump_functions_for_edge): Same.
99 (ipa_write_jump_function): Same.
100 * tree-ssa-dom.c (simplify_stmt_for_jump_threading): Same.
101 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
103 * vr-values.c (set_value_range_to_nonnegative): Same.
104 (set_value_range_to_truthvalue): Same.
105 (vr_values::get_value_range): Same.
106 (vr_values::set_defs_to_varying): Same.
107 (vr_values::update_value_range): Same.
108 (symbolic_range_based_on_p): Same.
109 (vr_values::op_with_boolean_value_range_p): Same.
110 (vr_values::extract_range_for_var_from_comparison_expr): Same.
111 (vr_values::extract_range_from_ssa_name): Same.
112 (vr_values::extract_range_from_binary_expr): Same.
113 (vr_values::extract_range_from_unary_expr): Same.
114 (vr_values::extract_range_from_cond_expr): Same.
115 (vr_values::extract_range_from_comparison): Same.
116 (vr_values::check_for_binary_op_overflow): Same.
117 (vr_values::extract_range_basic): Same.
118 (vr_values::extract_range_from_assignment): Same.
119 (compare_ranges): Same.
120 (compare_range_with_value): Same.
121 (vr_values::adjust_range_with_scev): Same.
122 (vrp_valueize): Same.
123 (vrp_valueize_1): Same.
124 (vr_values::get_vr_for_comparison): Same.
125 (vr_values::compare_name_with_value): Same.
126 (vr_values::compare_names): Same.
127 (vr_values::vrp_evaluate_conditional): Same.
128 (find_case_label_ranges): Same.
129 (vr_values::vrp_visit_switch_stmt): Same.
130 (vr_values::extract_range_from_phi_node): Same.
131 (vr_values::simplify_div_or_mod_using_ranges): Same.
132 (vr_values::simplify_bit_ops_using_ranges): Same.
133 (test_for_singularity): Same.
134 (range_fits_type_p): Same.
135 (vr_values::simplify_cond_using_ranges_1): Same.
136 (vr_values::simplify_switch_using_ranges): Same.
137 (vr_values::simplify_float_conversion_using_ranges): Same.
138 (vr_values::two_valued_val_range_p): Same.
139 (vr_values::add_equivalence): Move to value_range::equiv_add.
140 * vr-values.h (vr_values::add_equivalence): Remove.
141 (VR_INITIALIZER): Remove.
142 * tree-vrp.c (value_range::set): New.
143 (value_range::equiv_add): New.
144 (value_range::value_range): New.
145 (value_range::deep_copy): New.
146 (value_range::check): New.
147 (value_range::equal_p): New.
148 (value_range::ignore_equivs_equal_p): New.
149 (value_range::operator==): New.
150 (value_range::operator!=): New.
151 (value_range::symbolic_p): New.
152 (value_range::numeric_p): New.
153 (value_range::set_undefined): New.
154 (value_range::set_varying): New.
155 (value_range::may_contain_p): New.
156 (value_range::equiv_clear): New.
157 (value_range::singleton_p): New.
158 (value_range::intersect): New.
159 (value_range::dump): New.
160 (value_range::set_and_canonicalize): New.
161 (set_value_range): Adjust for value_range API.
162 (set_value_range_to_undefined): Same.
163 (set_value_range_to_varying): Same.
164 (set_and_canonicalize_value_range): Same.
165 (set_value_range_to_nonnull): Same.
166 (set_value_range_to_null): Same.
167 (range_is_null): Same.
168 (range_is_nonnull): Same.
169 (range_int_cst_p): Same.
170 (range_int_cst_singleton_p): Same.
171 (symbolic_range_p): Same.
172 (range_includes_zero_p): Same.
173 (value_range_constant_singleton): Same.
174 (vrp_set_zero_nonzero_bits): Same.
175 (ranges_from_anti_range): Same.
176 (extract_range_into_wide_ints): Same.
177 (extract_range_from_multiplicative_op): Same.
178 (set_value_range_with_overflow): Same.
179 (extract_range_from_binary_expr_1): Same.
180 (extract_range_from_unary_expr): Same.
181 (dump_value_range): Same.
182 (debug_value_range): Same.
183 (vrp_prop::check_array_ref): Same.
184 (vrp_prop::check_mem_ref): Same.
185 (vrp_prop::vrp_initialize): Same.
186 (vrp_prop::visit_stmt): Same.
187 (intersect_ranges): Same.
188 (vrp_prop::visit_phi): Same.
189 (vrp_prop::vrp_finalize): Same.
190 (determine_value_range_1): Same.
191 (determine_value_range): Same.
192 (vrp_intersect_ranges_1): Rename to...
193 (vrp_intersect_1): this.
194 (vrp_intersect_ranges): Rename to...
195 (value_range::intersect_helper): ...this.
196 (vrp_meet_1): Rename to...
197 (value_range::union_helper): ...this.
198 (vrp_meet): Rename to...
199 (value_range::union_): ...this.
200 (copy_value_range): Remove.
201 * tree-vrp.h (struct value_range): Rewrite into a proper class.
202 (value_range::vrtype): New.
203 (value_range::type): New.
204 (value_range::equiv): New.
205 (value_range::min): New.
206 (value_range::max): New.
207 (value_range::varying_p): New.
208 (value_range::undefined_p): New.
209 (value_range::null_p): New.
210 (value_range::equiv_add): New.
211 (copy_value_range): Remove.
213 2018-10-17 David Malcolm <dmalcolm@redhat.com>
215 * Makefile.in (SELFTEST_TARGETS): New.
216 (selftest) Change from s-selftest-c to $(SELFTEST_TARGETS).
217 (C_SELFTEST_FLAGS, C_SELFTEST_DEPS, s-selftest-c, selftest-c-gdb)
218 (selftest-gdb, selftest-c-valgrind, selftest-valgrind): Move to
220 (CPP_SELFTEST_FLAGS, CPP_SELFTEST_DEPS, s-selftest-c++)
221 (selftest-c++-gdb, selftest-c++-valgrind): Move to
223 * configure: Regenerate.
224 * configure.ac (selftest_languages): New.
226 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
228 * tree-vrp.c (extract_range_from_multiplicative_op): Remove
229 overflow wraps argument.
230 (extract_range_from_binary_expr_1): Do not pass overflow wraps to
231 wide_int_range_multiplicative_op.
232 * wide-int-range.cc (wide_int_range_mult_wrapping): Remove
233 overflow wraps argument.
234 (wide_int_range_multiplicative_op): Same.
235 (wide_int_range_lshift): Same.
236 (wide_int_range_div): Same.
237 * wide-int-range.h (wide_int_range_multiplicative_op): Same.
238 (wide_int_range_lshift): Same.
239 (wide_int_range_div): Same.
241 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
243 * wide-int-range.h (wide_int_range_shift_undefined_p): Adjust to
244 use sign as argument.
245 * tree-vrp.c (extract_range_from_binary_expr_1): Pass sign to
246 wide_int_range_shift_undefined_p.
248 2018-10-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
250 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>):
252 (@despeculate_copy<ALLI_TI:mode>): ... This.
253 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): Remove
256 2018-10-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
258 * config.gcc: Obsolete *-*-solaris2.10*.
259 * doc/install.texi (Specific, *-*-solaris2*): Document it.
261 2018-10-12 Jeff Law <law@redhat.com>
263 * config/ft32/ft32.md (ft32_general_movsrc_operand): Disable
264 reg + sym +- const_int addressing modes.
266 2018-10-15 David Malcolm <dmalcolm@redhat.com>
268 * common.opt (fdiagnostics-minimum-margin-width=): New option.
269 * diagnostic-show-locus.c (layout::layout): Apply the minimum
271 (layout::start_annotation_line): Only print up to 3 of the
272 margin character, to avoid touching the left-hand side.
273 (selftest::test_diagnostic_show_locus_fixit_lines): Update for
274 minimum margin width, as set by test_diagnostic_context's ctor.
275 (selftest::test_fixit_insert_containing_newline): Likewise.
276 (selftest::test_fixit_insert_containing_newline_2): Likewise.
277 (selftest::test_line_numbers_multiline_range): Clear
279 * diagnostic.c (diagnostic_initialize): Initialize
281 * diagnostic.h (struct diagnostic_context): Add field
283 * doc/invoke.texi: Add -fdiagnostics-minimum-margin-width=.
284 * opts.c (common_handle_option): Handle
285 OPT_fdiagnostics_minimum_margin_width_.
286 * selftest-diagnostic.c
287 (selftest::test_diagnostic_context::test_diagnostic_context):
288 Initialize min_margin_width to 6.
289 * toplev.c (general_init): Initialize global_dc->min_margin_width.
291 2018-10-15 David Malcolm <dmalcolm@redhat.com>
293 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
294 Fix usage of "error_at_rich_loc" in the comment.
296 2018-10-15 Renlin Li <renlin.li@arm.com>
299 * tree-vectorizer.c (try_vectorize_loop_1): Don't use
300 if-conversioned loop when it contains ifn with types not
301 supported by backend.
302 * internal-fn.c (expand_direct_optab_fn): Add an assert.
303 (direct_internal_fn_supported_p): New helper function.
304 * internal-fn.h (direct_internal_fn_supported_p): Declare.
306 2018-10-15 Jakub Jelinek <jakub@redhat.com>
309 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
310 Remove OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
311 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
313 2018-10-15 Bin Cheng <bin.cheng@linux.alibaba.com>
315 PR tree-optimization/87022
316 * tree-loop-distribution.c (pg_add_dependence_edges): Check all
317 bits in dist vector rather than the first one.
319 2018-10-15 Richard Biener <rguenther@suse.de>
322 * tree-ssa-structalias.c (struct vls_data): Add escaped_p member.
323 (visit_loadstore): When a used restrict tag escaped verify that
324 the points-to solution of "other" pointers do not include
326 (compute_dependence_clique): If a used restrict tag escaped
327 communicated that down to visit_loadstore.
329 2018-10-15 Andreas Krebbel <krebbel@linux.ibm.com>
331 * config/s390/s390.c (s390_expand_vec_init): Force vector element
332 into reg if it isn't a general operand.
334 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
337 * config/i386/sse.md (*vec_dupv2di): Add register source to
340 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
343 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
344 Add OPTION_MASK_ISA_AVX512IFMA_UNSET,
345 OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
346 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
348 2018-10-13 Eric Botcazou <ebotcazou@adacore.com>
350 * dwarf2cfi.c (struct dw_trace_info): Add args_size_defined_for_eh.
351 (notice_args_size): Set it in the current trace if no insn that can
352 throw internally has been seen yet.
353 (connect_traces): When connecting args_size between traces, allow the
354 incoming values not to match if there is an insn setting it before the
355 first insn that can throw internally; in that case, force the creation
356 of a CFI note on this latter insn.
358 2018-10-13 Jonathan Wakely <jwakely@redhat.com>
360 * opt-problem.h (opt_wrapper): Use template-argument-list when naming
361 the base class, because using the injected-class-name was not clearly
362 specified until DR 176.
364 2018-10-12 Paul Koning <ni1d@arrl.net>
366 * config/pdp11/pdp11.md (doloop_end): New expander.
367 (doloop_end_insn): renamed from "doloop_end".
368 (addqi3): New pattern.
369 (subqi3): New pattern.
370 * config/pdp11/predicates.md (incdec_operand): New predicate.
372 2018-10-12 Yury Gribov <tetra2005@gmail.com>
375 * real.c (format_helper::can_represent_integral_type_p): New function
376 * real.h (format_helper::can_represent_integral_type_p): Ditto.
377 * match.pd: New pattern.
379 2018-10-12 Alexandre Oliva <oliva@adacore.com>
381 * configure.ac: Introduce --enable-large-address-aware
382 to define MINGW_DEFAULT_LARGE_ADDR_AWARE.
383 * doc/install.texi: Document it.
384 * configure, config.in: Rebuilt.
385 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Define,
386 based on MINGW_DEFAULT_LARGE_ADDR_AWARE.
387 (LINK_SPEC): Insert it.
388 * config/i386/mingw-264.h: Likewise.
390 * cygming.h (HAVE_GAS_ALIGNED_COMM): Fallback-define.
392 2018-10-12 Peter Bergner <bergner@linux.ibm.com>
394 PR rtl-optimization/87600
395 * ira-lives (non_conflicting_reg_copy_p): Disable for non LRA targets.
397 2018-10-12 Paul Koning <ni1d@arrl.net>
399 * doc/md.texi (doloop_end): Document that the pattern code may
400 need to check operand mode.
402 2018-10-12 Wilco Dijkstra <wdijkstr@arm.com>
404 * config/aarch64/aarch64.md (zero_extendsidi2_aarch64): Add alternatives
405 to zero-extend between int and floating-point registers.
406 (load_pair_zero_extendsidi2_aarch64): Add alternative for zero-extended
407 ldp into floating-point registers. Add type and arch attributes.
408 (zero_extend<SHORT:mode><GPI:mode>2_aarch64): Add arch attribute.
409 Use f_loads for type attribute.
411 2018-10-11 Martin Sebor <msebor@redhat.com>
413 * doc/extend.texi (attribute packed): Correct typos.
415 2018-10-11 Martin Sebor <msebor@redhat.com>
417 * doc/extend.texi (attribute flatten): Mention interaction with
420 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
423 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
424 Set new_decl virtual flag to zero.
426 2018-10-11 Martin Sebor <msebor@redhat.com>
429 * doc/extend.texi (attribute format_arg): Discuss using multiple
430 attributes on a single function.
432 2018-10-11 Giuliano Belinassi <giuliano.belinassi@usp.br>
434 PR tree-optimization/86829
435 * match.pd (sin (atan (x))): New simplification rules.
436 (cos (atan (x))): Likewise.
437 * real.c (build_sinatan_real): New function.
438 * real.h (build_sinatan_real): Prototype.
440 2018-10-11 Will Schmidt <will_schmidt@vnet.ibm.com>
442 * config/rs6000/rs6000.c (map_to_integral_tree_type): New helper
444 (fold_mergeeo_helper): New helper function.
445 (rs6000_gimple_fold_builtin): Add hooks for vec_mergee and vec_mergeo
446 intrinsics. Correct some whitespace indentation issues.
448 2018-10-11 Wilco Dijkstra <wdijkstr@arm.com>
451 * config/aarch64/aarch64.c (aarch64_mask_and_shift_for_ubfiz_p):
452 Use HOST_WIDE_INT_1U for shift.
454 2018-10-11 Doug Rupp <rupp@adacore.com>
455 Olivier Hainque <hainque@adacore.com>
457 * config/rs6000/vxworks.h (VXWORKS_RELAX_LINK_SPEC): New macro.
458 Pass --relax to the linker for RTPs.
459 (LINK_SPEC): Append VXWORKS_RELAX_LINK_SPEC.
461 2018-10-11 Andrew Stubbs <ams@codesourcery.com>
462 Jan Hubicka <jh@suse.cz>
463 Martin Jambor <mjambor@suse.cz>
465 * print-rtl.c (print_rtx_operand_codes_E_and_V): Print how many times
466 the same elements are repeated rather than printing all of them.
467 * read-rtl.c (rtx_reader::read_rtx_operand): Recognize and expand
469 * read-rtl-function.c (test_loading_repeat): New function.
470 (read_rtl_function_c_tests): Call test_loading_repeat.
471 * rtl-tests.c (test_dumping_repeat): New function.
472 (rtl_tests_c_tests): Call test_dumping_repeat.
474 2018-10-11 Richard Biener <rguenther@suse.de>
476 * config/i386/x86-tune-costs.h (bdver?_memcpy, bdver?_memset,
477 bdver?_cost): Unify to ...
478 (bdver_memcpy, bdver_memset, bdver_cost): ... this.
479 * config/i386/i386.c (processor_cost_table): Adjust.
481 2018-10-10 Eric Botcazou <ebotcazou@adacore.com>
484 * cgraphunit.c (cgraph_node::expand_thunk): Force DECL_IGNORED_P on
485 the thunk when expanding to GIMPLE.
487 2018-10-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
489 * varasm.c (mergeable_string_section): Don't try to move zero-length
490 strings to the merge section.
492 2018-10-10 Uros Bizjak <ubizjak@gmail.com>
495 * config/i386/mmx.md (const_vector 0 -> mem splitter): New splitter.
497 2018-10-10 Jakub Jelinek <jakub@redhat.com>
500 * config/i386/i386-builtin.def (IX86_BUILTIN_RDPMC): Move from args set
503 2018-10-10 Richard Biener <rguenther@suse.de>
505 * config/i386/sse.md (reduc_plus_scal_v8df, reduc_plus_scal_v4df,
506 reduc_plus_scal_v2df, reduc_plus_scal_v16sf, reduc_plus_scal_v8sf,
507 reduc_plus_scal_v4sf): Merge into pattern reducing to half width
508 and recursing and pattern terminating the recursion on SSE
509 vector width using ix86_expand_reduc.
510 (reduc_sminmax_scal_<mode>): Split into part reducing to half
511 width and recursing and SSE2 vector variant doing the final
512 reduction with ix86_expand_reduc.
513 (reduc_uminmax_scal_<mode>): Likewise for the AVX512 variants
514 with terminating the recursion at AVX level, splitting that
517 2018-10-09 David Malcolm <dmalcolm@redhat.com>
519 * genmatch.c (error_cb): Rename to...
520 (diagnostic_cb): ...this, converting int params to enums.
521 (fatal_at): Update for renaming.
522 (warning_at): Likewise.
524 * input.c (selftest::ebcdic_execution_charset::apply):
525 Update for renaming of...
526 (selftest::ebcdic_execution_charset::on_error): ...this, renaming
528 (selftest::ebcdic_execution_charset::on_diagnostic): ...this,
529 converting level and reason to enums.
530 (class selftest::lexer_error_sink): Rename to...
531 (class selftest::lexer_test_options): ...this, renaming field
532 "m_errors" to "m_diagnostics".
533 (selftest::lexer_test_options::apply): Update for renaming of...
534 (selftest::lexer_test_options::on_error): ...this, renaming to...
535 (selftest::lexer_test_options::on_diagnostic): ...this
536 converting level and reason to enums.
537 (selftest::test_lexer_string_locations_raw_string_unterminated):
538 Update for renamings.
539 * opth-gen.awk (struct cpp_reason_option_codes_t): Use enum for
542 2018-10-09 Paul A. Clarke <pc@us.ibm.com>
544 * config.gcc (powerpc*-*-*): Add pmmintrin.h to extra_headers.
545 * config/rs6000/pmmintrin.h: New file.
547 2018-10-09 Eric Botcazou <ebotcazou@adacore.com>
549 PR tree-optimization/86659
550 * gimple-match.h (gimple_match_op constructors): Initialize reverse.
552 2018-10-09 Richard Biener <rguenther@suse.de>
554 PR tree-optimization/63155
555 * tree-ssa-structalias.c: Include tree-ssa.h.
556 (get_constraint_for_ssa_var): For undefs return nothing_id.
557 (find_func_aliases): Cleanup PHI handling.
559 2018-10-09 Richard Biener <rguenther@suse.de>
561 * tree-vectorizer.c (dump_stmt_cost): Fix cut&paste missing
564 2018-10-09 Martin Liska <mliska@suse.cz>
566 * asan.c (asan_emit_stack_protection): If a stack variable
567 is located in a same file as current function, then emit
568 line info into variable definition string.
570 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
572 * print-rtl.c (rtx_writer::print_rtx_operand_code_i): Print column
575 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
577 * cgraphunit.c (cgraph_node::expand_thunk): Do not force DECL_IGNORED_P
580 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
582 PR tree-optimization/86659
583 * gimple-match.h (struct gimple_match_op): Add reverse field.
584 (gimple_match_op::set_op): New overloaded method.
585 * gimple-match-head.c (maybe_build_generic_op) <BIT_FIELD_REF>: Set
586 the REF_REVERSE_STORAGE_ORDER flag on the value.
587 (gimple_simplify) <GIMPLE_ASSIGN>: For BIT_FIELD_REF, propagate the
588 REF_REVERSE_STORAGE_ORDER flag and avoid simplifying if it is set.
590 2018-10-08 Richard Sandiford <richard.sandiford@arm.com>
593 * gimple-ssa-backprop.c (backprop::intersect_uses): Use
594 FOR_EACH_IMM_USE_FAST instead of FOR_EACH_IMM_USE_STMT.
596 2018-10-08 H.J. Lu <hongjiu.lu@intel.com>
599 * config/i386/avx512fintrin.h (_mm512_mask_fmaddsub_round_pd):
600 Defined with __builtin_ia32_vfmaddsubpd512_mask.
602 2018-10-08 Richard Biener <rguenther@suse.de>
604 * config/i386/x86-tune-costs.h (znver1_cost): Make AVX256 vector loads
605 cost the same as AVX128 ones.
607 2018-10-08 Paul Koning <ni1d@arrl.net>
609 * config/pdp11/pdp11-protos.h (output_block_move): Remove.
610 (expand_block_move): New function.
611 * config/pdp11/pdp11.c (output_block_move): Remove.
612 (expand_block_move): New function.
613 * config/pdp11/pdp11.h (MOVE_RATIO): New definition.
614 * config/pdp11/pdp11.md (movmemhi): Use expand_block_move.
615 (*movmemhi1): Remove.
617 2018-10-08 Robin Dapp <rdapp@linux.ibm.com>
619 * config/s390/2827.md: Increase latencies for some FP instructions.
621 2018-10-08 Richard Biener <rguenther@suse.de>
623 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
625 * tree-vectorizer.c (dump_stmt_cost): Add cost param and dump it.
626 * tree-vectorizer.h (dump_stmt_cost): Adjust.
627 (add_stmt_cost): Dump return value of the hook.
629 2018-10-08 Richard Biener <rguenther@suse.de>
631 PR tree-optimization/63155
632 * tree-ssa-propagate.c (add_ssa_edge): Do cheap check first.
633 (ssa_propagation_engine::ssa_propagate): Remove redundant
636 2018-10-05 Peter Bergner <bergner@linux.ibm.com>
638 PR rtl-optimization/86939
639 PR rtl-optimization/87479
640 * ira.h (non_conflicting_reg_copy_p): New prototype.
641 * ira-lives.c (ignore_reg_for_conflicts): New static variable.
642 (make_hard_regno_dead): Don't add conflicts for register
643 ignore_reg_for_conflicts.
644 (make_object_dead): Likewise.
645 (non_conflicting_reg_copy_p): New function.
646 (process_bb_node_lives): Set ignore_reg_for_conflicts for copies.
647 Remove special conflict handling of REAL_PIC_OFFSET_TABLE_REGNUM.
648 * lra-lives.c (ignore_reg_for_conflicts): New static variable.
649 (make_hard_regno_dead): Don't add conflicts for register
650 ignore_reg_for_conflicts. Remove special conflict handling of
651 REAL_PIC_OFFSET_TABLE_REGNUM. Remove now unused argument
652 check_pic_pseudo_p and update callers.
653 (mark_pseudo_dead): Don't add conflicts for register
654 ignore_reg_for_conflicts.
655 (process_bb_lives): Set ignore_reg_for_conflicts for copies.
657 2018-10-05 Andrew Waterman <andrew@sifive.com>
658 Jim Wilson <jimw@sifive.com>
660 * config/riscv/riscv.md (f<quiet_pattern>_quiet<ANYF:mode><X:mode>4):
661 Add define_expand. Add ! HONOR_SNANS check to current pattern. Add
662 new pattern using HONOR_SNANS that emits one extra instruction.
664 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
666 * config/rs6000/rs6000.md (unnamed mfcr scc_comparison_operator
667 patterns): Merge SI and DI patterns to a GPR pattern.
668 (unnamed define_insn and define_split for record form of that): Merge
669 to a single define_insn_and_split pattern.
671 2018-10-05 David Malcolm <dmalcolm@redhat.com>
674 * input.c (expand_location_to_spelling_point): Add param "aspect"
675 and use rather than hardcoding LOCATION_ASPECT_CARET.
676 (get_substring_ranges_for_loc): Handle the case of a single token
677 within a macro expansion.
678 * input.h (expand_location_to_spelling_point): Add "aspect" param,
679 defaulting to LOCATION_ASPECT_CARET.
681 2018-10-05 Paul Koning <ni1d@arrl.net>
683 * config/pdp11/pdp11.c (TARGET_CXX_GUARD_TYPE): Define.
684 (TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT): Define.
685 (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
686 (pdp11_guard_type): New function.
688 2018-10-05 Paul Koning <ni1d@arrl.net>
690 * config/pdp11/pdp11.h (FLOAT_TYPE_SIZE): Always 32.
691 * config/pdp11/pdp11.opt (mfloat32): Remove.
693 * doc/invoke.texi (pdp11 -mfloat32): Remove:
694 (pdp11 -mfloat64): Remove.
696 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
698 * config/i386/i386.md (*cmpxf_cc_i387): Remove pattern.
699 (*cmp<mode>_cc_i387): Ditto.
700 (*cmpu<mode>_cc_i387): Ditto.
701 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
702 * config/i386/i386.c (ix86_expand_fp_compare): Remove
704 <case IX86_FPCMP_SAHF>: Do not generate pattern with HImode clobber.
705 Emit x86_sahf_1 pattern.
706 (ix86_expand_compare): Update call to ix86_expand_fp_compare.
707 (ix86_expand_carry_flag_compare): Ditto.
709 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
711 * config/i386/i386.md (*cmpxf_i387): Change operand 2 predicate
712 to reg_or_0_operand. Add "C" constraint.
713 (*cmpxf_cc_i387): Ditto.
714 (*cmp<mode>_i387): Change operand 2 predicate
715 to nonimm_or_0_operand. Add "C" constraint.
716 (*cmp<mode>_cc_i387): Ditto.
717 (*cmp<mode>_0_i387): Remove insn pattern.
718 (*cmp<mode>_0_cc_i387): Ditto.
720 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
722 * config/i386/constraints.md ("C"): Do not depend on TARGET_SSE.
723 * config/i386/predicates.md (nonimm_or_0_operand): Rename
724 from vector_move_operand. Update all uses.
726 2018-10-05 Martin Sebor <msebor@redhat.com>
728 PR tree-optimization/87490
729 * builtins.c (expand_builtin_strnlen): Handle a null data.decl
732 2018-10-05 Richard Biener <rguenther@suse.de>
734 PR tree-optimization/63155
735 * tree-ssa-ccp.c (ccp_propagate::visit_phi): Avoid excess
736 vertical space in dumpfiles.
737 * tree-ssa-propagate.h
738 (ssa_propagation_engine::process_ssa_edge_worklist): Remove.
739 * tree-ssa-propagate.c (cfg_blocks_back): New global.
740 (ssa_edge_worklist_back): Likewise.
741 (curr_order): Likewise.
742 (cfg_blocks_get): Remove abstraction.
743 (cfg_blocks_add): Likewise.
744 (cfg_blocks_empty_p): Likewise.
745 (add_ssa_edge): Add to current or next worklist based on
747 (add_control_edge): Likewise.
748 (ssa_propagation_engine::process_ssa_edge_worklist): Fold
750 (ssa_propagation_engine::ssa_propagate): ... here. Unify
751 iteration from CFG and SSA edge worklist so we process
752 everything in RPO order, prioritizing forward progress
754 (ssa_prop_init): Allocate new worklists, do not dump
756 (ssa_prop_fini): Free new worklists.
758 2018-10-05 Richard Biener <rguenther@suse.de>
760 * tree-core.h (tree_block::abstract_flag): Remove.
761 (tree_block::block_num): Make full 32bits.
762 * tree.def (BLOCK): Remove docs about BLOCK_ABSTRACT.
763 * tree.h (BLOCK_ABSTRACT): Remove.
764 * dwarf2out.c (gen_lexical_block_die): Remove dead code
765 resulting from BLOCK_ABSTRACT being always false.
766 (gen_inlined_subroutine_die): Likewise.
767 (gen_block_die): Likewise.
768 * tree.c (block_ultimate_origin): Likewise.
769 * tree-pretty-print.c (dump_block_node): Remove code dealing
771 * tree-ssa-live.c (dump_scope_block): Likewise.
772 * tree-streamer-in.c (unpack_ts_block_value_fields): Likewise.
773 * tree-streamer-out.c (pack_ts_block_value_fields): Likewise.
775 2018-10-05 Richard Biener <rguenther@suse.de>
777 * config/i386/i386.c (ix86_add_stmt_cost): When scalar cost
778 is asked for initialize mode to the component mode of the
781 2018-10-05 H.J. Lu <hongjiu.lu@intel.com>
784 * config/i386/gnu-user.h (ASM_SPEC): Don't pass -msse2avx to
786 * config/i386/gnu-user64.h (ASM_SPEC): Likewise.
788 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
791 * config/rs6000/rs6000-builtin.def (RS6000_BUILTIN_SET_FPSCR_DRN): Use
793 * config/rs6000/rs6000.md (rs6000_set_fpscr_rn): Require the operand
794 to be DImode. When using mffscrn, force the operand to a register.
796 2018-10-04 Uros Bizjak <ubizjak@gmail.com>
798 * config/i386/i386.md (*fop_<X87MODEF:mode>_2_i387): Macroize insn
799 from *fop_<MODEF:mode>_2_i387 and *fop_xf_2_i387 using
800 X87MODEF mode iterator.
801 (*fop_<X87MODEF:mode>_3_i387): Macroize insn from
802 *fop_<MODEF:mode>_3_i387 and *fop_xf_3_i387 using
803 X87MODEF mode iterator.
805 2018-10-04 Vinay Kumar <vinay.kumar@blackfigtech.com>
807 * doc/invoke.texi (-Wno-prio-ctor-dtor): Document new warning
810 2018-10-04 David Malcolm <dmalcolm@redhat.com>
812 * Makefile.in (OBJS): Add opt-problem.o.
813 * dump-context.h: Include "selftest.h.
814 (selftest::temp_dump_context): New forward decl.
815 (class dump_context): Make friend of class
816 selftest::temp_dump_context.
817 (dump_context::dump_loc_immediate): New decl.
818 (class dump_pretty_printer): Move here from dumpfile.c.
819 (class temp_dump_context): Move to namespace selftest.
820 (temp_dump_context::temp_dump_context): Add param
821 "forcibly_enable_dumping".
822 (selftest::verify_dumped_text):
823 (ASSERT_DUMPED_TEXT_EQ): Move here from dumpfile.c.
824 (selftest::verify_item):
825 (ASSERT_IS_TEXT): Move here from dumpfile.c.
826 (ASSERT_IS_TREE): Likewise.
827 (ASSERT_IS_GIMPLE): Likewise.
828 * dumpfile.c (dump_context::dump_loc): Move immediate dumping
830 (dump_context::dump_loc_immediate): ...this new function.
831 (class dump_pretty_printer): Move to dump-context.h.
832 (dump_switch_p_1): Don't enable MSG_PRIORITY_REEMITTED.
833 (opt_info_switch_p_1): Enable MSG_PRIORITY_REEMITTED.
834 (temp_dump_context::temp_dump_context): Move to "selftest"
835 namespace. Add param "forcibly_enable_dumping", and use it to
836 conditionalize the use of m_pp;
837 (selftest::verify_dumped_text): Make non-static.
838 (ASSERT_DUMPED_TEXT_EQ): Move to dump-context.h.
839 (selftest::verify_item): Make non-static.
840 (ASSERT_IS_TEXT): Move to dump-context.h.
841 (ASSERT_IS_TREE): Likewise.
842 (ASSERT_IS_GIMPLE): Likewise.
843 (selftest::test_capture_of_dump_calls): Pass "true" for new
844 param of temp_dump_context.
845 * dumpfile.h (enum dump_flag): Add MSG_PRIORITY_REEMITTED, adding
846 it to MSG_ALL_PRIORITIES. Update values of TDF_COMPARE_DEBUG and
848 * opt-problem.cc: New file.
849 * opt-problem.h: New file.
850 * optinfo-emit-json.cc
851 (selftest::test_building_json_from_dump_calls): Pass "true" for
852 new param of temp_dump_context.
853 * optinfo.cc (optinfo_kind_to_dump_flag): New function.
854 (optinfo::emit_for_opt_problem): New function.
855 (optinfo::emit): Clarity which emit_item is used.
856 * optinfo.h (optinfo::get_dump_location): New accessor.
857 (optinfo::emit_for_opt_problem): New decl.
858 (optinfo::emit): Make const.
859 * selftest-run-tests.c (selftest::run_tests): Call
860 selftest::opt_problem_cc_tests.
861 * selftest.h (selftest::opt_problem_cc_tests): New decl.
862 * tree-data-ref.c (dr_analyze_innermost): Convert return type from
863 bool to opt_result, converting fprintf messages to
864 opt_result::failure_at calls. Add "stmt" param for use by the
866 (create_data_ref): Pass "stmt" to the dr_analyze_innermost call.
867 (runtime_alias_check_p): Convert return type from bool to
868 opt_result, converting dump_printf calls to
869 opt_result::failure_at, using the statement DDR_A for their
871 (find_data_references_in_stmt): Convert return type from bool to
872 opt_result, converting "return false" to opt_result::failure_at
874 * tree-data-ref.h: Include "opt-problem.h".
875 (dr_analyze_innermost): Convert return type from bool to opt_result,
876 and add a const gimple * param.
877 (find_data_references_in_stmt): Convert return type from bool to
879 (runtime_alias_check_p): Likewise.
880 * tree-predcom.c (find_looparound_phi): Pass "init_stmt" to
881 dr_analyze_innermost.
882 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test):
883 Convert return type from bool to opt_result, adding a message for
884 the PARAM_VECT_MAX_VERSION_FOR_ALIAS_CHECKS zero case.
885 (vect_analyze_data_ref_dependence): Convert return type from bool
886 to opt_result. Change sense of return type from "false"
887 effectively meaning "no problems" to "false" meaning a problem,
888 so that "return false" becomes "return opt_result::success".
889 Convert "return true" calls to opt_result::failure_at, using
890 the location of statement A rather than vect_location.
891 (vect_analyze_data_ref_dependences): Convert return type from bool
893 (verify_data_ref_alignment): Likewise, converting dump_printf_loc
894 calls to opt_result::failure_at, using the stmt location rather
896 (vect_verify_datarefs_alignment): Convert return type from bool
898 (vect_enhance_data_refs_alignment): Likewise. Split local "stat"
899 into multiple more-tightly-scoped copies.
900 (vect_analyze_data_refs_alignment): Convert return type from bool
902 (vect_analyze_data_ref_accesses): Likewise, converting a
903 "return false" to a "return opt_result::failure_at", adding a
905 (vect_prune_runtime_alias_test_list): Convert return type from
906 bool to opt_result, converting dump_printf_loc to
907 opt_result::failure_at. Add a %G to show the pertinent statement,
908 and use the stmt's location rather than vect_location.
909 (vect_find_stmt_data_reference): Convert return type from
910 bool to opt_result, converting dump_printf_loc to
911 opt_result::failure_at, using stmt's location.
912 (vect_analyze_data_refs): Convert return type from bool to
913 opt_result. Convert "return false" to "return
914 opt_result::failure_at", adding messages as needed.
915 * tree-vect-loop.c (vect_determine_vf_for_stmt_1): Convert return
916 type from bool to opt_result.
917 (vect_determine_vf_for_stmt): Likewise.
918 (vect_determine_vectorization_factor): Likewise, converting
919 dump_printf_loc to opt_result::failure_at, using location of phi
920 rather than vect_location.
921 (vect_analyze_loop_form_1): Convert return type from bool to
922 opt_result, converting dump_printf_loc calls, retaining the use of
924 (vect_analyze_loop_form): Convert return type from loop_vec_info
925 to opt_loop_vec_info.
926 (vect_analyze_loop_operations): Convert return type from bool to
927 opt_result, converting dump_printf_loc calls, using the location
928 of phi/stmt rather than vect_location where available. Convert
929 various "return false" to "return opt_result::failure_at" with
930 "unsupported phi" messages.
931 (vect_get_datarefs_in_loop): Convert return type from bool to
932 opt_result. Add a message for the
933 PARAM_LOOP_MAX_DATAREFS_FOR_DATADEPS failure.
934 (vect_analyze_loop_2): Convert return type from bool to
935 opt_result. Ensure "ok" is set to a opt_result::failure_at before
936 each "goto again;", adding new messages where needed.
937 Add "unsupported grouped {store|load}" messages.
938 (vect_analyze_loop): Convert return type from loop_vec_info to
940 * tree-vect-slp.c (vect_analyze_slp): Convert return type from
942 * tree-vect-stmts.c (process_use): Likewise, converting
943 dump_printf_loc call and using stmt location, rather than
945 (vect_mark_stmts_to_be_vectorized): Likeise.
946 (vect_analyze_stmt): Likewise, adding a %G.
947 (vect_get_vector_types_for_stmt): Convert return type from bool to
948 opt_result, converting dump_printf_loc calls and using stmt
949 location, rather than vect_location.
950 (vect_get_mask_type_for_stmt): Convert return type from tree to
951 opt_tree, converting dump_printf_loc calls and using stmt location.
952 * tree-vectorizer.c: Include "opt-problem.h.
953 (try_vectorize_loop_1): Flag "Analyzing loop at" dump message as
954 MSG_PRIORITY_INTERNALS. Convert local "loop_vinfo" from
955 loop_vec_info to opt_loop_vec_info. If if fails, and dumping is
956 enabled, use it to report at the top level "couldn't vectorize
957 loop" followed by the problem.
958 * tree-vectorizer.h (opt_loop_vec_info): New typedef.
959 (vect_mark_stmts_to_be_vectorized): Convert return type from bool
961 (vect_analyze_stmt): Likewise.
962 (vect_get_vector_types_for_stmt): Likewise.
963 (tree vect_get_mask_type_for_stmt): Likewise.
964 (vect_analyze_data_ref_dependences): Likewise.
965 (vect_enhance_data_refs_alignment): Likewise.
966 (vect_analyze_data_refs_alignment): Likewise.
967 (vect_verify_datarefs_alignment): Likewise.
968 (vect_analyze_data_ref_accesses): Likewise.
969 (vect_prune_runtime_alias_test_list): Likewise.
970 (vect_find_stmt_data_reference): Likewise.
971 (vect_analyze_data_refs): Likewise.
972 (vect_analyze_loop): Convert return type from loop_vec_info to
974 (vect_analyze_loop_form): Likewise.
975 (vect_analyze_slp): Convert return type from bool to opt_result.
977 2018-10-04 David Malcolm <dmalcolm@redhat.com>
979 * doc/invoke.texi (-fopt-info): Document new "internals"
981 * dump-context.h (dump_context::apply_dump_filter_p): New decl.
982 * dumpfile.c (dump_options): Update for renaming of MSG_ALL to
984 (optinfo_verbosity_options): Add "internals".
985 (kind_as_string): Update for renaming of MSG_ALL to MSG_ALL_KINDS.
986 (dump_context::apply_dump_filter_p): New member function.
987 (dump_context::dump_loc): Use apply_dump_filter_p rather than
988 explicitly masking the dump_kind.
989 (dump_context::begin_scope): Increment the scope depth first. Use
990 apply_dump_filter_p rather than explicitly masking the dump_kind.
991 (dump_context::emit_item): Use apply_dump_filter_p rather than
992 explicitly masking the dump_kind.
993 (dump_dec): Likewise.
994 (dump_hex): Likewise.
995 (dump_switch_p_1): Default to MSG_ALL_PRIORITIES.
996 (opt_info_switch_p_1): Default to MSG_PRIORITY_USER_FACING.
997 (opt_info_switch_p): Update handling of default
998 MSG_OPTIMIZED_LOCATIONS to cope with default of
999 MSG_PRIORITY_USER_FACING.
1000 (dump_basic_block): Use apply_dump_filter_p rather than explicitly
1001 masking the dump_kind.
1002 (selftest::test_capture_of_dump_calls): Update test_dump_context
1003 instances to use MSG_ALL_KINDS | MSG_PRIORITY_USER_FACING rather
1004 than MSG_ALL. Generalize scope test to be run at all four
1005 combinations of with/without MSG_PRIORITY_USER_FACING and
1006 MSG_PRIORITY_INTERNALS, adding examples of explicit priority
1007 for each of the two values.
1008 * dumpfile.h (enum dump_flag): Add comment about the MSG_* flags.
1009 Rename MSG_ALL to MSG_ALL_KINDS. Add MSG_PRIORITY_USER_FACING,
1010 MSG_PRIORITY_INTERNALS, and MSG_ALL_PRIORITIES, updating the
1011 values for TDF_COMPARE_DEBUG and TDF_ALL_VALUES.
1012 (AUTO_DUMP_SCOPE): Add a note to the comment about the interaction
1013 with MSG_PRIORITY_*.
1014 * tree-vect-loop-manip.c (vect_loop_versioning): Mark versioning
1015 dump messages as MSG_PRIORITY_USER_FACING.
1016 * tree-vectorizer.h (DUMP_VECT_SCOPE): Add a note to the comment
1017 about the interaction with MSG_PRIORITY_*.
1019 2018-10-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1021 * varasm.c (output_constant): Add new parameter merge_strings.
1022 Make strings properly zero terminated in merge string sections.
1023 (mergeable_string_section): Don't fail if the last char is non-zero.
1024 (assemble_variable_contents): Handle merge string sections.
1025 (assemble_variable): Likewise.
1026 (assemble_constant_contents): Likewise.
1027 (output_constant_def_contents): Likewise.
1028 (output_constructor_array_range,
1029 output_constructor_regular_field): Adjust call to output_constant.
1030 (output_object_block): Adjust call to assemble_constant_contents
1031 and assemble_variable_contents.
1033 2018-10-04 Martin Liska <mliska@suse.cz>
1036 * cgraphunit.c (process_function_and_variable_attributes):
1037 Warn about a function with alias attribute and a body.
1039 2018-10-04 Martin Liska <mliska@suse.cz>
1042 * multiple_target.c (redirect_to_specific_clone): New function.
1043 (ipa_target_clone): Use it.
1044 * tree-inline.c: Fix comment.
1046 2018-10-04 David Malcolm <dmalcolm@redhat.com>
1048 * dumpfile.c (gcc::dump_manager::dump_manager): Initialize new
1050 (gcc::dump_manager::~dump_manager): Free m_optinfo_filename.
1051 (gcc::dump_manager::register_pass): New member function, adapted
1052 from loop body in gcc::pass_manager::register_pass, adding a
1053 call to update_dfi_for_opt_info.
1054 (gcc::dump_manager::opt_info_enable_passes): Store the
1055 -fopt-info options into the new fields. Move the loop
1057 (gcc::dump_manager::update_dfi_for_opt_info): ...this new member
1059 * dumpfile.h (struct opt_pass): New forward decl.
1060 (gcc::dump_manager::register_pass): New decl.
1061 (gcc::dump_manager::update_dfi_for_opt_info): New decl.
1062 (class gcc::dump_manager): Add fields "m_optgroup_flags",
1063 "m_optinfo_flags", and "m_optinfo_filename".
1064 * passes.c (gcc::pass_manager::register_pass): Move all of the
1065 dump-handling code to gcc::dump_manager::register_pass.
1067 2018-10-04 Peter Bergner <bergner@linux.ibm.com>
1069 PR rtl-optimization/87466
1070 * target.def (setjmp_preserves_nonvolatile_regs_p): New target hook.
1071 * doc/tm.texi.in (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P): New hook.
1072 * doc/tm.texi: Regenerate.
1073 * ira-lives.c (process_bb_node_lives): Use the new target hook.
1074 * lra-lives.c (process_bb_lives): Likewise.
1075 * config/rs6000/rs6000.c (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P):
1078 2018-10-04 Tamar Christina <tamar.christina@arm.com>
1080 * params.c (add_params): Fix initialization.
1082 2018-10-04 Martin Liska <mliska@suse.cz>
1084 PR gcov-profile/84107
1085 * tree-profile.c (init_ic_make_global_vars):
1086 Remove ic_void_ptr_var and ic_gcov_type_ptr_var.
1087 Come up with new ic_tuple* variables. Emit
1088 __gcov_indirect_call{,_topn} variables.
1089 (gimple_gen_ic_profiler): Access the variable
1091 (gimple_gen_ic_func_profiler): Access
1092 __gcov_indirect_call.callee field.
1093 (gimple_init_gcov_profiler): Use ptr_type_node.
1094 * value-prof.c (gimple_ic): Use ptr_type_node.
1096 2018-10-04 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1098 PR tree-optimization/85787
1099 * ipa-pure-const.c (malloc_candidate_p_1): Move most of malloc_candidate_p
1100 into this function and add support for detecting multiple phis.
1101 (DUMP_AND_RETURN): Move from malloc_candidate_p into top-level macro.
1103 2018-10-04 Martin Liska <mliska@suse.cz>
1106 * ipa-inline.c (inline_to_all_callers_1):
1107 Call ultimate_alias_target for node being inlined.
1109 2018-10-03 Jeff Law <law@redhat.com>
1111 * gimple-ssa-sprintf.c (format_string): Do not hardcode size of
1113 * tree.c (get_typenode_from_name): Moved from fortran/trans-types.c.
1114 * tree.h (get_typenode_from_name): Prototype.
1116 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
1118 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
1119 Change operand 2 predicate to nonimmediate_operand.
1120 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
1122 2018-10-03 Martin Sebor <msebor@redhat.com>
1123 Jeff Law <law@redhat.com>
1125 * gimple-ssa-sprintf.c (struct fmtresult): Add new member and
1127 (get_string_length): Detect unterminated arrays.
1128 (format_string): Same.
1129 (format_directive): Warn about unterminated arrays.
1130 (handle_gimple_call): Mark statements with no_warning as needed.
1132 2018-10-03 Jim Wilson <jimw@sifive.com>
1134 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): For ABI_ILP32E,
1135 also define __riscv_abi_rve. Delete trailing white space.
1137 2018-10-03 Paul Koning <ni1d@arrl.net>
1139 Enable LRA register allocator for PDP11.
1140 * config/pdp11/constraints.md (Q): Use define_memory_constraint.
1143 * config/pdp11/pdp11.c (pdp11_lra_p): New function.
1144 * config/pdp11/pdp11.opt (-mlra): New option.
1145 * doc/invoke.texi (PDP-11 Options): Document -mlra.
1147 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
1149 * config/i386/i386.md (*<absneg:code>extendsfdf2): Remove.
1150 (*<absneg:code>extend<mode>xf2): Ditto.
1152 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
1154 PR tree-optimization/87415
1155 * tree-vrp.c (set_value_range_with_overflow): Special case one bit
1158 2018-10-02 Jeff Law <law@redhat.com>
1160 * gimple-fold.c (get_range_strlen): Only set *nonstr when
1161 an unterminated string is discovered. Bubble up range
1162 even for unterminated strings.
1163 (gimple_fold_builtin_strlen): Do not fold if get_range_strlen
1164 indicates the string was not terminated via NONSTR.
1166 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
1168 * tree-vrp.c (extract_range_from_unary_expr): Special case all
1169 pointer conversions.
1170 Do not do anything special for anti-ranges.
1172 2018-10-03 Jérôme Lambourg <lambourg@adacore.com>
1174 * config/arm/vxworks.h (ARM_TARGET2_DWARF_FORMAT): Adjust to
1175 DW_EH_PE_pcrel | DW_EH_PE_indirect for RTPs.
1177 2018-10-03 Martin Liska <mliska@suse.cz>
1179 PR gcov-profile/86109
1180 * coverage.c (coverage_begin_function): Do not
1181 mark lambdas as artificial.
1182 * tree-core.h (struct GTY): Remove tm_clone_flag
1183 and introduce new lambda_function.
1184 * tree.h (DECL_LAMBDA_FUNCTION): New macro.
1186 2018-10-02 Aaron Sawdey <acsawdey@linux.ibm.com>
1189 * config/rs6000/rs6000-string.c (expand_strn_compare): Check that both
1190 P8_VECTOR and VSX are enabled.
1192 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
1194 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
1195 0x3907 as CPU model number.
1197 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
1199 * common/config/s390/s390-common.c: Rename PF_ARCH12 to PF_Z14.
1200 * config/s390/s390.h (enum processor_flags): Rename PF_ARCH12 to
1201 PF_Z14. Rename TARGET_CPU_ARCH12 to TARGET_CPU_Z14,
1202 TARGET_CPU_ARCH12_P to TARGET_CPU_Z14_P, TARGET_ARCH12 to
1203 TARGET_Z14, and TARGET_ARCH12_P to TARGET_Z14_P.
1204 * config/s390/s390.md: Likewise. Rename also the cpu attribute
1205 value from arch12 to z14.
1207 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
1209 * config/i386/i386.md (fxam<mode>2_i387_with_temp): Remove.
1211 (isinf<mode>2): Ditto.
1213 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
1215 * config/i386/i386.c (ix86_emit_i387_round): Extend op1 to XFmode
1216 before emitting fxam. Perform calculations in XFmode.
1218 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
1220 * match.pd (((X /[ex] A) +- B) * A): New transformation.
1222 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
1225 * fold-const.c (fold_plusminus_mult_expr): Handle complex and vectors.
1226 * tree.c (signed_or_unsigned_type_for): Handle complex.
1228 2018-10-02 Jeff Law <law@redhat.com>
1230 * gimple-fold.c (get_range_strlen): Remove dead code.
1232 2018-10-02 Martin Sebor <msebor@redhat.com>
1233 Jeff Law <law@redhat.com>
1235 * builtins.c (unterminated_array): Add new arguments.
1236 If argument is not terminated, bubble up size and exact
1238 (expand_builtin_strnlen): Detect, avoid expanding
1239 and diagnose unterminated arrays.
1240 (c_strlen): Fill in offset of start of unterminated strings.
1241 * builtins.h (unterminated_array): Update prototype.
1243 2018-10-02 Richard Biener <rguenther@suse.de>
1245 * config/i386/sse.md (reduc_plus_scal_v4df): Avoid the use
1246 of haddv4df, first reduce to SSE width and exploit the fact
1247 that we only need element zero with the reduction result.
1248 (reduc_plus_scal_v2df): Likewise.
1250 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
1252 * dojump.h (do_jump): Delete.
1253 (do_jump_1): Likewise.
1254 (split_comparison): Move around.
1255 * dojump.c (do_jump): Make static.
1256 (do_jump_1): Likewise.
1257 (jumpifnot): Move around.
1258 (jumpifnot_1): Likewise.
1260 (jumpif_1): Likewise.
1261 * expr.c (expand_expr_real_1): Call jumpif[not] instead of do_jump.
1263 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
1265 * reorg.c (make_return_insns): Use emit_copy_of_insn_after for the
1266 insns in the delay slot and add_insn_after for the jump insn.
1268 2018-10-02 Richard Biener <rguenther@suse.de>
1270 * tree-inline.c (expand_call_inline): Use the location of
1271 the callee declaration for the inline-entry marker.
1272 * final.c (notice_source_line): Remove special-casing of
1273 NOTE_INSN_INLINE_ENTRY.
1275 2018-10-01 Carl Love <cel@us.ibm.com>
1278 * config/rs6000/rs6000-builtin.def (__builtin_mffsl): New.
1279 (__builtin_mtfsb0): New.
1280 (__builtin_mtfsb1): New.
1281 ( __builtin_set_fpscr_rn): New.
1282 (__builtin_set_fpscr_drn): New.
1283 * config/rs6000/rs6000.c (rs6000_expand_mtfsb_builtin): Add.
1284 (rs6000_expand_set_fpscr_rn_builtin): Add.
1285 (rs6000_expand_set_fpscr_drn_builtin): Add.
1286 (rs6000_expand_builtin): Add case statement entries for
1287 RS6000_BUILTIN_MTFSB0, RS6000_BUILTIN_MTFSB1,
1288 RS6000_BUILTIN_SET_FPSCR_RN, RS6000_BUILTIN_SET_FPSCR_DRN,
1289 RS6000_BUILTIN_MFFSL.
1290 (rs6000_init_builtins): Add ftype initialization and def_builtin
1291 calls for __builtin_mffsl, __builtin_mtfsb0, __builtin_mtfsb1,
1292 __builtin_set_fpscr_rn, __builtin_set_fpscr_drn.
1293 * config/rs6000.md (rs6000_mtfsb0, rs6000_mtfsb1, rs6000_mffscrn,
1294 rs6000_mffscdrn): Add define_insn.
1295 (rs6000_set_fpscr_rn, rs6000_set_fpscr_drn): Add define_expand.
1296 * doc/extend.texi: Add documentation for the builtins.
1298 2018-10-01 Richard Biener <rguenther@suse.de>
1300 PR tree-optimization/87465
1301 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Fix typo
1302 causing branch miscounts.
1304 2018-10-01 Tamar Christina <tamar.christina@arm.com>
1306 * common/config/aarch64/aarch64-common.c (TARGET_OPTION_DEFAULT_PARAM,
1307 aarch64_option_default_param): New.
1308 (params.h): Include.
1309 (TARGET_OPTION_VALIDATE_PARAM, aarch64_option_validate_param): New.
1310 * config/aarch64/aarch64.c (aarch64_override_options_internal): Simplify
1311 stack-clash protection validation code.
1313 2018-10-01 Tamar Christina <tamar.christina@arm.com>
1315 * params.c (validate_param): New.
1316 (add_params): Use it.
1317 (set_param_value): Refactor param validation into validate_param.
1318 (diagnostic.h): Include.
1319 * diagnostic.h (diagnostic_ready_p): New.
1321 2018-10-01 Tamar Christina <tamar.christina@arm.com>
1323 * params.c (set_param_value):
1324 Add index of parameter being validated.
1325 * common/common-target.def (option_validate_param): New.
1326 * common/common-targhooks.h (default_option_validate_param): New.
1327 * common/common-targhooks.c (default_option_validate_param): New.
1328 * doc/tm.texi.in (TARGET_OPTION_VALIDATE_PARAM): New.
1329 * doc/tm.texi: Regenerate.
1331 2018-10-01 Tamar Christina <tamar.christina@arm.com>
1334 * config/aarch64/aarch64.c (aarch64_override_options_internal):
1335 Add validation for stack-clash parameters and set defaults.
1337 2018-10-01 Tamar Christina <tamar.christina@arm.com>
1340 * configure.ac: Add stack-clash-protection-guard-size.
1341 * doc/install.texi: Document it.
1342 * config.in (DEFAULT_STK_CLASH_GUARD_SIZE): New.
1343 * params.def: Update comment for guard-size.
1344 (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE,
1345 PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Update description.
1346 * configure: Regenerate.
1348 2018-10-01 Tamar Christina <tamar.christina@arm.com>
1351 * config/aarch64/aarch64.h (STACK_CLASH_MIN_BYTES_OUTGOING_ARGS,
1352 STACK_DYNAMIC_OFFSET): New.
1353 * config/aarch64/aarch64.c (aarch64_layout_frame):
1354 Update outgoing args size.
1355 (aarch64_stack_clash_protection_alloca_probe_range,
1356 TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
1358 2018-10-01 Tamar Christina <tamar.christina@arm.com>
1361 * explow.c (anti_adjust_stack_and_probe_stack_clash): Support custom
1363 * target.def (stack_clash_protection_alloca_probe_range): New.
1364 (stack_clash_protection_final_dynamic_probe): Remove.
1365 * targhooks.h (default_stack_clash_protection_alloca_probe_range) New.
1366 (default_stack_clash_protection_final_dynamic_probe): Remove.
1367 * targhooks.c: Likewise.
1368 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
1369 (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE): Remove.
1370 * doc/tm.texi: Regenerate.
1372 2018-10-01 Tamar Christina <tamar.christina@arm.com>
1375 * config/aarch64/aarch64-protos.h (aarch64_output_probe_sve_stack_clash): New.
1376 * config/aarch64/aarch64.c (aarch64_output_probe_sve_stack_clash,
1377 aarch64_clamp_to_uimm12_shift): New.
1378 (aarch64_allocate_and_probe_stack_space): Add SVE specific section.
1379 * config/aarch64/aarch64.md (probe_sve_stack_clash): New.
1381 2018-10-01 Tamar Christina <tamar.christina@arm.com>
1384 * config/aarch64/aarch64.c (aarch64_layout_frame): Add assert.
1386 2018-10-01 Jeff Law <law@redhat.com>
1387 Richard Sandiford <richard.sandiford@linaro.org>
1388 Tamar Christina <tamar.christina@arm.com>
1391 * config/aarch64/aarch64.md
1392 (probe_stack_range): Add k (SP) constraint.
1393 * config/aarch64/aarch64.h (STACK_CLASH_CALLER_GUARD,
1394 STACK_CLASH_MAX_UNROLL_PAGES): New.
1395 * config/aarch64/aarch64.c (aarch64_output_probe_stack_range): Emit
1396 stack probes for stack clash.
1397 (aarch64_allocate_and_probe_stack_space): New.
1398 (aarch64_expand_prologue): Use it.
1399 (aarch64_expand_epilogue): Likewise and update IP regs re-use criteria.
1400 (aarch64_sub_sp): Add emit_move_imm optional param.
1402 2018-10-01 MCC CS <deswurstes@users.noreply.github.com>
1404 PR tree-optimization/87261
1405 * match.pd: Remove trailing whitespace.
1406 Add (x & y) | ~(x | y) -> ~(x ^ y),
1407 (~x | y) ^ (x ^ y) -> x | ~y and (x ^ y) | ~(x | y) -> ~(x & y)
1409 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
1411 * config/arc/arc.md (*add_n): Clean up pattern, update instruction
1413 (ashlsi3_insn): Update instruction constraints.
1414 (ashrsi3_insn): Likewise.
1415 (rotrsi3): Likewise.
1416 (add_shift): Likewise.
1417 * config/arc/constraints.md (Csz): New 32 bit constraint. It
1418 avoids placing in the limm field small constants which, otherwise,
1419 could end into a small instruction.
1421 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
1423 * config/arc/arc.md (maddsidi4_split): Don't use dmac if the
1424 destination register is not odd-even.
1425 (umaddsidi4_split): Likewise.
1427 2018-10-01 Richard Biener <rguenther@suse.de>
1429 * tree-inline.c (expand_call_inline): Store origin of fn
1430 in BLOCK_ABSTRACT_ORIGIN for the inline BLOCK.
1431 * tree.c (block_ultimate_origin): Simplify and do some
1434 2018-09-30 Uros Bizjak <ubizjak@gmail.com>
1436 * config/i386/mmx.md (EMMS): New int iterator.
1437 (emms): New int attribute.
1438 (mmx_<emms>): Macroize insn from *mmx_emms and *mmx_femms using
1439 EMMS int iterator. Explicitly declare clobbers.
1440 (mmx_emms): Remove expander.
1442 * config/i386/predicates.md (emms_operation): Remove predicate.
1443 (vzeroall_pattern): New predicate.
1444 (vzeroupper_pattern): Rename from vzeroupper_operation.
1445 * config/i386/i386.c (ix86_avx_u128_mode_after): Use
1446 vzeroupper_pattern and vzeroall_pattern predicates.
1448 2018-09-30 Peter Bergner <bergner@linux.ibm.com>
1450 PR rtl-optimization/86939
1451 * ira-lives.c (make_hard_regno_born): Rename from this...
1452 (make_hard_regno_live): ... to this. Remove update to conflict
1453 information. Update function comment.
1454 (make_hard_regno_dead): Add conflict information update. Update
1456 (make_object_born): Rename from this...
1457 (make_object_live): ... to this. Remove update to conflict information.
1458 Update function comment.
1459 (make_object_dead): Add conflict information update. Update function
1461 (mark_pseudo_regno_live): Call make_object_live.
1462 (mark_pseudo_regno_subword_live): Likewise.
1463 (mark_hard_reg_dead): Update function comment.
1464 (mark_hard_reg_live): Call make_hard_regno_live.
1465 (process_bb_node_lives): Likewise.
1466 * lra-lives.c (make_hard_regno_born): Rename from this...
1467 (make_hard_regno_live): ... to this. Remove update to conflict
1468 information. Remove now uneeded check_pic_pseudo_p argument.
1469 Update function comment.
1470 (make_hard_regno_dead): Add check_pic_pseudo_p argument and add update
1471 to conflict information. Update function comment.
1472 (mark_pseudo_live): Remove update to conflict information. Update
1474 (mark_pseudo_dead): Add conflict information update.
1475 (mark_regno_live): Call make_hard_regno_live.
1476 (mark_regno_dead): Call make_hard_regno_dead with new arguement.
1477 (process_bb_lives): Call make_hard_regno_live and make_hard_regno_dead.
1479 2018-09-29 H.J. Lu <hongjiu.lu@intel.com>
1482 * config/i386/i386.c (construct_container): Use TImode for
1483 BLKmode values in 2 integer registers.
1485 2018-09-29 Jeff Law <law@redhat.com>
1487 * builtins.c (unterminated_array): Pass in c_strlen_data * to
1488 c_strlen rather than just a tree *.
1489 (c_strlen): Change NONSTR argument to a c_strlen_data pointer.
1490 Update recursive calls appropriately. If caller did not provide a
1491 suitable data pointer, create a local one. When a non-terminated
1492 string is discovered, bubble up information about the string via the
1493 c_strlen_data object.
1494 * builtins.h (c_strlen): Update prototype.
1495 (c_strlen_data): New structure.
1496 * gimple-fold.c (get_range_strlen): Update calls to c_strlen.
1497 For a type 2 call, if c_strlen indicates a non-terminated string
1498 use the length of the non-terminated string.
1499 (gimple_fold_builtin_stpcpy): Update calls to c_strlen.
1501 2018-09-29 Jakub Jelinek <jakub@redhat.com>
1504 * config/i386/avx512fintrin.h (_mm512_abs_pd, _mm512_mask_abs_pd): Use
1505 __m512d type for __A argument rather than __m512.
1507 2018-09-28 John David Anglin <danglin@gcc.gnu.org>
1509 * match.pd (simple_comparison): Don't optimize if either operand is
1510 a function pointer when target needs function pointer canonicalization.
1512 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
1514 * config/rs6000/driver-rs6000.c (asm_names): Adjust the entries for
1515 power5 .. power9 to remove indirection.
1516 * config/rs6000/rs6000.h (ASM_CPU_POWER5_SPEC, ASM_CPU_POWER6_SPEC,
1517 ASM_CPU_POWER7_SPEC, ASM_CPU_POWER8_SPEC, ASM_CPU_POWER9_SPEC,
1518 ASM_CPU_476_SPEC): Delete.
1519 (ASM_CPU_SPEC): Adjust.
1520 (EXTRA_SPECS): Delete asm_cpu_power5, asm_cpu_power6, asm_cpu_power7,
1521 asm_cpu_power8, asm_cpu_power9, asm_cpu_476.
1523 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
1525 * config.in: Delete HAVE_AS_DCI.
1526 * config/powerpcspe/powerpcspe.h: Treat HAVE_AS_DCI as always true.
1527 * config/rs6000/rs6000.h: Ditto.
1528 * configure.ac: Delete HAVE_AS_DCI.
1529 * configure: Regenerate.
1531 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
1533 * config.in (HAVE_AS_LWSYNC): Delete.
1534 * config/powerpcspe/powerpcspe.h (TARGET_LWSYNC_INSTRUCTION): Delete.
1535 * config/powerpcspe/sync.md (*lwsync): Always generate lwsync, never
1537 * config/rs6000/rs6000.h (TARGET_LWSYNC_INSTRUCTION): Delete.
1538 * config/rs6000/sync.md (*lwsync): Always generate lwsync, never do it
1540 * configure.ac: Delete HAVE_AS_LWSYNC.
1541 * configure: Regenerate.
1543 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
1544 Pierre-Marie de Rodat <derodat@adacore.com>
1546 * calls.c (expand_call): Try to do a tail call for thunks at -O0 too.
1547 * cgraph.h (struct cgraph_thunk_info): Add indirect_offset.
1548 (cgraph_node::create_thunk): Add indirect_offset parameter.
1549 (thunk_adjust): Likewise.
1550 * cgraph.c (cgraph_node::create_thunk): Add indirect_offset parameter
1551 and initialize the corresponding field with it.
1552 (cgraph_node::dump): Dump indirect_offset field.
1553 * cgraphclones.c (duplicate_thunk_for_node): Deal with indirect_offset.
1554 * cgraphunit.c (cgraph_node::analyze): Be prepared for external thunks.
1555 (thunk_adjust): Add indirect_offset parameter and deal with it.
1556 (cgraph_node::expand_thunk): Deal with the indirect_offset field and
1557 pass it to thunk_adjust. Do not call the target hook if it's non-zero
1558 or if the thunk is external or local. Fix formatting. Do not chain
1559 the RESULT_DECL to BLOCK_VARS. Pass the static chain to the target,
1560 if any, in the GIMPLE representation.
1561 * ipa-icf.c (sem_function::equals_wpa): Deal with indirect_offset.
1562 * lto-cgraph.c (lto_output_node): Write indirect_offset field.
1563 (input_node): Read indirect_offset field.
1564 * tree-inline.c (expand_call_inline): Pass indirect_offset field in the
1565 call to thunk_adjust.
1566 * tree-nested.c (struct nesting_info): Add thunk_p field.
1567 (create_nesting_tree): Set it.
1568 (convert_all_function_calls): Copy static chain from targets to thunks.
1569 (finalize_nesting_tree_1): Return early for thunks.
1570 (unnest_nesting_tree_1): Do not finalize thunks.
1571 (gimplify_all_functions): Do not gimplify thunks.
1573 2018-09-28 David Malcolm <dmalcolm@redhat.com>
1575 * opt-suggestions.c (option_proposer::build_option_suggestions):
1576 Release "option_values".
1578 2018-09-28 David Malcolm <dmalcolm@redhat.com>
1580 * coverage.c (get_coverage_counts): Convert problem-reporting dump
1581 messages from MSG_OPTIMIZED_LOCATIONS to MSG_MISSED_OPTIMIZATION.
1582 * dumpfile.c (kind_as_string): New function.
1583 (dump_loc): Rather than a hardcoded prefix of "note: ", use
1584 kind_as_string to vary the prefix based on dump_kind.
1585 (selftest::test_capture_of_dump_calls): Update for above.
1587 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
1589 * config/i386/i386.h (SSE_REGNO): Fix check for FIRST_REX_SSE_REG.
1590 (GET_SSE_REGNO): Rename from SSE_REGNO. Update all uses for rename.
1592 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
1594 * config/i386/i386.h (CC_REGNO): Remove FPSR_REG.
1595 * config/i386/i386.c (ix86_fixed_condition_code_regs): Use
1596 INVALID_REGNUM instead of FPSR_REG.
1597 (ix86_md_asm_adjust): Do not clobber FPSR_REG.
1598 * config/i386/i386.md: Update comment of FP compares.
1599 (fldenv): Do not clobber FPSR_REG.
1601 2018-09-28 Richard Biener <rguenther@suse.de>
1603 * tree.h (BLOCK_ORIGIN): New.
1604 * omp-expand.c (grid_expand_target_grid_body): Assign
1605 BLOCK_ORIGIN to BLOCK_ABSTRACT_ORIGIN.
1606 * tree-inline.c (remap_block): Likewise.
1607 * auto-profile.c (get_function_decl_from_block): Simplify
1608 by eliding the BLOCK_ABSTRACT_ORIGIN chasing.
1609 * langhooks.c (lhd_print_error_function): Likewise.
1610 * optinfo-emit-json.cc (optrecord_json_writer::inlining_chain_to):
1612 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
1613 * tree.c (block_nonartificial_location): Likewise.
1614 (block_ultimate_origin): Likewise.
1615 * tree-pretty-print.c (percent_K_format): Likewise. Remove
1616 no longer needed LTO case.
1618 2018-09-28 Andrew Stubbs <ams@codesourcery.com>
1619 Jan Hubicka <jh@suse.cz>
1620 Martin Jambor <mjambor@suse.cz>
1622 * simplify-rtx.c (simplify_merge_mask): New function.
1623 (simplify_ternary_operation): Use it, also see if VEC_MERGEs with the
1624 same masks are used in op1 or op2.
1625 (test_vec_merge): New function.
1626 (test_vector_ops): Call test_vec_merge.
1628 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
1630 * config/sparc/sparc-protos.h (sparc_branch_cost): Declare.
1631 * config/sparc/sparc.h (BRANCH_COST): Call sparc_branch_cost.
1632 * config/sparc/sparc.c (struct processor_costs): Add branch_cost field.
1633 (cypress_costs): Set it.
1634 (supersparc_costs): Likewise.
1635 (hypersparc_costs): Likewise.
1636 (leon_cost): Likewise.
1637 (leon3_costs): Likewise.
1638 (sparclet_costs): Likewise.
1639 (ultrasparc_costs): Likewise.
1640 (ultrasparc_costs): Likewise.
1641 (niagara_costs): Likewise.
1642 (niagara2_costs): Likewise.
1643 (niagara3_costs): Likewise.
1644 (niagara4_costs): Likewise.
1645 (niagara7_costs): Likewise.
1646 (m8_costs): Likewise.
1647 (TARGET_CAN_FOLLOW_JUMP): Define.
1648 (pass_work_around_errata::gate): Minor tweak.
1649 (sparc_option_override): Remove MASK_FSMULD mask for V7 processors.
1650 Do not set both MASK_VIS4 and MASK_VIS4B for M8 processor.
1651 Automaitcally clear MASK_FSMULD mask for V7 processors.
1652 (sparc_can_follow_jump): New static function.
1653 (output_ubranch): Deal with CROSSING_JUMP_P.
1654 (sparc_use_sched_lookahead): Rewrite using switch statement.
1655 (sparc_issue_rate): Reorder.
1656 (sparc_branch_cost): New function.
1658 2018-09-27 Martin Sebor <msebor@redhat.com>
1660 * tree.h (tree_to_shwi): Add attribute nonnull and pure.
1661 (tree_to_poly_int64, tree_to_uhwi, tree_to_poly_uint64): Same.
1662 (int_fits_type_p): Same.
1664 2018-09-27 Uros Bizjak <ubizjak@gmail.com>
1666 * config/i386/i386.md (FPCR_REG): Remove.
1667 (UNSPEC_FLDCW): Remove.
1668 (x86_fnstcw_1): Use (const_int 0) instead of FPCR_REG.
1669 (x86_fldcw_1): Remove insn pattern.
1670 (fnstenv): Do not clobber FPCR_REG.
1672 * config/i386/i386.h (FIXED_REGISTERS) Remove fpsr register.
1673 (CALL_USED_REGISTERS): Ditto.
1674 (REG_ALLOC_ORDER): Ditto.
1675 (REG_CLASS_CONTENTS): Ditto.
1676 (HI_REGISTER_NAMES): Ditto.
1677 (ADDITIONAL_REGISTER_NAMES): Use defines instead
1678 of numerical constants.
1679 * config/i386/i386.c (regclass_map): Remove fpsr register.
1680 (dbx_register_map): Ditto.
1681 (dbx64_register_map): Ditto.
1682 (svr4_dbx_register_map): Ditto.
1683 (print_reg): Do not handle FPCR_REG.
1685 2018-09-27 Segher Boessenkool <segher@kernel.crashing.org>
1688 * config.in (HAVE_AS_CMPB, HAVE_AS_DFP, HAVE_AS_FPRND, HAVE_AS_MFPGPR,
1689 HAVE_AS_POPCNTB, HAVE_AS_POPCNTD, HAVE_AS_POWER8, HAVE_AS_POWER9):
1690 Delete, always treat as true.
1691 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
1692 Ditto. Simplify remaining code.
1693 * config/powerpcspe/powerpcspe.h: Ditto.
1694 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ditto.
1695 Simplify remaining code.
1696 (rs6000_expand_builtin): Ditto.
1697 * config/rs6000/rs6000.h: Ditto.
1698 * configure.ac: Ditto.
1699 * configure: Regenerate.
1701 2018-09-27 Martin Liska <mliska@suse.cz>
1703 * coverage.c (get_coverage_counts): Revert the formatting
1704 of missing profile opt info.
1706 2018-09-27 Richard Biener <rguenther@suse.de>
1710 * dwarf2out.c (set_block_origin_self): Do not mark outermost
1711 block as we do not output that.
1712 (gen_inlined_subroutine_die): Elide the originally outermost
1713 block, matching what we do for concrete instances.
1714 (decls_for_scope): Add parameter specifying whether to recurse
1717 2018-09-27 Andrew Stubbs <ams@codesourcery.com>
1718 Tom de Vries <tom@codesourcery.com>
1722 * expmed.c (emit_cstore): Fix handling of result_mode == BImode and
1723 STORE_FLAG_VALUE == 1.
1725 2018-09-27 Andreas Krebbel <krebbel@linux.ibm.com>
1727 * config/s390/s390.md (PPA_TX_ABORT, PPA_OOO_BARRIER): New
1728 constant definitions.
1729 ("tx_assist"): Replace magic number with PPA_TX_ABORT.
1730 ("*ppa"): Enable pattern also for -march=zEC12 -mno-htm.
1731 ("speculation_barrier"): New expander definition.
1733 2018-09-26 Indu Bhagat <indu.bhagat@oracle.com>
1735 PR gcov-profile/86957
1736 * common.opt: New warning option -Wmissing-profile.
1737 * coverage.c (get_coverage_counts): Add warning for missing .gcda file.
1738 * doc/invoke.texi: Document -Wmissing-profile.
1740 2018-09-26 Jim Wilson <jimw@sifive.com>
1742 * config/riscv/riscv.md (subsi3_extended2): Add J constraint.
1743 (negdi2, negsi2, negsi2_extended, negsi2_extended2): New.
1745 2018-09-26 Martin Sebor <msebor@redhat.com>
1747 * tree.c (zerop): Change return type to bool.
1748 (integer_zerop, integer_onep, integer_each_onep): Same.
1749 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
1750 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
1751 (real_onep, real_minus_onep, chain_index): Same.
1752 (print_type_hash_statistics, type_list_equal): Same.
1753 * tree.h (zerop): Same.
1754 (zerop, integer_zerop, integer_onep, integer_each_onep): Same.
1755 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
1756 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
1757 (real_onep, real_minus_onep, chain_index): Same.
1758 (print_type_hash_statistics, type_list_equal): Same.
1760 2018-09-26 Jim Wilson <jimw@sifive.com>
1762 * config/riscv/riscv.h (FUNCTION_ARG_REGNO_P): Fix comment.
1764 2018-09-26 Jakub Jelinek <jakub@redhat.com>
1767 * config/i386/i386.c: Include debug.h and dwarf2out.h.
1768 (output_indirect_thunk): Emit DW_CFA_def_cfa_offset after the
1771 2018-09-25 Andrew Stubbs <ams@codesourcery.com>
1773 * builtins.c (get_builtin_sync_mem): Force address mode conversion.
1775 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
1777 * config/i386/i386.h (enum reg_class): Remove FP_TOP_SSE_REGS
1778 and FP_SECOND_SSE_REGS.
1779 (REG_CLASS_NAMES): Ditto.
1780 (REG_CLASS_CONTENTS): Ditto.
1781 * config/i386/i386.c (ix86_preferred_reload_class) Do not handle
1782 FP_TOP_SSE_REGS and FP_SECOND_SSE_REGS classes.
1783 (ix86_preferred_output_reload_class): Ditto.
1784 * config/i386/i386.md (fix_trunc<mode>_i387_fisttp): Change "=&1f"
1785 clobber constraint to "=&f".
1786 (fix_truncdi_i387): Ditto.
1787 (lrintxfdi2): Ditto.
1788 (fistdi2_<rounding>): Ditto.
1789 (fpremxf4_i387): Change "=u" constraint to "=f".
1790 (fprem1xf4_i387): Ditto.
1792 (fptanxf4_i387): Ditto.
1793 (fxtractxf3_i387): Ditto.
1794 (fscalexf4_i387): Ditto.
1795 (atan2xf3): Change "u" constraint to "f".
1796 (fyl2xxf3_i387): Ditto.
1797 (fyl2xp1xf3_i387): Ditto.
1799 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
1802 * config/i386/i386.h (NUM_MODES_FOR_MODE_SWITCHING): Update
1803 for removed I387_MASK_PM entity.
1806 2018-09-26 Jeff Law <law@redhat.com>
1808 2018-09-26 Alexey Neyman <stilor@att.net>
1810 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
1811 headers are no longer pulled in by <isl/val.h>.
1813 2018-09-26 Richard Biener <rguenther@suse.de>
1816 * dwarf2out.c (gen_lexical_block_die): Do not equate inline
1817 or concrete instance DIE to the tree. Create abstract origin
1818 attributes also for concrete instances.
1820 2018-09-26 Alexey Neyman <stilor@att.net>
1822 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
1823 headers are no longer pulled in by <isl/val.h>.
1825 2018-09-26 Matthew Malcomson <matthew.malcomson@arm.com>
1827 * config/arm/arm.c (arm_split_compare_and_swap, arm_split_atomic_op):
1828 Use new helper functions.
1829 * config/arm/sync.md (atomic_load<mode>, atomic_store<mode>):
1830 Use new helper functions.
1831 * config/arm/aarch-common-protos.h (aarch_mm_needs_acquire,
1832 aarch_mm_needs_release): New declarations.
1833 * config/arm/aarch-common.c (aarch_mm_needs_acquire,
1834 aarch_mm_needs_release): New.
1836 2018-09-26 Eric Botcazou <ebotcazou@adacore.com>
1838 * config/arm/arm.c (arm_reorg): Skip Thumb reorg pass for thunks.
1839 (arm32_output_mi_thunk): Deal with long calls.
1841 2018-09-26 Richard Biener <rguenther@suse.de>
1845 * tree-inline.c (expand_call_inline): When the location
1846 of the call is UNKNOWN_LOCATION use DECL_SOURCE_LOCATION
1847 or BUILTINS_LOCATION for the BLOCK_SOURCE_LOCATION of
1848 the inserted BLOCK to make inlined_function_outer_scope_p
1850 * dwarf2out.c (add_call_src_coords_attributes): Do not add
1851 coords for reserved locations.
1853 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
1855 * config/rs6000/rs6000.md (*movcc_internal1): Use set_attr_alternative.
1856 (*call_indirect_nonlocal_sysv<mode>): Ditto.
1857 (*call_value_indirect_nonlocal_sysv<mode>): Ditto.
1858 (*sibcall_nonlocal_sysv<mode>): Ditto.
1859 (*sibcall_value_nonlocal_sysv<mode>): Ditto.
1860 (<bd>_<mode>): Ditto.
1861 (<bd>tf_<mode>): Ditto.
1863 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
1865 * config/rs6000/altivec.md (*altivec_mov<mode>): Write the output
1866 control string as a list of templates instead of as C code.
1867 (*altivec_movti): Ditto.
1868 * config/rs6000/darwin.md (movdf_low_di): Ditto.
1870 2018-09-25 Jim Wilson <jimw@sifive.com>
1872 * config/riscv/riscv.c (riscv_split_symbol): Mark auipc label as weak
1873 when target symbol is weak.
1875 2018-09-25 Bernd Edlinger <bernd.edlinger@hotmail.de>
1878 * builtins.c (unterminated_array): Simplify.
1879 * expr.c (string_constant): Handle SSA_NAME. Add more exceptions
1880 where pointer arithmetic is safe.
1882 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
1885 * config/rs6000/altivec.md (altivec_vspltb): Use
1886 const_0_to_15_operand instead of u5bit_cint_operand.
1887 (*altivec_vspltb_internal): Ditto.
1888 (altivec_vspltb_direct): Ditto.
1889 (altivec_vsplth): Use const_0_to_7_operand instead of
1891 (*altivec_vsplth_internal): Ditto.
1892 (altivec_vsplth_direct): Ditto.
1893 (altivec_vspltw): Use const_0_to_3_operand instead of
1895 (*altivec_vspltw_internal): Ditto.
1896 (altivec_vspltw_direct): Ditto.
1897 (altivec_vspltsf): Ditto.
1898 (*altivec_vspltsf_internal): Ditto.
1899 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Handle the
1900 various splats with the proper size immediate. Reorder the various
1901 cases by ascending size of immediate, and put all such together.
1903 2018-09-25 Richard Biener <rguenther@suse.de>
1906 * dwarf2out.c (add_AT_external_die_ref): Remove now redundant
1908 (maybe_create_die_with_external_ref): Do not create
1909 DW_TAG_imported_unit here.
1910 (add_abstract_origin_attribute): Handle external BLOCK refs.
1911 (dwarf2out_abstract_function): Simplify LTO case.
1912 (dwarf2out_early_finish): Create DW_TAG_imported_unit explicitely
1913 rather than using maybe_create_die_with_external_ref.
1915 2018-09-25 Uros Bizjak <ubizjak@gmail.com>
1918 * config/i386/i386.md (frndintxf2_mask_pm): Remove.
1919 (frndintxf2_mask_pm_i387): Ditto.
1920 (nearbyintxf2): Rewrite expander pattern to match rintxf2.
1921 Enable for !flag_trapping_math.
1922 (nearbyint<mode>2): Enable x87 modes for !flag_trapping_math.
1923 Enable SSE modes for TARGET_SSE4_1 and expand them with round insn.
1924 Change operand 1 predicate to nonimmediate_operand.
1925 (attr "i387_cw"): Remove mask_pm.
1926 * config/i386/i386.h (enum ix86_stack_slot): Remove SLOT_CW_MASK_PM.
1927 (enum ix86_entity): Remove I387_MASK_PM.
1928 * config/i386/i386.c (ix86_i387_mode_needed): Do not
1929 handle I387_MASK_PM.
1930 (ix86_mode_needed): Ditto.
1931 (ix86_mode_after): Ditto.
1932 (ix86_mode_entry): Ditto.
1933 (ix86_mode_exit): Ditto.
1934 (emit_i387_cw_initialization): Do not handle I387_CW_MASK_PM.
1936 2018-09-25 Jakub Jelinek <jakub@redhat.com>
1938 * vr-values.c (vr_values::vr_values): Initialize to_remove_edges and
1939 to_update_switch_stmts to vNULL instead of calling create on them
1942 2018-09-25 Richard Biener <rguenther@suse.de>
1944 PR tree-optimization/87402
1945 * tree-ssa-sccvn.c (SSA_VISITED): Remove unused function.
1946 (visit_phi): Re-instantiate handling of supposed to be VARYING
1947 but non-VARYING backedge value.
1949 2018-09-25 Richard Biener <rguenther@suse.de>
1952 * dwarf2out.c (struct sym_off_pair): New.
1953 (external_die_map): New global.
1954 (lookup_decl_die): When in LTO create DIEs lazily from the
1956 (lookup_block_die): New function, create DIEs lazily in LTO.
1957 (equate_block_to_die): New function.
1958 (dwarf2out_die_ref_for_decl): During WPA get the association
1959 from the external DIE map.
1960 (dwarf2out_register_external_die): Record mapping into the
1962 (maybe_create_die_with_external_ref): New function split out from
1963 DIE generation part of old dwarf2out_register_external_die.
1964 (add_abstract_origin_attribute): Do not return the DIE. When
1965 in LTO reference externals directly.
1966 (dwarf2out_abstract_function): When in LTO ignore calls for
1967 decls with external DIEs (already present abstract instances).
1968 (gen_call_site_die): Adjust.
1969 (add_high_low_attributes): Likewise.
1970 (gen_lexical_block_die): Likewise.
1971 (gen_inlined_subroutine_die): Likewie.
1972 (gen_block_die): Likewise.
1973 (dwarf2out_inline_entry): Likewise.
1974 (dwarf2out_early_finish): In LTRANS phase create DW_TAG_imported_unit
1977 2018-09-25 Martin Liska <mliska@suse.cz>
1979 * ipa-fnsummary.c (estimate_node_size_and_time): Scale by two
1980 integers and not by a float value.
1982 2018-09-25 Martin Liska <mliska@suse.cz>
1985 * dbgcnt.c (dbg_cnt_process_single_pair): Return false
1987 * dumpfile.c (dump_enable_all): Remove extra parenthesis.
1988 * gcov-tool.c: Declare the function with ATTRIBUTE_NORETURN.
1989 * godump.c (go_format_type): Remove extra parenthesis.
1991 2018-09-25 Martin Liska <mliska@suse.cz>
1993 * alias.c (set_dest_equal_p): Remove unused function.
1994 * config/i386/i386.c (def_builtin_pure2): Likewise.
1995 * diagnostic-show-locus.c (class layout): Remove
1997 (layout::layout): Likewise here.
1998 * dump-context.h (class temp_dump_context): Likewise.
1999 * dwarf2out.c (add_AT_fde_ref): Remove unused function.
2000 (add_AT_loclistsptr): Likewise.
2001 (add_AT_offset): Likewise.
2002 (get_AT_hi_pc): Likewise.
2003 (is_comdat_die): Likewise.
2004 (type_is_enum): Likewise.
2005 (ceiling): Likewise.
2006 (add_AT_vms_delta): Likewise.
2007 (is_class_die): Likewise.
2008 * edit-context.c (class line_event): Remove unused field.
2009 * graphite-sese-to-poly.c (tree_int_to_gmp): Remove
2011 * ipa-cp.c (ipa_get_vr_lat): Likewise.
2012 * lra-constraints.c (ok_for_index_p_nonstrict): Likewise.
2013 (ok_for_base_p_nonstrict): Likewise.
2014 * tree-chrec.c (is_not_constant_evolution): Likewise.
2015 (chrec_fold_poly_cst): Likewise.
2016 * tree-if-conv.c (has_pred_critical_p): Likewise.
2017 * tree-ssa-coalesce.c (print_exprs): Likewise.
2018 * tree-ssa-pre.c (bitmap_set_contains_expr): Likewise.
2019 * tree-ssa-uninit.c (is_and_or_or_p): Likewise.
2020 * tree-vrp.c (value_ranges_intersect_p): Likewise.
2021 (value_range_nonnegative_p): Likewise.
2023 2018-09-25 Martin Liska <mliska@suse.cz>
2025 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
2026 Do not handle "GNU Pascal".
2027 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
2029 * config/sparc/sparc.c (sparc_pass_by_reference): Remove Pascal
2030 from documentation. Likewise.
2031 * dbxout.c (dbxout_range_type): Likewise.
2032 * doc/cpp.texi: Likewise.
2033 * doc/extend.texi: Likewise.
2034 * doc/frontends.texi: Likewise.
2035 * doc/invoke.texi: Remove Pascal entry.
2036 * tree.def (CLEANUP_POINT_EXPR): Likewise.
2037 * doc/rtl.texi (MODE_FUNCTION): Remove not used entry.
2039 2018-09-25 Martin Liska <mliska@suse.cz>
2042 * doc/invoke.texi: Document all parameters and remove default
2045 2018-09-25 Ilya Leoshkevich <iii@linux.ibm.com>
2048 * rtl.c (rtx_code_size): Take into account that EQ_ATTR_ALT
2049 contains HOST_WIDE_INTs when computing its size.
2051 2018-09-24 Jim Wilson <jimw@sifive.com>
2054 * config/riscv/riscv.h (STACK_BOUNDARY): Test riscv_abi == ABI_ILP32E
2056 (ABI_STACK_BOUNDARY, MAX_ARGS_IN_REGISTERS): Likewise.
2058 2018-09-24 Andrew Pinski <apinski@marvell.com>
2060 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Don't
2061 access prev before checking it for NULLness in the
2062 AARCH64_FUSE_CMP_BRANCH case.
2064 2018-09-24 H.J. Lu <hongjiu.lu@intel.com>
2067 * config/i386/i386.c (rest_of_insert_endbranch): Set
2068 endbr_queued_at_entrance to true and don't insert ENDBR if
2069 x86_function_profiler will be called.
2070 (x86_function_profiler): Insert ENDBR if endbr_queued_at_entrance
2072 * config/i386/i386.h (machine_function): Add
2073 endbr_queued_at_entrance.
2075 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
2077 * genattrtab.c (mk_attr_alt): Use alternative_mask.
2078 (attr_rtx_1): Adjust caching to match the new EQ_ATTR_ALT field
2080 (check_attr_test): Use alternative_mask.
2081 (get_attr_value): Likewise.
2082 (compute_alternative_mask): Use alternative_mask and XWINT.
2083 (make_alternative_compare): Use alternative_mask.
2084 (attr_alt_subset_p): Use XWINT.
2085 (attr_alt_subset_of_compl_p): Likewise.
2086 (attr_alt_intersection): Use alternative_mask and XWINT.
2087 (attr_alt_union): Likewise.
2088 (attr_alt_complement): Use HOST_WIDE_INT and XWINT.
2089 (mk_attr_alt): Use alternative_mask and HOST_WIDE_INT.
2090 (simplify_test_exp): Use alternative_mask and XWINT.
2091 (write_test_expr): Use alternative_mask and XWINT, adjust bit
2092 number calculation to support 64 bits. Generate code that
2093 checks 64-bit masks.
2094 (main): Use alternative_mask.
2095 * rtl.def (EQ_ATTR_ALT): Change field types from ii to ww.
2097 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
2100 * config/s390/s390.c (s390_emit_epilogue): Do not use PARALLEL
2101 RETURN+USE when returning via %r14.
2103 2018-09-24 Martin Liska <mliska@suse.cz>
2105 * gcov.c (output_lines): Print colorization legend
2106 for both flag_use_colors and flag_use_hotness_colors.
2109 2018-09-24 Martin Liska <mliska@suse.cz>
2111 * coverage.c (get_coverage_counts): Use warning_at
2112 with current_function_decl location. Use %qD in warning
2115 2018-09-24 Martin Liska <mliska@suse.cz>
2117 * memory-block.h (memory_block_pool::release): Annotate with
2118 valgrind that the memory is not accessible.
2120 2018-09-24 Martin Liska <mliska@suse.cz>
2123 * asan.c: Make asan_handled_variables extern.
2125 * cfgexpand.c (expand_stack_vars): Make sure
2126 a representative is unpoison if another
2127 variable in the partition is handled by
2128 use-after-scope sanitization.
2130 2018-09-24 Richard Biener <rguenther@suse.de>
2132 PR tree-optimization/63155
2133 * tree-ssa-propagate.c (add_ssa_edge): Avoid adding PHIs to
2134 the worklist when the edge of the respective argument isn't
2137 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
2139 * config/i386/i386.h (enum reg_class): Rename MASK_REGS to
2140 ALL_MASK_REGS and MASK_EVEX_REGS to MASK_REGS.
2141 (MASK_CLASS_P): Update for rename.
2142 (MAYBE_MASK_CLASS_P): Ditto.
2143 (REG_CLASS_NAMES): Update.
2144 (REG_CLASS_CONTENT): Update.
2145 * config/i386/i386.c (regclass_map): Update for MASK_REG
2146 and ALL_MASK_REGS rename.
2147 * config/i386/constraints.md (Yk): Update for rename.
2150 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
2152 * config/i386/i386.h (enum reg_class): Remove
2153 EVEX_SSE_REGS and MOD4_SSE_REGS.
2154 (REG_CLASS_NAMES): Update.
2155 (REG_CLASS_CONTENT): Update.
2156 * config/i386/i386.c (regclass_map): Declare AVX-512 SSE
2157 registers as ALL_SSE_REGS.
2158 (ix86_additional_allocno_class_p): Remove.
2159 (TARGET_ADDITIONAL_ALLOCNO_CLASS_P): Remove.
2160 (ix86_register_priority): Lower priority of EVEX SSE registers.
2161 Use IN_RANGE macro where appropriate.
2162 (ix86_hard_regno_mode_ok): Merge AVX-5124FMAPS and
2163 AVX-5124VNNIW checks.
2164 (ix86_modes_tieable_p): Tie 512-bit SSE modes.
2165 * config/i386/sse.md (avx5124fmaddps_4fmaddps)
2166 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddps_maskz)
2167 (avx5124fmaddps_4fmaddss, avx5124fmaddps_4fmaddss_mask)
2168 (avx5124fmaddps_4fmaddss_maskz, avx5124fmaddps_4fnmaddps)
2169 (avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddps_maskz)
2170 (avx5124fmaddps_4fnmaddss, avx5124fmaddps_4fnmaddss_mask)
2171 (avx5124fmaddps_4fnmaddss_maskz, avx5124vnniw_vp4dpwssd)
2172 (avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssd_maskz)
2173 (avx5124vnniw_vp4dpwssds, avx5124vnniw_vp4dpwssds_mask)
2174 (avx5124vnniw_vp4dpwssds_maskz): Use "v" instead of "Yh" constraint.
2175 * config/i386/constraints.md (Yh): Remove.
2177 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
2179 * config/i386/i386.c (regclass_map): Declare integer REX registers
2182 2018-09-23 Gerald Pfeifer <gerald@pfeifer.com>
2184 * doc/service.texi (Service): Switch the fsf.org link to https.
2186 2018-09-22 Chung-Ju Wu <jasonwucj@gmail.com>
2189 * config/nds32/nds32.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
2190 Define to speculation_safe_value_not_needed.
2192 2018-09-21 Florian Weimer <fweimer@redhat.com>
2195 * doc/extend.texi (Common Function Attributes): Mention that
2196 noreturn suppresses tail call optimization.
2198 2018-09-21 Jeff Law <law@redhat.com>
2200 * gimple-ssa-evrp.c (evrp_dom_walker::cleanup): Call
2201 vr_values::cleanup_edges_and_switches.
2202 * tree-vrp.c (to_remove_edges, to_update_switch_stmts): Moved into
2204 (identify_jump_threads): Remove EDGE_IGNORE handling.
2205 (execute_vrp): Move handling of to_remove_edges and
2206 to_update_switch_stmts into vr_values class member functions.
2207 * tree-vrp.h (switch_update, to_remove_edges): Remove declarations.
2208 (to_update_switch_stmts): Likewise.
2209 * vr-values.c: Include cfghooks.h.
2210 (vr_values::vr_values): Initialize to_remove_edges and
2211 to_update_switch_stmts.
2212 (vr_values::~vr_values): Verify to_remove_edges and
2213 to_update_switch_stmts are empty.
2214 (vr_values::simplify_switch_using_ranges): Set EDGE_IGNORE as needed.
2215 (vr_values::cleanup_edges_and_switches): New member function.
2216 * vr-values.h (vr_values): Add cleanup_edges_and_switches member
2217 function. Add new data members.
2219 2018-09-21 David Malcolm <dmalcolm@redhat.com>
2221 PR tree-optimization/87309
2222 * dumpfile.c (dump_context::begin_scope): Filter the dump_loc
2223 calls with pflags and alt_flags.
2224 (selftest::test_capture_of_dump_calls): Add test of interaction of
2225 MSG_OPTIMIZED_LOCATIONS with AUTO_DUMP_SCOPE.
2227 2018-09-21 Olivier Hainque <hainque@adacore.com>
2229 * config.gcc: Factorize and comment inclusion of vxworks-dummy.h.
2231 2018-09-21 Olivier Hainque <hainque@adacore.com>
2233 * config/vxworks.h (CLEAR_INSN_CACHE): #define to 1.
2235 2018-09-21 Olivier Hainque <hainque@adacore.com>
2237 * config/vxworks.h (VXWORKS_LIBGCC_SPEC): Remove -lc_internal.
2238 Merge block comment with the one ahead of VXWORKS_LIBS_RTP. Then:
2239 (VXWORKS_LIBS_RTP): Minor reordering.
2241 2018-09-21 Olivier Hainque <hainque@adacore.com>
2243 * config/vxworks.h (STARTFILE_PREFIX_SPEC): Define.
2244 (VXWORKS_LIBS_DIR_RTP): Remove definition and use.
2246 2018-09-21 Olivier Hainque <hainque@adacore.com>
2248 * config/vxworks.h (SIZE_TYPE): Account for TARGET_VXWORKS64.
2249 (PTRDIFF_TYPE): Likewise.
2251 2018-09-21 Olivier Hainque <hainque@adacore.com>
2253 * config.gcc: Enforce def of TARGET_VXWORKS64 to 1 from
2254 triplet, similar to support for VxWorks7.
2255 * config/vxworks-dummy.h: Provide a default definition
2256 of TARGET_VXWORKS64 to 0.
2258 2018-09-21 Olivier Hainque <hainque@adacore.com>
2260 * config/vxworks.h (TARGET_VXWORKS7): Move default definition ...
2261 * config/vxworks-dummy.h: here.
2263 2018-09-21 Olivier Hainque <hainque@adacore.com>
2265 * config.gcc: Prepend vxworks-dummy.h to tm_file for powerpc*
2267 2018-09-21 Shaokun Zhang <zhangshaokun@hisilicon.com>
2268 Bo Zhou <zbo.zhou@hisilicon.com>
2270 * config/aarch64/aarch64-cores.def (tsv110): New CPU.
2271 * config/aarch64/aarch64-tune.md: Regenerated.
2272 * doc/invoke.texi (AArch64 Options/-mtune): Add "tsv110".
2273 * config/aarch64/aarch64.c (tsv110_tunings): New tuning table.
2274 * config/aarch64/aarch64-cost-tables.h: Add "tsv110" extra costs.
2276 2018-09-21 Andrew Stubbs <ams@codesourcery.com>
2277 Julian Brown <julian@codesourcery.com>
2279 * builtins.c (get_builtin_sync_mem): Handle address spaces.
2281 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
2283 * config/rs6000/rs6000.c (rs6000_function_ok_for_sibcall): Return false
2284 if the call takes a static chain.
2286 2018-09-21 Martin Liska <mliska@suse.cz>
2288 * auto-profile.c (autofdo_source_profile::read): Do not
2290 (read_profile): Do not add working sets.
2291 (read_autofdo_file): Remove sum_all.
2292 (afdo_callsite_hot_enough_for_early_inline): Remove const
2294 * coverage.c (struct counts_entry): Remove gcov_summary.
2295 (read_counts_file): Read new GCOV_TAG_OBJECT_SUMMARY,
2296 do not support GCOV_TAG_PROGRAM_SUMMARY.
2297 (get_coverage_counts): Remove summary and expected
2299 * coverage.h (get_coverage_counts): Likewise.
2300 * doc/gcov-dump.texi: Remove -w option.
2301 * gcov-dump.c (dump_working_sets): Remove.
2302 (main): Do not support '-w' option.
2303 (print_usage): Likewise.
2304 (tag_summary): Likewise.
2305 * gcov-io.c (gcov_write_summary): Do not dump
2307 (gcov_read_summary): Likewise.
2308 (gcov_histo_index): Remove.
2309 (gcov_histogram_merge): Likewise.
2310 (compute_working_sets): Likewise.
2311 * gcov-io.h (GCOV_TAG_OBJECT_SUMMARY): Mark
2313 (GCOV_TAG_PROGRAM_SUMMARY): Mark it obsolete.
2314 (GCOV_TAG_SUMMARY_LENGTH): Adjust.
2315 (GCOV_HISTOGRAM_SIZE): Remove.
2316 (GCOV_HISTOGRAM_BITVECTOR_SIZE): Likewise.
2317 (struct gcov_summary): Simplify rapidly just
2318 to runs and sum_max fields.
2319 (gcov_histo_index): Remove.
2320 (NUM_GCOV_WORKING_SETS): Likewise.
2321 (compute_working_sets): Likewise.
2322 * gcov-tool.c (print_overlap_usage_message): Remove
2323 trailing empty line.
2324 * gcov.c (read_count_file): Read GCOV_TAG_OBJECT_SUMMARY.
2325 (output_lines): Remove program related line.
2326 * ipa-profile.c (ipa_profile): Do not consider GCOV histogram.
2327 * lto-cgraph.c (output_profile_summary): Do not stream GCOV
2329 (input_profile_summary): Do not read it.
2330 (merge_profile_summaries): And do not merge it.
2331 (input_symtab): Do not call removed function.
2332 * modulo-sched.c (sms_schedule): Do not print sum_max.
2333 * params.def (HOT_BB_COUNT_FRACTION): Reincarnate param that was
2334 removed when histogram method was invented.
2335 (HOT_BB_COUNT_WS_PERMILLE): Mention that it's used only in LTO
2337 * postreload-gcse.c (eliminate_partially_redundant_load): Fix
2339 * predict.c (get_hot_bb_threshold): Use HOT_BB_COUNT_FRACTION
2340 and dump selected value.
2341 * profile.c (add_working_set): Remove.
2342 (get_working_sets): Likewise.
2343 (find_working_set): Likewise.
2344 (get_exec_counts): Do not work with working sets.
2345 (read_profile_edge_counts): Do not inform as sum_max is removed.
2346 (compute_branch_probabilities): Likewise.
2347 (compute_value_histograms): Remove argument for call of
2348 get_coverage_counts.
2349 * profile.h: Do not make gcov_summary const.
2351 2018-09-21 Monk Chiang <sh.chiang04@gmail.com>
2353 * config.gcc (nds32*-*-*): Set TARGET_DEFAULT_TLSDESC_TRAMPOLINE=0.
2355 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
2357 PR tree-optimization/86990
2358 * gimple-ssa-store-merging.c (imm_store_chain_info:coalesce_immediate):
2359 Check that the entire merged store group is made of constants only for
2362 2018-09-20 Allan Sandfeld Jensen <allan.jensen@qt.io>
2364 * gcc.c (LINK_COMMAND_SPEC): Handle -r like -nostdlib.
2365 (VTABLE_VERIFICATION_SPEC): Likewise.
2366 (SANITIZER_EARLY_SPEC): Likewise.
2367 (SANITIZER_SPEC): Likewise.
2368 * config/darwin.h (LINK_COMMAND_SPEC): Likewise.
2369 * doc/invoke.texi (Link Options): Document -r.
2371 2018-09-20 Richard Biener <rguenther@suse.de>
2374 * gimplify.c (gimplify_expr): Retain alignment of
2375 addressable lvalue in dereference.
2377 2018-09-20 Alexandre Oliva <aoliva@redhat.com>
2380 * configure.ac: Check for .loc is_stmt support.
2381 * configure, config.in: Rebuilt.
2382 * dwarf2out.c (dwarf2out_source_line): Skip is_stmt
2385 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
2387 * config/rs6000/rs6000.opt (misel=no, misel=yes): Delete.
2388 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -misel=yes and
2391 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
2393 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
2395 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete
2398 2018-09-20 Marek Polacek <polacek@redhat.com>
2400 * doc/invoke.texi: Add -Wno-init-list-lifetime to C++ Language Options.
2402 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
2404 PR tree-optimization/87288
2405 * tree-vect-loop.c (vect_analyze_loop_2): Take PEELING_FOR_GAPS
2406 into account when determining PEELING_FOR_NITERS.
2408 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
2410 PR tree-optimization/86877
2411 * tree-vect-loop.c (vect_analyze_loop_2): Call
2412 vect_verify_datarefs_alignment.
2414 2018-09-19 Marek Polacek <polacek@redhat.com>
2416 * doc/invoke.texi: Document -Wclass-conversion.
2418 2018-09-19 John David Anglin <danglin@gcc.gnu.org>
2420 * config/pa/pa.c (pa_adjust_priority): Delete.
2421 (TARGET_SCHED_ADJUST_PRIORITY): Delete define.
2423 * config/pa/pa.md (atomic_storeqi): Restore deleted expander.
2424 (atomic_storehi): Likewise.
2425 (atomic_storesi): Likewise.
2426 (atomic_loaddi): Restore compare and swap exchange loop code.
2428 2018-09-19 Segher Boessenkool <segher@kernel.crashing.org>
2430 PR rtl-optimization/86902
2431 * combine.c (try_combine): When changing the CC mode used, don't change
2432 an unrelated mode in other_insn to that new CC mode.
2434 2018-09-19 David Malcolm <dmalcolm@redhat.com>
2436 * tree-data-ref.c (runtime_alias_check_p): Use formatted printing
2437 with %T in place of calls to dump_generic_expr.
2438 (prune_runtime_alias_test_list): Likewise.
2439 (create_runtime_alias_checks): Likewise.
2440 * tree-vect-data-refs.c (vect_check_nonzero_value): Likewise.
2441 (vect_analyze_data_ref_dependence): Likewise.
2442 (vect_slp_analyze_data_ref_dependence): Likewise.
2443 (vect_record_base_alignment): Likewise. Use %G in place of call
2444 to dump_gimple_stmt.
2445 (vect_compute_data_ref_alignment): Likewise.
2446 (verify_data_ref_alignment): Likewise.
2447 (vect_find_same_alignment_drs): Likewise.
2448 (vect_analyze_group_access_1): Likewise.
2449 (vect_analyze_data_ref_accesses): Likewise.
2450 (dependence_distance_ge_vf): Likewise.
2451 (dump_lower_bound): Likewise.
2452 (vect_prune_runtime_alias_test_list): Likewise.
2453 (vect_find_stmt_data_reference): Likewise.
2454 (vect_analyze_data_refs): Likewise.
2455 (vect_create_addr_base_for_vector_ref): Likewise.
2456 (vect_create_data_ref_ptr): Likewise.
2457 * tree-vect-loop-manip.c (vect_set_loop_condition): Likewise.
2458 (vect_can_advance_ivs_p): Likewise.
2459 (vect_update_ivs_after_vectorizer): Likewise.
2460 (vect_gen_prolog_loop_niters): Likewise.
2461 (vect_prepare_for_masked_peels): Likewise.
2462 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
2463 (vect_determine_vectorization_factor): Likewise.
2464 (vect_is_simple_iv_evolution): Likewise.
2465 (vect_analyze_scalar_cycles_1): Likewise.
2466 (vect_analyze_loop_operations): Likewise.
2467 (report_vect_op): Likewise.
2468 (vect_is_slp_reduction): Likewise.
2469 (check_reduction_path): Likewise.
2470 (vect_is_simple_reduction): Likewise.
2471 (vect_create_epilog_for_reduction): Likewise.
2472 (vect_finalize_reduction:): Likewise.
2473 (vectorizable_induction): Likewise.
2474 (vect_transform_loop_stmt): Likewise.
2475 (vect_transform_loop): Likewise.
2476 (optimize_mask_stores): Likewise.
2477 * tree-vect-patterns.c (vect_pattern_detected): Likewise.
2478 (vect_split_statement): Likewise.
2479 (vect_recog_over_widening_pattern): Likewise.
2480 (vect_recog_average_pattern): Likewise.
2481 (vect_determine_min_output_precision_1): Likewise.
2482 (vect_determine_precisions_from_range): Likewise.
2483 (vect_determine_precisions_from_users): Likewise.
2484 (vect_mark_pattern_stmts): Likewise.
2485 (vect_pattern_recog_1): Likewise.
2486 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
2487 (vect_record_max_nunits): Likewise.
2488 (vect_build_slp_tree_1): Likewise.
2489 (vect_build_slp_tree_2): Likewise.
2490 (vect_print_slp_tree): Likewise.
2491 (vect_analyze_slp_instance): Likewise.
2492 (vect_detect_hybrid_slp_stmts): Likewise.
2493 (vect_detect_hybrid_slp_1): Likewise.
2494 (vect_slp_analyze_operations): Likewise.
2495 (vect_slp_analyze_bb_1): Likewise.
2496 (vect_transform_slp_perm_load): Likewise.
2497 (vect_schedule_slp_instance): Likewise.
2498 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
2499 (vect_mark_stmts_to_be_vectorized): Likewise.
2500 (vect_init_vector_1): Likewise.
2501 (vect_get_vec_def_for_operand): Likewise.
2502 (vect_finish_stmt_generation_1): Likewise.
2503 (vect_check_load_store_mask): Likewise.
2504 (vectorizable_call): Likewise.
2505 (vectorizable_conversion): Likewise.
2506 (vectorizable_operation): Likewise.
2507 (vectorizable_load): Likewise.
2508 (vect_analyze_stmt): Likewise.
2509 (vect_is_simple_use): Likewise.
2510 (vect_get_vector_types_for_stmt): Likewise.
2511 (vect_get_mask_type_for_stmt): Likewise.
2512 * tree-vectorizer.c (increase_alignment): Likewise.
2514 2018-09-19 Andrew Stubbs <ams@codesourcery.com>
2516 * doc/rtl.texi: Adjust vec_select description.
2517 * simplify-rtx.c (simplify_binary_operation_1): Allow VEC_SELECT to use
2518 non-constant selectors.
2520 2018-09-19 Matthew Malcomson <matthew.malcomson@arm.com>
2522 * config/aarch64/aarch64-protos.h
2523 (aarch64_offset_9bit_signed_unscaled_p): New declaration.
2524 * config/aarch64/aarch64.md (arches): New "rcpc8_4" attribute value.
2525 (arch_enabled): Add check for "rcpc8_4" attribute value of "arch".
2526 * config/aarch64/aarch64.h (AARCH64_FL_RCPC8_4): New bitfield.
2527 (AARCH64_FL_FOR_ARCH8_4): Include AARCH64_FL_RCPC8_4.
2528 (AARCH64_FL_PROFILE): Move index so flags are ordered.
2529 (AARCH64_ISA_RCPC8_4): New flag.
2530 * config/aarch64/aarch64.c (offset_9bit_signed_unscaled_p): Renamed
2531 to aarch64_offset_9bit_signed_unscaled_p.
2532 * config/aarch64/atomics.md (atomic_store<mode>): Allow offset
2534 * config/aarch64/constraints.md (Ust): New constraint.
2535 * config/aarch64/predicates.md.
2536 (aarch64_9bit_offset_memory_operand): New predicate.
2537 (aarch64_rcpc_memory_operand): New predicate.
2539 2018-09-19 Eric Botcazou <ebotcazou@adacore.com>
2541 PR rtl-optimization/87361
2542 * rtlanal.c (nonzero_bits1): Revert accidental change.
2544 2018-09-19 Richard Biener <rguenther@suse.de>
2546 PR tree-optimization/87349
2547 PR tree-optimization/87342
2548 * tree-ssa-sccvn.c (do_rpo_vn): Iterate max_rpo computation.
2550 2018-09-18 Marek Polacek <polacek@redhat.com>
2552 P1064R0 - Allowing Virtual Function Calls in Constant Expressions
2553 * gimple-fold.c (gimple_get_virt_method_for_vtable): Adjust assert.
2555 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
2557 * config/rs6000/rs6000.md: Remove old "Cygnus sibcall" comment.
2559 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
2561 PR rtl-optimization/86882
2562 * rtlanal.c (reg_overlap_mentioned_p): Handle CLOBBER.
2564 2018-09-18 Uros Bizjak <ubizjak@gmail.com>
2566 * config/i386/i386.md (*<code>extend<mode>xf2): Macroize insn from
2567 *<code>extendsfxf2 and *<code>extenddfxf2 using MODEF mode iterator.
2569 2018-09-18 Jonathan Wakely <jwakely@redhat.com>
2572 * doc/invoke.texi (Link Options): Fix formatting and grammar.
2574 2018-09-18 Richard Biener <rguenther@suse.de>
2577 * tree-ssa-coalesce.c (tree_int_map_hasher): Remove.
2578 (compute_samebase_partition_bases): Likewise.
2579 (coalesce_ssa_name): Always use compute_optimized_partition_bases.
2580 (gimple_can_coalesce_p): Simplify.
2582 2018-09-18 Hans-Peter Nilsson <hp@bitrange.com>
2584 Handle a library implementation of ffs calling __builtin_ffs.
2585 * config/mmix/mmix.c (TARGET_INIT_LIBFUNCS): Override with...
2586 (mmix_init_libfuncs): New function: make __builtin_ffs expand
2589 2018-09-17 David Malcolm <dmalcolm@redhat.com>
2591 * diagnostic-show-locus.c (class layout_range): Add field
2593 (layout_range::layout_range): Add "original_idx" param and use it
2594 to initialize new field.
2595 (make_range): Use 0 for original_idx.
2596 (layout::layout): Pass in index to calls to
2597 maybe_add_location_range.
2598 (layout::maybe_add_location_range): Add param "original_idx" and
2599 pass it on to layout_range.
2600 (layout::print_any_labels): Pass on range->m_original_idx to
2602 (gcc_rich_location::add_location_if_nearby): Use 0 for
2604 * gcc-rich-location.h (text_range_label::get_text): Update for new
2606 (range_label_for_type_mismatch::get_text): Likewise.
2608 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
2610 * config/i386/i386.c (ix86_emit_i387_log1p): Emit fldln2 earlier.
2612 2018-09-17 David Malcolm <dmalcolm@redhat.com>
2614 * gimple-ssa-sprintf.c (fmtwarn): Update for introduction of
2615 format_string_diagnostic_t.
2616 (fmtwarn_n): Likewise.
2617 * substring-locations.c
2618 (format_string_diagnostic_t::format_string_diagnostic_t) New ctor.
2619 (format_warning_n_va): Convert to...
2620 (format_string_diagnostic_t::emit_warning_n_va): ...this.
2621 (format_warning_va): Convert to...
2622 (format_string_diagnostic_t::emit_warning_va): ...this.
2623 (format_warning_at_substring): Convert to...
2624 (format_string_diagnostic_t::emit_warning): ...this.
2625 (format_warning_at_substring_n): Convert to...
2626 (format_string_diagnostic_t::emit_warning_n): ...this.
2627 * substring-locations.h (class format_string_diagnostic_t): New
2629 (format_warning_va): Convert to
2630 format_string_diagnostic_t::emit_warning_va.
2631 (format_warning_n_va): Convert to
2632 format_string_diagnostic_t::emit_warning_n_va.
2633 (format_warning_at_substring): Convert to
2634 format_string_diagnostic_t::emit_warning.
2635 (format_warning_at_substring_n): Convert to
2636 format_string_diagnostic_t::emit_warning_n.
2638 2018-09-17 Cesar Philippidis <cesar@codesourcery.com>
2639 Bernd Schmidt <bernds_cb1@t-online.de>
2641 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with
2644 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
2646 * config/i386/i386.md (truncxf<mode>2_i387_noop_unspec): Change
2647 operand 0 predicate to nonimmediate operand.
2648 (rint<mode>2_frndint): Remove insn pattern.
2649 (rint<mode>2): Change operand 1 predicate to general_operand.
2650 Extend operand 1 to XFmode and generate rintxf2 insn.
2651 (frndintxf2_<rounding>): Rename from frndint<mode>2_<rounding>.
2652 Do not use X87MODEF mode macro.
2653 (frndintxf2_<rounding>_i387): Rename from
2654 frndint<mode>2_<rounding>_i387. Do not use X87MODEF mode macro.
2655 (<rounding_insn><mode>2): For non-SSE modes, extend operand 1
2656 to XFmode and generate significandxf3 insn.
2658 2018-09-17 Richard Biener <rguenther@suse.de>
2660 PR tree-optimization/87328
2661 * tree-ssa-sccvn.c (process_bb): Remove assertion about not
2662 visiting unexecutable backedges when not iterating.
2663 (do_rpo_vn): Mark all edges not executable even when not
2666 2018-09-17 Martin Jambor <mjambor@suse.cz>
2669 * doc/invoke.texi (Warning Options): Likewise.
2671 2018-09-17 Richard Biener <rguenther@suse.de>
2673 PR tree-optimization/87301
2674 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup): Properly
2675 clean EH info from leftover copy assignments.
2677 2018-09-17 Martin Liska <mliska@suse.cz>
2679 PR gcov-profile/85871
2680 * gcov.c (output_intermediate_file): Fix out of bounds
2683 2018-09-17 Vineet Gupta <vgupta@synopsys.com>
2685 * config/arc/arc.c: Object attributes for core4 not reflected
2687 * config/arc/arc.h: Don't restrict DBNZ to core3 (core4 includes
2690 2018-09-17 Alexey Brodkin <abrodkin@synopsys.com>
2692 * config/arc/linux.h (LINK_EH_SPEC): Add missing space.
2694 2018-09-17 Martin Liska <mliska@suse.cz>
2696 * doc/gcov.texi: Document new option --use-hotness-colors.
2697 * gcov.c (struct source_info): Declare new field.
2698 (source_info::source_info): Set default for maximum_count.
2699 (print_usage): Add new -q option.
2700 (process_args): Process it.
2701 (accumulate_line_info): Save src->maximum_count.
2702 (output_line_beginning): Make color line number if
2703 flag_use_hotness_colors is set.
2704 (output_line_details): Pass default argument value.
2705 (output_lines): Pass src->maximum_count.
2707 2018-09-17 Martin Liska <mliska@suse.cz>
2709 * common/config/i386/i386-common.c (ix86_get_valid_option_values):
2710 Use processor_names table.
2711 * config/i386/i386.c (ix86_default_align): Use
2712 processor_cost_table for alignment values.
2713 (ix86_option_override_internal): Use processor_names.
2714 (ix86_function_specific_print): Likewise.
2715 * config/i386/i386.h (struct processor_costs):
2716 Add alignment values.
2717 (struct ptt): Remove and replace with const char *.
2718 * config/i386/x86-tune-costs.h (struct processor_costs):
2719 Declare default alignments for all costs.
2721 2018-09-17 Aldy Hernandez <aldyh@redhat.com>
2723 * tree-vrp.c (extract_range_from_unary_expr): Do not special case
2724 symbolics or VR_VARYING ranges for ABS_EXPR.
2725 * wide-int-range.cc (wide_int_range_abs): Return positive numbers
2726 when range will wrap.
2728 2018-09-15 Eric Botcazou <ebotcazou@adacore.com>
2731 * cfgexpand.c (expand_gimple_basic_block): Be prepared for a BARRIER
2732 before and after a JUMP_TABLE_DATA.
2734 2018-09-14 John David Anglin <danglin@gcc.gnu.org>
2737 * dojump.c (do_compare_and_jump): Canonicalize function pointers
2738 when one operand is a function pointer. Use POINTER_TYPE_P and
2739 FUNC_OR_METHOD_TYPE_P.
2740 * expr.c (do_store_flag): Use POINTER_TYPE_P and FUNC_OR_METHOD_TYPE_P.
2741 * fold-const.c (build_range_check): Likewise.
2742 * match.pd (simple_comparison): Likewise.
2744 2018-09-14 David Malcolm <dmalcolm@redhat.com>
2747 * spellcheck.c (get_edit_distance_cutoff): New function.
2748 (selftest::test_edit_distance_unit_test_oneway): Rename to...
2749 (selftest::test_get_edit_distance_one_way): ...this.
2750 (selftest::test_get_edit_distance_unit): Rename to...
2751 (selftest::test_get_edit_distance_both_ways): ...this.
2752 (selftest::test_edit_distances): Move tests to this new function,
2753 and test some more pairs of strings. Update for above renaming.
2754 (selftest::get_old_cutoff): New function.
2755 (selftest::test_get_edit_distance_cutoff): New function.
2756 (selftest::assert_suggested_for): New function.
2757 (ASSERT_SUGGESTED_FOR): New macro.
2758 (selftest::assert_not_suggested_for): New function.
2759 (ASSERT_NOT_SUGGESTED_FOR): New macro.
2760 (selftest::test_suggestions): New function.
2761 (selftest::spellcheck_c_tests): Move test_get_edit_distance_unit
2762 tests to selftest::test_edit_distances and call it. Add calls to
2763 selftest::test_get_edit_distance_cutoff and
2764 selftest::test_suggestions.
2765 * spellcheck.h (get_edit_distance_cutoff): New function declaration.
2766 (best_match::consider): Replace hard-coded cutoff calculation with
2768 (best_match::get_cutoff): New declaration.
2769 (best_match::get_best_meaningful_candidate): Likewise.
2771 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
2773 * builtins.c (fold_builtin_strlen): Remove TODO comment.
2775 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
2778 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
2780 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
2781 terminated string literal.
2783 2018-09-14 Martin Sebor <msebor@redhat.com>
2785 * builtins.c (unterminated_array): Handle ARRAY_REF.
2786 (expand_builtin_stpcpy_1): Detect unterminated char arrays.
2787 * builtins.h (unterminated_array): Declare extern.
2788 * gimple-fold.c (gimple_fold_builtin_stpcpy): Detect unterminated
2790 (gimple_fold_builtin_sprintf): Propagate NO_WARNING to transformed
2793 2018-09-14 Martin Sebor <msebor@redhat.com>
2794 Jeff Law <law@redhat.com>
2796 * builtins.c (unterminated_array): New.
2797 (expand_builtin_strcpy): Adjust.
2798 (expand_builtin_strcpy_args): Detect unterminated arrays.
2799 * gimple-fold.c (get_maxval_strlen): Add argument. Detect
2800 unterminated arrays.
2801 * gimple-fold.h (get_maxval_strlen): Add argument.
2802 (gimple_fold_builtin_strcpy): Detec unterminated arrays.
2804 * gimple-fold.c (get_range_strlen): Add argument.
2805 (get_maxval_strlen): Adjust.
2806 * gimple-fold.h (get_range_strlen): Add argument.
2808 2018-09-14 Wei Xiao <wei3.xiao@intel.com>
2810 * config/i386/movdirintrin.h: Fix copyright year.
2812 2018-09-14 Uros Bizjak <ubizjak@gmail.com>
2814 * reg-stack.c: Include regs.h.
2815 (replace_reg): Assert that mode is MODE_FLOAT or MODE_COMPLEX_FLOAT.
2816 (emit_pop_insn): Default pop insn mode to the reg_raw_mode of
2817 FIRST_STACK_REG, not DFmode.
2818 (emit_swap_insn): Default swap insn mode to the reg_raw_mode of
2819 FIRST_STACK_REG, not XFmode. Explicitly construct swap RTX.
2820 (change stack): Default register mode to the reg_raw_mode of
2821 FIRST_STACK_REG, not DFmode.
2822 * config/i386/i386.md (*swap<mode>): Remove insn pattern.
2823 (*swapxf): Rename from swapxf.
2825 2018-09-14 Carl Love <cel@us.ibm.com>
2827 * config/rs6000/emmintrin.h: Add _MM_SHUFFLE2.
2828 * config/rs6000/xmmintrin.h: Add _MM_SHUFFLE.
2830 2018-09-14 Segher Boessenkool <segher@kernel.crashing.org>
2833 * config/rs6000/rs6000.md (*mov<mode>_hardfloat64): Add Z to the Y
2836 2018-09-14 Sam Tebbs <sam.tebbs@arm.com>
2839 * config/aarch64/aarch64.md (*aarch64_bfxilsi_uxtw): Define.
2841 2018-09-14 Jason Merrill <jason@redhat.com>
2843 Fix --enable-gather-detailed-mem-stats.
2844 * hash-table.c (hash_table_usage): Change from variable to function.
2845 * hash-table.h: Adjust.
2846 * Makefile.in: Add missing dependencies on hash-table.h.
2848 2018-09-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2850 PR tree-optimization/87259
2852 (pass_cse_reciprocals::execute): Run optimize_recip_sqrt after
2853 execute_cse_reciprocals_1 has tried transforming.
2855 2018-09-14 Aldy Hernandez <aldyh@redhat.com>
2857 * tree-vrp.c (extract_range_from_binary_expr_1): Normalize
2858 VR_VARYING for PLUS/MINUS_EXPR.
2860 2018-09-14 Ilya Leoshkevich <iii@linux.ibm.com>
2862 * config/s390/s390-passes.def (INSERT_PASS_BEFORE): Improve
2865 2018-09-14 Richard Biener <rguenther@suse.de>
2868 * tree-ssa-coalesce.c (ssa_conflicts_merge): Remove conflict
2869 bits for the merged partition.
2871 2018-09-13 Martin Sebor <msebor@redhat.com>
2872 Bernd Edlinger <bernd.edlinger@hotmail.de>
2874 * builtins.h (c_srlen): Add argument.
2875 * builtins.c (warn_string_no_nul): New function.
2876 (c_strlen): Add argument and use it. Update recursive calls.
2877 Pass DECL argument to string_constant to get info on non
2878 terminated strings. Update *NONSTR as needed.
2879 (fold_builtin_strlen): Add argument to calls to c_strlen.
2880 Warn for unterminated arrays.
2881 (warn_string_no_null): Add prototype.
2882 * expr.c (string_constant): Update arguments. Update recursive
2883 calls appropriately. Detect missing NUL terminator and outermost
2884 declaration its missing in.
2885 Improve checks for arrays with nonzero lower bound or elements
2886 that are not a single byte. Simplify offset computation.
2887 Simplify checks for non-NUL terminated strings.
2888 * gimple-fold.c (get_range_strlen): Add argument to c_strlen call.
2889 * gimple-ssa-sprintf.c (get_string_length): Remove unnecessary code.
2891 2018-09-13 Bernd Edlinger <bernd.edlinger@hotmail.de>
2893 * builtins.c (c_strlen): Handle not zero terminated STRING_CSTs
2895 * fold-const.c (c_getstr): Fix function comment. Remove unused third
2896 argument. Fix range checks.
2897 * fold-const.h (c_getstr): Adjust protoype.
2898 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid folding when
2899 string is constant but contains no NUL byte.
2901 * expr.c (string_constant): Adjust function comment.
2902 Remove bogus check for zero termination.
2904 * fold-const.c (c_getstr): Clamp STRING_LENGTH to STRING_SIZE.
2906 * varasm.c (compare_constant): Compare type size of STRING_CSTs.
2907 (get_constant_size): Don't make STRING_CSTs larger than they are.
2908 (check_string_literal): New check function for STRING_CSTs.
2909 (output_constant): Use it.
2911 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
2914 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
2916 2018-09-13 Richard Biener <rguenther@suse.de>
2918 PR tree-optimization/87263
2919 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
2920 (struct unwind_state): Add max_rpo field.
2921 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
2922 Compute max_rpo, the max RPO number a block can be backwards reached
2923 from. Re-write non-iterating mode to a RPO ordered worklist approach,
2924 separating it from the iterating mode.
2926 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
2928 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
2929 (rfs_decision): New scheduling decision.
2931 2018-09-13 Richard Biener <rguenther@suse.de>
2934 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
2935 (vn_nary_op_insert_pieces_predicated): Do not write useless
2936 valid_dominated_by_p entry outside of the allocated storage.
2938 2018-09-13 Omar Sandoval <osandov@osandov.com>
2939 Tom de Vries <tdevries@suse.de>
2942 * dwarf2out.c (is_c): New function.
2943 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
2945 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
2948 * config/aarch64/aarch64.md (*aarch64_bfxil):
2950 * config/aarch64/constraints.md (Ulc): Define.
2951 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
2953 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
2956 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
2958 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
2959 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
2960 aarch64_layout_frame call.
2961 (aarch64_expand_epilogue): Likewise.
2962 (aarch64_initial_elimination_offset): Likewise.
2963 (aarch64_get_separate_components): Likewise.
2964 (aarch64_use_return_insn_p): Likewise.
2965 (aarch64_layout_frame): Remove unneeded check.
2967 2018-09-13 Jakub Jelinek <jakub@redhat.com>
2969 * configure.ac: Only append
2970 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
2971 gcc_config_arguments if it was never reconfigured or last reconfigure
2972 was with different arguments.
2973 * configure: Regenerated.
2975 2018-09-13 Jakub Jelinek <jakub@redhat.com>
2976 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2979 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
2980 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
2982 2018-09-13 Jakub Jelinek <jakub@redhat.com>
2984 PR tree-optimization/87287
2985 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
2986 X % C == 0 to X % (unsigned) C == 0 optimization to ...
2987 * match.pd (X % C == 0): ... here. New optimization.
2989 2018-09-12 Jakub Jelinek <jakub@redhat.com>
2992 * expr.h (maybe_optimize_mod_cmp): Declare.
2993 * expr.c (mod_inv): New function.
2994 (maybe_optimize_mod_cmp): New function.
2995 (do_store_flag): Use it.
2996 * cfgexpand.c (expand_gimple_cond): Likewise.
2998 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
2999 Julian Brown <julian@codesourcery.com>
3002 * gimplify.c (gimplify_scan_omp_clauses): Set
3003 target_firstprivatize_array_bases in OpenACC parallel and kernels
3004 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
3005 OpenACC data regions.
3007 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
3009 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
3010 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
3012 2018-09-12 Richard Biener <rguenther@suse.de>
3014 PR tree-optimization/87280
3015 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
3016 edge but unreachable target.
3017 (do_rpo_vn): For conservatively handling a PHI only mark
3018 the backedge executable but not the block reachable.
3020 2018-09-12 Richard Biener <rguenther@suse.de>
3022 PR tree-optimization/87266
3023 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
3026 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
3028 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
3030 ("trunc<BFP:mode><DFP_ALL:mode>2")
3031 ("trunc<DFP_ALL:mode><BFP:mode>2")
3032 ("extend<BFP:mode><DFP_ALL:mode>2")
3033 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
3034 according to the target operand type.
3036 2018-09-12 Jakub Jelinek <jakub@redhat.com>
3037 Andreas Krebbel <krebbel@linux.ibm.com>
3039 PR tree-optimization/86844
3040 * gimple-ssa-store-merging.c
3041 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
3042 there are any overlapping stores in between them, make sure they are
3043 also coalesced or we give up completely.
3045 2018-09-12 Jakub Jelinek <jakub@redhat.com>
3048 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
3049 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
3051 2018-09-12 Tom de Vries <tdevries@suse.de>
3053 * common.opt (gdescribe-dies): Add option.
3054 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
3055 attribute for artifical and nameless decls.
3056 (dwarf2out_register_external_die): Add description attribute to
3057 external reference die.
3058 (add_desc_attribute): New functions.
3059 (gen_subprogram_die): Add description attribute to
3060 DW_TAG_call_site_parameter.
3061 * tree-pretty-print.c (print_generic_expr_to_str): New function.
3062 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
3063 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
3065 (@item -gdescribe-dies): Add.
3067 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
3069 * tree-vrp.c (vrp_shift_undefined_p): Remove.
3070 (extract_range_from_binary_expr_1: Call
3071 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
3072 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
3075 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
3077 * gimple-ssa-warn-alloca.c
3078 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
3079 field for ALLOCA_BOUND_*_LARGE.
3081 2018-09-11 Nathan Sidwell <nathan@acm.org>
3083 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
3085 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
3087 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
3088 for clobbers. Remove obsolete comment.
3090 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
3092 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
3093 mpxchk, mpxld and mpxst types.
3094 (define_attr length_immediate): Remove all processing of mpx types.
3095 (define_attr prefix_0f): Ditto.
3096 (define_attr memory): Ditto.
3098 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
3100 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
3101 (log<mode>2): Change operand 1 predicate to general_operand.
3102 Extend operand 1 to XFmode and generate logxf3 insn.
3103 (log10<mode>2): Change operand 1 predicate to general_operand.
3104 Extend operand 1 to XFmode and generate log10xf3 insn.
3105 (log2<mode>2): Change operand 1 predicate to general_operand.
3106 Extend operand 1 to XFmode and generate log2xf3 insn.
3107 (fyl2xp1_extend<mode>xf3_i387): Remove.
3108 (log1p<mode>2): Change operand 1 predicate to general_operand.
3109 Extend operand 1 to XFmode and generate log1pxf3 insn.
3110 (fxtract_extend<mode>xf3_i387): Remove.
3111 (logb<mode>2): Change operand 1 predicate to general_operand.
3112 Extend operand 1 to XFmode and generate logbxf3 insn.
3113 (ilogb<mode>2): Change operand 1 predicate to general_operand.
3114 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
3115 (significand<mode>2): Change operand 1 predicate to general_operand.
3116 Extend operand 1 to XFmode and generate significandxf3 insn.
3118 2018-09-11 Nathan Sidwell <nathan@acm.org>
3120 * gcc.c (perror_with_name, pfatal_with_name): Delete.
3121 (load_specs): Use fatal_error.
3122 (DELETE_IF_ORDINARY, process_command): Use error.
3123 (execute, run_attempt): Use fatal_error.
3125 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
3127 * diagnostic-core.h (sorry_at): New prototype.
3128 * diagnostic.c (sorry_at): New function.
3130 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
3132 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
3133 by zero as VR_UNDEFINED.
3135 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
3137 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
3138 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
3139 (<sincos>mode2): New expander.
3140 (sincos_extend<mode>xf3_i387): Remove insn pattern.
3141 (sincos -> sin, cos splitters): Remove splitter patterns.
3142 (sincos<mode>3): Change operand 2 predicate to general_operand.
3143 Extend operand 2 to XFmode and generate sincosxf3 insn.
3144 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
3145 Change operand 3 predicate to const1_operand.
3146 (fptan_extend<mode>xf4_i387): Remove insn pattern.
3147 (tanxf2): Update operands in the call to fptanxf4_i387.
3148 (tan<mode>2): Change operand 1 predicate to general_operand.
3149 Extend operand 1 to XFmode and generate tanxf3 insn.
3150 (atan2xf3): Rename from *fpatanxf3_i387.
3151 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
3152 (atan2xf3): Remove expander.
3153 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
3154 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
3155 (atan<mode>2): Change operand 1 predicate to general_operand.
3156 Extend operand 1 to XFmode and generate atanxf3 insn.
3158 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
3160 * config/i386/i386.md (x87/SSE constant load splitter): Use
3161 memory_operand instead of nonimmediate_operand for input operand
3164 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
3166 * config/i386/i386.md (float partial SSE register stall splitter): Move
3167 splitter near its instruction pattern.
3168 (float_extend partial SSE register stall splitter): Ditto.
3169 (float_truncate partial SSE register stall splitter): Ditto.
3171 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
3174 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
3175 to speculation_safe_value_not_needed.
3178 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
3179 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
3180 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
3181 leaf_function_p, instead use has_hard_reg_initial_val.
3183 2018-09-09 Nathan Sidwell <nathan@acm.org>
3185 * gcc.h (pfatal_with_name): Don't declare here.
3186 * gcc.c (pfatal_with_name): Make static.
3188 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
3190 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
3193 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
3195 PR rtl-optimization/85458
3196 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
3197 priority hook to reduce the priority of EXPR.
3199 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
3201 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
3202 DImode for x87 on 32bit targets. Conditionally disable x87 modes
3203 with X87_ENABLE_FLOAT. Remove preparation code.
3204 (*float<SWI48:mode><MODEF:mode>2): Rename from
3205 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
3206 math using "enabled" attribute.
3207 (*floatdi<MODEF:mode>2_i387): Rename from
3208 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
3209 enable for 32bit targets only.
3210 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
3212 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
3213 as operand 1 predicate. Rewrite as define_insn_and_split.
3214 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
3216 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
3218 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
3219 to fallthru to FLOAT case.
3221 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
3224 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
3225 around folding of vec_sl to handle out of range shift values.
3227 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
3229 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
3230 Update callers to gen_fix_trunc<mode>_i387_fisttp
3231 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
3232 nonimmediate_operand.
3233 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
3234 and corresponding splitters.
3235 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
3236 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
3237 (fix_truncdi_i387_with_temp): Remove insn pattern
3238 and corresponding splitters.
3239 (fix_trunc<mode>_i387): Change operand 0 predicate to
3240 nonimmediate_operand.
3241 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
3242 and corresponding splitters.
3243 (*fistdi2_1): Remove.
3245 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
3246 (lrintxfdi2): Remove expander. Reimplement as define_insn.
3247 (*fist<mode>2_1): Remove.
3248 (fist<mode>2): Ditto.
3249 (fist<mode>2_with_temp): Remove insn pattern and corresponding
3251 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
3252 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
3253 (fistdi2_<rounding>): Change operand 0 predicate to
3254 nonimmediate_operand.
3255 (fistdi2_<rounding>_with_temp): Remove insn pattern
3256 and corresponding splitters.
3257 (fist<mode>2_<rounding>): Change operand 0 predicate to
3258 nonimmediate_operand.
3259 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
3260 and corresponding splitters.
3262 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
3264 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
3266 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
3269 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
3271 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3272 early gimple folding of vec_splat().
3273 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
3274 * gimple-fold.h: Add an extern define for tree_vec_extract().
3276 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
3278 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
3279 wrappers around TREE_TYPE comparisons.
3281 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
3284 * config/s390/predicates.md: Add nonsym_memory_operand.
3285 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
3286 contains a SYMBOL_REF, load it into an intermediate pseudo.
3287 (s390_emit_compare_and_swap): Legitimize operand.
3288 * config/s390/s390.md: Use the new nonsym_memory_operand
3289 with UNSPECV_CAS patterns.
3291 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
3294 * config/s390/s390-passes.def: New file.
3295 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
3297 (make_pass_s390_early_mach): Add declaration.
3298 * config/s390/s390.c (make_pass_s390_early_mach):
3299 (s390_option_override): Remove dynamic registration.
3300 * config/s390/t-s390: Add s390-passes.def.
3302 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
3304 * config/s390/s390.c (s390_decompose_constant_pool_ref):
3305 Remove UNSPEC_LTREL_BASE check.
3306 (annotate_constant_pool_refs): Likewise.
3307 (find_constant_pool_ref): Likewise.
3308 (find_ltrel_base): Removed.
3309 (replace_ltrel_base): Removed.
3310 (s390_mainpool_finish): Remove replace_ltrel_base call.
3311 (s390_chunkify_start): Remove pending LTREL_BASE logic.
3312 (s390_chunkify_finish): Remove replace_ltrel_base call.
3313 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
3315 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
3318 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
3319 to speculation_safe_value_not_needed.
3321 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
3322 Bernd Schmidt <bernds_cb1@t-online.de>
3324 * config/nvptx/nvptx-opts.h: New file.
3325 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
3326 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
3328 (TARGET_SM35): New macro.
3329 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
3331 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
3333 (misa=): New option.
3334 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
3336 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
3338 * config/i386/i386.md (truncdfsf2): Remove expander.
3339 (truncdfsf2_with_temp): Ditto.
3340 (truncxf<mode>2): Ditto.
3341 (*truncdfsf_fast_mixed): Remove insn pattern.
3342 (*truncdfsf_fast_i387): Ditto.
3343 (*truncdfsf_mixed): Ditto.
3344 (*truncdfsf_i387): Ditto.
3345 (*truncdfsf2_i387_1): Ditto.
3346 (*truncxfsf2_mixed): Ditto.
3347 (*truncxfdf2_mixed): Ditto.
3348 (*truncxf<mode>2_i387_noop): Ditto. Update callers
3349 to call gen_truncxf<mode>2 instead.
3350 (*truncxf<mode>2_i387): Remove.
3351 (reg->reg splitters): Remove splitter pattern.
3352 (reg->mem splitters): Ditto.
3354 (truncdfsf2): New insn pattern.
3355 (truncxf<mode>2): Ditto.
3357 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3359 * tree-ssa-math-opts.c (is_mult_by): New function.
3360 (is_square_of): Use the above.
3361 (optimize_recip_sqrt): New function.
3362 (pass_cse_reciprocals::execute): Use the above.
3364 2018-09-05 Richard Biener <rguenther@suse.de>
3367 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
3368 to zero-init the emplaced vec.
3370 2018-09-05 Martin Liska <mliska@suse.cz>
3372 PR tree-optimization/87205
3373 * tree-switch-conversion.c (pass_lower_switch::execute):
3374 Group cases for switch statements.
3376 2018-09-05 Richard Biener <rguenther@suse.de>
3378 PR tree-optimization/87217
3379 * tree-ssa-sccvn.c (vuse_valueize): New.
3380 (vn_reference_lookup_pieces): Use it.
3381 (vn_reference_lookup): Likewise.
3383 2018-09-05 Nathan Sidwell <nathan@acm.org>
3386 * stor-layout.c (place_field): Scan forwards to check last
3387 bitfield when ms_bitfield_placement is in effect.
3389 2018-09-05 Richard Biener <rguenther@suse.de>
3392 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
3395 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
3396 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
3398 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
3399 * config.gcc (extra_objs): Build it.
3400 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
3402 * config/aarch64/aarch64-passes.def
3403 (pass_tag_collision_avoidance): New pass.
3404 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
3405 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
3406 (aarch64_classify_address): Remove static qualifier.
3407 (aarch64_address_info, aarch64_address_type): Move to...
3408 * config/aarch64/aarch64-protos.h: ... here.
3409 (make_pass_tag_collision_avoidance): New function.
3410 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
3413 2018-09-05 Martin Liska <mliska@suse.cz>
3415 * doc/gcov.texi: Update documentation of humar
3417 * gcov.c (format_count): Print one decimal place, it provides
3418 more fine number of situations like '1G' vs. '1.4G'.
3420 2018-09-05 Martin Liska <mliska@suse.cz>
3423 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
3424 * optc-gen.awk: Allow 'Var' for Deprecated options in order
3425 to generate a MASK value.
3427 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
3430 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
3431 if frame pointer isn't used.
3432 (compute_frame_pointer_to_fb_displacement): Likewise.
3434 2018-09-04 Jakub Jelinek <jakub@redhat.com>
3437 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
3438 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
3439 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
3440 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
3441 and OPTION_MASK_ISA_XSAVEC_UNSET.
3443 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
3445 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
3446 XOR operations in NAND case.
3448 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
3450 * wide-int-range.cc (wide_int_range_convert): New.
3451 * wide-int-range.h (wide_int_range_convert): New.
3452 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
3453 code into wide_int_range_convert.
3454 (extract_range_into_wide_ints): Do not munge anti range constants
3455 into the entire domain. Just return the range back.
3457 2018-09-04 Martin Liska <mliska@suse.cz>
3459 * genmatch.c (output_line_directive): Add new argument
3461 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
3463 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
3465 * doc/invoke.texi (Option Summary): Add whitespace.
3467 * doc/invoke.texi (Option Summary): Add -Waligned-new.
3469 2018-09-04 Richard Biener <rguenther@suse.de>
3471 PR tree-optimization/87211
3472 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
3473 backedge value we're supposed to treat as VARYING also number
3474 the PHI to VARYING in case it got a different value-number already.
3476 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
3478 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
3479 (extract_range_from_binary_expr_1): Do not call
3480 vrp_can_optimize_bit_op.
3481 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
3483 (wide_int_range_get_mask_and_bounds): New.
3484 (wide_int_range_optimize_bit_op): New.
3485 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
3486 (wide_int_range_bit_and): Same.
3487 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
3488 (wide_int_range_optimize_bit_op): New.
3489 (wide_int_range_get_mask_and_bounds): New.
3491 2018-09-04 Richard Biener <rguenther@suse.de>
3493 PR tree-optimization/87176
3494 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
3495 variable. When value-numbering a virtual PHI node make sure
3496 to not value-number to the backedge value.
3498 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
3500 * doc/extend.texi (Long Long, Hex Floats): Document support for
3501 long long and hex floats in more recent versions of ISO C++.
3503 2018-09-03 Richard Biener <rguenther@suse.de>
3505 PR tree-optimization/87177
3506 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
3509 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
3511 * bb-reorder.c (edge_order): Convert to C-qsort-style
3512 tri-state comparator.
3513 (reorder_basic_blocks_simple): Change std::stable_sort to
3516 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
3518 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
3519 tri-state comparator.
3520 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
3522 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
3524 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
3525 (mergesort): ... here as maximum count for using netsort.
3526 (gcc_qsort): Set nlim to 3 if stable sort is requested.
3527 (gcc_stablesort): New.
3528 * system.h (gcc_stablesort): Declare.
3530 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
3532 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
3533 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
3534 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
3536 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
3538 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
3539 lxsdx and stxsdx alternatives.
3540 (*mov<mode>_hardfloat64): Ditto.
3541 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
3543 2018-09-03 Richard Biener <rguenther@suse.de>
3545 PR tree-optimization/87200
3546 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
3549 2018-09-03 Martin Liska <mliska@suse.cz>
3551 PR tree-optimization/87201
3552 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
3553 Fix parenthesis in an expression.
3555 2018-09-03 Richard Biener <rguenther@suse.de>
3557 PR tree-optimization/87197
3558 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
3559 visited. CSE the VN_INFO hashtable lookup.
3561 PR tree-optimization/87169
3562 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
3563 iterating make sure there's no extra backedges from irreducible
3564 regions feeding the header. Mark the destination block
3567 2018-09-03 Martin Liska <mliska@suse.cz>
3570 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
3571 * common/common-targhooks.c (default_get_valid_option_values):
3573 * common/common-targhooks.h (default_get_valid_option_values):
3575 * common/config/i386/i386-common.c: Move processor_target_table
3577 (ix86_get_valid_option_values): New function.
3578 (TARGET_GET_VALID_OPTION_VALUES): New macro.
3579 * config/i386/i386.c (struct ptt): Move to i386-common.c.
3580 (PTA_*): Move all defined masks into i386-common.c.
3581 (ix86_function_specific_restore): Use new processor_cost_table.
3582 * config/i386/i386.h (struct ptt): Moved from i386.c.
3583 (struct pta): Likewise.
3584 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
3585 * doc/tm.texi.in: Likewise.
3586 * opt-suggestions.c (option_proposer::suggest_option):
3587 Pass prefix to build_option_suggestions.
3588 (option_proposer::get_completions): Likewise.
3589 (option_proposer::build_option_suggestions): Use the new target
3591 * opts.c (struct option_help_tuple): New struct.
3592 (print_filtered_help): Use the new target hook.
3594 2018-09-03 Martin Liska <mliska@suse.cz>
3597 * predict.c (set_even_probabilities): Add likely_edges
3598 argument and handle cases where we have precisely one
3600 (combine_predictions_for_bb): Catch also likely_edges.
3601 (tree_predict_by_opcode): Handle gswitch statements.
3602 * tree-cfg.h (find_case_label_for_value): New declaration.
3603 (find_taken_edge_switch_expr): Likewise.
3604 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
3605 Find pivot in decision tree based on probabily, not by number of
3608 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
3610 * doc/standards.texi (Standards): Update Objective-C reference.
3612 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
3614 * doc/install.texi (Prerequisites): Update link for MPC.
3616 2018-09-01 Michael Matz <matz@suse.de>
3618 PR tree-optimization/87074
3619 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
3620 PHIs for outer-loop uses.
3622 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
3624 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
3625 * doc/invoke.texi (C Dialect Options): Ditto.
3627 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
3629 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
3631 2018-08-31 Richard Biener <rguenther@suse.de>
3633 PR tree-optimization/87168
3634 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
3635 (rpo_elim::eliminate_avail): When OP was not visited it must
3638 2018-08-31 David Malcolm <dmalcolm@redhat.com>
3640 * tree-vrp.c (copy_value_range): Convert param "from" from
3641 "value_range *" to "const value_range *".
3642 (range_is_null): Likewise for param "vr".
3643 (range_int_cst_p): Likewise.
3644 (range_int_cst_singleton_p): Likewise.
3645 (symbolic_range_p): Likewise.
3646 (value_ranges_intersect_p): Likewise for both params.
3647 (value_range_nonnegative_p): Likewise for param "vr".
3648 (value_range_constant_singleton): Likewise.
3649 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
3650 (extract_range_into_wide_ints): Likewise for param "vr".
3651 (extract_range_from_multiplicative_op): Likewise for params "vr0"
3653 (vrp_can_optimize_bit_op): Likewise.
3654 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
3656 (extract_range_from_unary_expr): Likewise.
3657 (debug_value_range): Likewise for param "vr".
3658 (value_range::dump): Add "const" qualifier.
3659 (vrp_prop::check_array_ref): Convert local "vr" from
3660 "value_range *" to "const value_range *".
3661 (vrp_prop::check_mem_ref): Likewise.
3662 (vrp_prop::visit_stmt): Likewise for local "old_vr".
3663 (vrp_intersect_ranges_1): Likewise for param "vr_1".
3664 (vrp_intersect_ranges): Likewise.
3665 (simplify_stmt_for_jump_threading): Likewise for local "vr".
3666 (vrp_prop::vrp_finalize): Likewise.
3667 * tree-vrp.h (value_range::dump): Add "const" qualifier.
3668 (vrp_intersect_ranges): Add "const" qualifier to params as above.
3669 (extract_range_from_unary_expr): Likewise.
3670 (value_range_constant_singleton): Likewise.
3671 (symbolic_range_p): Likewise.
3672 (copy_value_range): Likewise.
3673 (extract_range_from_binary_expr_1): Likewise.
3674 (range_int_cst_p): Likewise.
3675 (vrp_set_zero_nonzero_bits): Likewise.
3676 (range_int_cst_singleton_p): Likewise.
3678 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
3680 * config/aarch64/arm_neon.h (vabsd_s64): New.
3681 (vnegd_s64): Likewise.
3683 2018-08-31 Martin Jambor <mjambor@suse.cz>
3685 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
3687 2018-08-31 Martin Liska <mliska@suse.cz>
3689 * ipa-icf.c (sem_item::add_type): Use
3690 sem_item::m_type_hash_cache.
3691 * ipa-icf.h: Move the cache from sem_item_optimizer
3694 2018-08-31 Nathan Sidwell <nathan@acm.org>
3696 * doc/extend.texi (Backwards Compatibility): Remove implicit
3697 extern C leeway of () being (...).
3699 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3701 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
3703 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
3707 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
3709 2018-08-31 Jakub Jelinek <jakub@redhat.com>
3712 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
3713 gen_int_mode. Formatting fixes.
3715 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
3717 * target.def (custom_function_descriptors): Improve documentation.
3718 * doc/tm.texi.in (Trampolines): Expand discussion of function
3719 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
3720 beginning of the section.
3721 * doc/tm.texi: Regenerated.
3723 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
3725 * cfg.h (class auto_edge_flag): Spell out the template-id of the
3726 base class in the initializer list. This is a workaround for
3727 building with older GCC.
3728 (class auto_bb_flag): Likewise.
3730 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
3732 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
3733 (altivec_vcmpequ<VI_char>_p): Remove star.
3734 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
3736 (expand_strncmp_vec_sequence): New function.
3737 (emit_final_str_compare_vec): New function.
3738 (expand_strn_compare): Add support for vector strncmp.
3739 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
3740 length specification to bytes.
3741 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
3742 (vcmpnezb_p): New pattern.
3743 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
3744 for option -mstring-compare-inline-limit.
3746 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
3748 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
3749 (PTA_SKYLAKE): Add PTA_AES.
3750 (PTA_GOLDMONT): Likewise.
3752 2018-08-29 Jan Hubicka <jh@suse.cz>
3755 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
3756 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
3758 2018-08-29 Jan Hubicka <jh@suse.cz>
3760 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
3762 (hash_tree): Do not visit TYPE_STUB_DECL.
3763 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
3764 stream TYPE_STUB_DECL.
3765 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
3766 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
3767 after free_lang_data.
3768 (type_in_anonymous_namespace_p): Likewise.
3770 2018-08-29 Jan Hubicka <jh@suse.cz>
3772 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
3773 comment that it has to be even number.
3774 (class sreal): Change m_sig type to int32_t.
3775 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
3776 int64_t for temporary calculations.
3777 (sreal_verify_basics): Drop one bit from minimum and maximum.
3779 2018-08-30 Richard Biener <rguenther@suse.de>
3781 PR tree-optimization/87147
3782 * tree-ssa-sccvn.c (SSA_VISITED): New function.
3783 (visit_phi): When the degenerate result is from the backedge and
3784 we didn't visit its definition yet drop to VARYING.
3785 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
3787 2018-08-29 Jan Hubicka <jh@suse.cz>
3789 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
3791 (hash_tree): Likewise.
3793 2018-08-29 Jan Hubicka <jh@suse.cz>
3795 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
3796 and TYPE_NEXT_REF_TO.
3798 2018-08-29 Jan Hubicka <jh@suse.cz>
3800 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
3801 comment that it has to be even number.
3802 (class sreal): Change m_sig type to int32_t.
3803 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
3804 int64_t for temporary calculations.
3805 (sreal_verify_basics): Drop one bit from minimum and maximum.
3807 2018-08-30 Tamar Christina <tamar.christina@arm.com>
3809 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
3811 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
3814 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
3815 if to_add is negative.
3817 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
3820 * builtins.c (c_strlen): Improve range checks.
3822 2018-08-29 Martin Sebor <msebor@redhat.com>
3823 Jeff Law <law@redhat.com>
3825 PR tree-optimization/86714
3826 PR tree-optimization/86711
3827 * builtins.c (c_strlen): Add arguments to call to string_constant.
3828 * expr.c (string_constant): Add argument. Detect missing nul
3829 terminator and outermost declaration it's missing in.
3830 * expr.h (string_constant): Add argument.
3831 * fold-const.c (read_from_constant_string): Add arguments to call to
3833 (c_getstr): Likewise.
3834 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
3836 * tree-ssa-strlen.c (get_stridx): Likewise.
3838 2018-08-29 Jan Hubicka <jh@suse.cz>
3840 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
3841 Do not stream DECL_VINDEX.
3842 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
3843 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
3844 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
3846 2018-08-29 Richard Biener <rguenther@suse.de>
3848 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
3849 virtual operands that are not default defs to honor region
3851 (rpo_vn_valueize): Remove ineffective code here.
3853 2018-08-29 Richard Biener <rguenther@suse.de>
3855 PR tree-optimization/87132
3856 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
3857 when skipping defs reachable over backedges.
3859 2018-08-29 Richard Biener <rguenther@suse.de>
3861 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
3862 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
3863 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
3864 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
3865 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
3866 (vn_reference_lookup_3): Remove use of const_parms.
3867 (free_rpo_vn): Do not free const_parms.
3868 (do_rpo_vn): Do not call init_const_parms.
3869 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
3870 SSA_NAME_POINTS_TO_READONLY_MEMORY.
3871 (call_may_clobber_ref_p_1): Likewise.
3873 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
3876 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
3878 (-ftree-scev-cprop): Document.
3880 2018-08-29 Jan Hubicka <jh@suse.cz>
3882 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
3884 (sreal constructor): Update.
3885 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
3886 sreal:operator/): Update.
3888 2018-08-29 Martin Liska <mliska@suse.cz>
3890 * tree-switch-conversion.c (switch_conversion::expand):
3891 Strenghten assumption about gswitch statements.
3893 2018-08-29 Richard Biener <rguenther@suse.de>
3895 PR tree-optimization/87117
3896 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
3897 re-value-number released SSA VDEFs.
3899 2018-08-29 Richard Biener <rguenther@suse.de>
3901 PR tree-optimization/87126
3902 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
3904 2018-08-28 Jim Wilson <jimw@sifive.com>
3906 * config/riscv/pic.md: Rewrite.
3907 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
3909 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
3910 (SOFTF, default_load, softload, softstore): New.
3912 2018-08-28 Jeff Law <law@redhat.com>
3914 * fold-const.c (fold_binary_loc): Remove recently added assert.
3916 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
3918 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
3919 to OP parmeter of generated function.
3921 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
3923 PR tree-optimization/87009
3924 * match.pd: Add boolean optimizations.
3926 2018-08-28 Martin Sebor <msebor@redhat.com>
3929 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
3930 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
3931 (pass_walloca::gate): Use it.
3932 (alloca_call_type): Same.
3933 (pass_walloca::execute): Same.
3934 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
3936 2018-08-28 David Malcolm <dmalcolm@redhat.com>
3938 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
3939 GCC_VERSION for usage of "__gcc_dump_printf__" format from
3942 2018-08-28 Richard Biener <rguenther@suse.de>
3944 PR tree-optimization/87124
3945 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
3946 constants before looking up avail.
3948 2018-08-28 Jakub Jelinek <jakub@redhat.com>
3951 * calls.c (maybe_warn_nonstring_arg): Punt early if
3952 warn_stringop_overflow is zero. Don't call get_range_strlen
3953 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
3954 Swap comparison operands to have constants on rhs. Only use
3955 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
3956 increment lenrng[0].
3958 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
3960 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
3961 use of tree_to_shwi. Remove duplicated test for the size being
3962 a whole number of bytes.
3964 2018-08-28 Richard Biener <rguenther@suse.de>
3966 PR tree-optimization/87117
3967 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
3968 Handle removed stmt without LHS (GIMPLE_NOP).
3970 2018-08-28 Richard Biener <rguenther@suse.de>
3972 PR tree-optimization/87117
3973 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
3974 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
3976 2018-08-28 Richard Biener <rguenther@suse.de>
3978 PR tree-optimization/87117
3979 * tree-ssa-pre.c (compute_avail): Do not make expressions
3980 with predicated values available.
3981 (get_expr_value_id): Assert we do not run into predicated value
3984 2018-08-28 Richard Biener <rguenther@suse.de>
3986 PR tree-optimization/87117
3987 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
3988 get virtual operands.
3989 (get_expr_operands): Handle STRING_CST like other decls.
3991 2018-08-28 Martin Liska <mliska@suse.cz>
3993 * tree.h: Update documentation of fndecl_built_in_p
3997 2018-08-27 Jeff Law <law@redhat.com>
3998 PR tree-optimization/87110
3999 * tree-ssa-dse.c (compute_trims): Handle non-constant
4002 2018-08-27 Martin Sebor <msebor@redhat.com>
4004 PR tree-optimization/86914
4005 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
4007 2018-08-27 Martin Sebor <msebor@redhat.com>
4009 PR tree-optimization/87112
4010 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
4011 the type of the bound argument.
4013 2018-08-27 Jeff Law <law@redhat.com>
4015 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
4016 type does not have a TYPE_SIZE_UNIT.
4018 2018-08-27 Steve Ellcey <sellcey@cavium.com>
4020 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
4021 with include of backend.h.
4023 2018-08-27 Richard Biener <rguenther@suse.de>
4025 PR tree-optimization/86927
4026 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
4027 use const cond reduction code.
4029 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
4031 PR tree-optimization/85758
4032 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
4034 2018-08-27 David Malcolm <dmalcolm@redhat.com>
4037 * diagnostic-show-locus.c (class layout_range): Update for
4038 conversion of show_caret_p to a tri-state.
4039 (layout_range::layout_range): Likewise.
4040 (make_range): Likewise.
4041 (layout::maybe_add_location_range): Likewise.
4042 (layout::should_print_annotation_line_p): Don't show annotation
4043 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
4044 (layout::get_state_at_point): Update for conversion of
4045 show_caret_p to a tri-state. Bail out early for
4046 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
4047 underlining or source colorization.
4048 (gcc_rich_location::add_location_if_nearby): Update for conversion
4049 of show_caret_p to a tri-state.
4050 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
4051 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
4053 (selftest::test_one_liner_labels): Likewise.
4054 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
4055 conversion of show_caret_p to a tri-state.
4056 * pretty-print.c (text_info::set_location): Likewise.
4057 * pretty-print.h (text_info::set_location): Likewise.
4058 * substring-locations.c (format_warning_n_va): Likewise.
4059 * tree-diagnostic.c (default_tree_printer): Likewise.
4060 * tree-pretty-print.c (newline_and_indent): Likewise.
4062 2018-08-27 David Malcolm <dmalcolm@redhat.com>
4065 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
4066 line above for line-insertion fix-it hints.
4067 (selftest::test_fixit_insert_containing_newline): Update the
4068 expected results, and add a test with line-numbering enabled.
4070 2018-08-27 Martin Liska <mliska@suse.cz>
4073 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
4074 params with DECL_HAS_VALUE_EXPR_P.
4076 2018-08-27 Martin Liska <mliska@suse.cz>
4078 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
4079 selected expansion strategy.
4081 2018-08-27 Martin Liska <mliska@suse.cz>
4083 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
4084 * builtins.c (is_builtin_fn): Likewise.
4085 * attribs.c (diag_attr_exclusions): Use new function
4086 fndecl_built_in_p and remove check for FUNCTION_DECL if
4088 (builtin_mathfn_code): Likewise.
4089 (fold_builtin_expect): Likewise.
4090 (fold_call_expr): Likewise.
4091 (fold_builtin_call_array): Likewise.
4092 (fold_call_stmt): Likewise.
4093 (set_builtin_user_assembler_name): Likewise.
4094 (is_simple_builtin): Likewise.
4095 * calls.c (gimple_alloca_call_p): Likewise.
4096 (maybe_warn_nonstring_arg): Likewise.
4097 * cfgexpand.c (expand_call_stmt): Likewise.
4098 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
4099 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
4100 (cgraph_node::verify_node): Likewise.
4101 * cgraphclones.c (build_function_decl_skip_args): Likewise.
4102 (cgraph_node::create_clone): Likewise.
4103 * config/arm/arm.c (arm_insert_attributes): Likewise.
4104 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
4105 * dse.c (scan_insn): Likewise.
4106 * expr.c (expand_expr_real_1): Likewise.
4107 * fold-const.c (operand_equal_p): Likewise.
4108 (fold_binary_loc): Likewise.
4109 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
4110 * gimple-low.c (lower_stmt): Likewise.
4111 * gimple-pretty-print.c (dump_gimple_call): Likewise.
4112 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
4114 * gimple.c (gimple_build_call_from_tree): Likewise.
4115 (gimple_call_builtin_p): Likewise.
4116 (gimple_call_combined_fn): Likewise.
4117 * gimplify.c (gimplify_call_expr): Likewise.
4118 (gimple_boolify): Likewise.
4119 (gimplify_modify_expr): Likewise.
4120 (gimplify_addr_expr): Likewise.
4121 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
4122 * ipa-cp.c (determine_versionability): Likewise.
4123 * ipa-fnsummary.c (compute_fn_summary): Likewise.
4124 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
4125 * ipa-split.c (visit_bb): Likewise.
4126 (split_function): Likewise.
4127 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
4128 * lto-cgraph.c (input_node): Likewise.
4129 * lto-streamer-out.c (write_symbol): Likewise.
4130 * omp-low.c (setjmp_or_longjmp_p): Likewise.
4131 (lower_omp_1): Likewise.
4132 * predict.c (strip_predict_hints): Likewise.
4133 * print-tree.c (print_node): Likewise.
4134 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
4135 * trans-mem.c (is_tm_irrevocable): Likewise.
4136 (is_tm_load): Likewise.
4137 (is_tm_simple_load): Likewise.
4138 (is_tm_store): Likewise.
4139 (is_tm_simple_store): Likewise.
4140 (is_tm_abort): Likewise.
4141 (tm_region_init_1): Likewise.
4142 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
4143 * tree-cfg.c (verify_gimple_call): Likewise.
4144 (move_stmt_r): Likewise.
4145 (stmt_can_terminate_bb_p): Likewise.
4146 * tree-eh.c (lower_eh_constructs_2): Likewise.
4147 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
4148 * tree-inline.c (remap_gimple_stmt): Likewise.
4149 (copy_bb): Likewise.
4150 (estimate_num_insns): Likewise.
4151 (fold_marked_statements): Likewise.
4152 * tree-sra.c (scan_function): Likewise.
4153 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
4154 (optimize_stack_restore): Likewise.
4155 (pass_fold_builtins::execute): Likewise.
4156 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
4157 (mark_all_reaching_defs_necessary_1): Likewise.
4158 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
4159 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
4160 (pass_forwprop::execute): Likewise.
4161 * tree-ssa-loop-im.c (stmt_cost): Likewise.
4162 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
4163 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
4164 * tree-ssa-strlen.c (get_string_length): Likewise.
4165 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
4166 (find_func_aliases_for_call): Likewise.
4167 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
4168 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
4169 * tree-tailcall.c (find_tail_calls): Likewise.
4170 * tree.c (need_assembler_name_p): Likewise.
4171 (free_lang_data_in_decl): Likewise.
4172 (get_call_combined_fn): Likewise.
4173 * ubsan.c (is_ubsan_builtin_p): Likewise.
4174 * varasm.c (incorporeal_function_p): Likewise.
4175 * tree.h (DECL_BUILT_IN): Remove and replace with
4177 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
4178 (fndecl_built_in_p): New.
4180 2018-08-27 Martin Liska <mliska@suse.cz>
4182 PR tree-optimization/86847
4183 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
4184 Dump also subtree probability.
4185 (switch_decision_tree::do_jump_if_equal): New function.
4186 (switch_decision_tree::emit_case_nodes): Handle special
4187 situations in balanced tree that can be emitted much simpler.
4188 Fix calculation of probabilities that happen in tree expansion.
4189 * tree-switch-conversion.h (struct cluster): Add
4191 (struct simple_cluster): Likewise.
4192 (struct case_tree_node): Add new function has_child.
4193 (do_jump_if_equal): New.
4195 2018-08-27 Martin Liska <mliska@suse.cz>
4197 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
4198 Add new argument to bit_test_cluster constructor.
4199 (bit_test_cluster::emit): Set bits really number of values
4201 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
4202 probability argument.
4203 * tree-switch-conversion.h (struct bit_test_cluster):
4204 Add m_handles_entire_switch.
4206 2018-08-27 Martin Liska <mliska@suse.cz>
4208 PR tree-optimization/86702
4209 * tree-switch-conversion.c (jump_table_cluster::emit):
4210 Make probabilities even for values in jump table
4211 according to number of cases handled.
4212 (switch_decision_tree::compute_cases_per_edge): Pass
4213 argument to reset_out_edges_aux function.
4214 (switch_decision_tree::analyze_switch_statement): Likewise.
4215 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
4218 2018-08-27 Martin Liska <mliska@suse.cz>
4220 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
4221 cfun argument explicitly.
4222 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
4223 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
4224 function gimple_switch_default_bb.
4225 (convert_switch_statements):
4227 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
4228 * stmt.c (label_to_block_fn): Use label_to_block and pass
4229 cfun argument explicitly and use gimple_switch_label_bb.
4230 (expand_case): Likewise.
4231 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
4232 cfun argument explicitly. Likewise.
4233 (make_edges_bb): Likewise.
4234 (make_cond_expr_edges): Likewise.
4235 (get_cases_for_edge): Likewise.
4236 (make_gimple_switch_edges): Likewise.
4237 (label_to_block_fn): Likewise.
4238 (label_to_block): Likewise.
4239 (make_goto_expr_edges): Likewise.
4240 (make_gimple_asm_edges): Likewise.
4241 (main_block_label): Likewise.
4242 (group_case_labels_stmt): Likewise.
4243 (find_taken_edge_computed_goto): Likewise.
4244 (find_taken_edge_switch_expr): Likewise.
4245 (gimple_verify_flow_info): Likewise.
4246 (gimple_redirect_edge_and_branch): Likewise.
4247 (gimple_switch_label_bb): New function.
4248 (gimple_switch_default_bb): Likewise.
4249 (gimple_switch_edge): Likewise.
4250 (gimple_switch_default_edge): Likewise.
4251 * tree-cfg.h (label_to_block_fn): Remove and replace ...
4252 (label_to_block): ... with this.
4253 (gimple_switch_label_bb): New.
4254 (gimple_switch_default_bb): Likewise.
4255 (gimple_switch_edge): Likewise.
4256 (gimple_switch_default_edge): Likewise.
4257 * tree-cfgcleanup.c (convert_single_case_switch): Use
4258 new gimple functions and pass new argument to label_to_block.
4259 (cleanup_control_flow_bb):
4260 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
4261 cfun argument explicitly.
4262 (make_eh_edges): Likewise.
4263 (redirect_eh_dispatch_edge): Likewise.
4264 (lower_resx): Likewise.
4265 (lower_eh_dispatch): Likewise.
4266 (maybe_remove_unreachable_handlers): Likewise.
4267 (unsplit_eh): Likewise.
4268 (cleanup_empty_eh): Likewise.
4269 (verify_eh_edges): Likewise.
4270 (verify_eh_dispatch_edge): Likewise.
4271 * tree-ssa-dom.c (record_edge_info): Likewise.
4272 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
4273 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
4274 (thread_through_normal_block): Likewise.
4275 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
4276 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
4277 * tree-switch-conversion.c (switch_conversion::collect): Use new
4279 (switch_conversion::check_final_bb): Likewise.
4280 (switch_conversion::gather_default_values): Pass new argument
4282 (switch_conversion::build_constructors): Likewise.
4283 (switch_decision_tree::compute_cases_per_edge): Use new
4284 gimple_switch_edge function.
4285 (switch_decision_tree::analyze_switch_statement): Pass new argument
4287 (switch_decision_tree::try_switch_expansion): Use
4288 gimple_switch_default_edge.
4289 * tree-vrp.c (find_switch_asserts): Pass new argument
4291 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
4292 (vr_values::simplify_switch_using_ranges): Likewise.
4294 2018-08-27 Richard Biener <rguenther@suse.de>
4296 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
4297 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
4299 * tree-ssa-sccvn.h (struct vn_pval): New structure.
4300 (struct vn_nary_op_s): Add unwind_to member. Add
4301 predicated_values flag and put result into a union together
4302 with a linked list of vn_pval.
4303 (struct vn_ssa_aux): Add name member to make maintaining
4304 a map of SSA name to vn_ssa_aux possible. Remove no longer
4305 needed info, dfsnum, low, visited, on_sccstack, use_processed
4306 and range_info_anti_range_p members.
4307 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
4308 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
4310 (vn_valueize): New global.
4311 (vn_context_bb): Likewise.
4312 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
4313 VN_INFO_PTR_INFO): Remove.
4314 * tree-ssa-sccvn.c: ... (rewrite)
4315 (pass_fre::execute): For -O2+ initialize loops and run
4316 RPO VN in optimistic mode (iterating). For -O1 and -Og
4317 run RPO VN in non-optimistic mode.
4318 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
4319 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
4320 * doc/invoke.texi (sccvn-max-scc-size): Remove.
4321 (rpo-vn-max-loop-depth): Document.
4322 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
4323 when valuezing the VUSE signals we walked out of the region.
4324 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
4325 (phi_translate): Set VN context block to use for availability
4327 (compute_avail): Likewise.
4328 (pre_valueize): New function.
4329 (pass_pre::execute): Adjust to the RPO VN API.
4331 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
4332 (propagate_constants_for_unrolling): Remove.
4333 (tree_unroll_loops_completely): Perform value-numbering
4334 on the unrolled bodies loop parent.
4336 2018-08-27 Richard Biener <rguenther@suse.de>
4338 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
4339 for partial antic compute.
4341 2018-08-27 Jakub Jelinek <jakub@redhat.com>
4343 PR rtl-optimization/87065
4344 * combine.c (simplify_if_then_else): Formatting fix.
4345 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
4347 (known_cond): Don't return const_true_rtx for vector modes. Use
4348 CONST0_RTX instead of const0_rtx. Formatting fixes.
4350 2018-08-27 Martin Liska <mliska@suse.cz>
4352 PR gcov-profile/87069
4353 * gcov.c (process_file): Record files already processed
4354 and warn about a file being processed multiple times.
4356 2018-08-27 Martin Liska <mliska@suse.cz>
4359 * config/aarch64/aarch64.c (aarch64_override_options_internal):
4360 Set default values for x_aarch64_*_string strings.
4361 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
4362 prefix. For -mabi do not print '=ABI' in help and use
4363 <option_value> format for -msve-vector-bits and -moverride
4366 2018-08-26 Jeff Law <law@redhat.com>
4368 * config/mips/frame-header-opt.c: Include "backend.h" rather than
4371 2018-08-26 Marek Polacek <polacek@redhat.com>
4373 PR c++/87029, Implement -Wredundant-move.
4374 * doc/invoke.texi: Document -Wredundant-move.
4376 2018-08-25 Martin Sebor <msebor@redhat.com>
4378 PR tree-optimization/87059
4379 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
4380 to the same type as the other.
4381 * fold-const.c (fold_binary_loc): Assert expectation.
4383 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
4385 * config/darwin.c (machopic_legitimize_pic_address): Clean up
4386 extraneous parentheses, dead code section and formatting.
4388 2018-08-24 David Malcolm <dmalcolm@redhat.com>
4391 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
4392 wide enough for jumps in the line-numbering to be visible.
4393 (layout::print_gap_in_line_numbering): New member function.
4394 (layout::calculate_line_spans): When using line numbering, merge
4395 line spans that are only 1 line apart.
4396 (diagnostic_show_locus): When printing line numbers, show gaps in
4397 line numbering directly, rather than printing headers.
4398 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
4399 line-numbering with multiple line spans.
4400 (selftest::test_fixit_insert_containing_newline_2): Add test of
4401 line-numbering, in which the spans are close enough to be merged.
4403 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
4405 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
4406 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
4407 * tree-vrp.c (range_is_nonnull): Remove.
4408 (range_includes_zero_p): Accept value_range instead of min/max.
4409 (extract_range_from_binary_expr_1): Do not early bail on
4411 Use range_includes_zero_p instead of range_is_nonnull.
4412 (extract_range_from_unary_expr): Use range_includes_zero_p instead
4413 of range_is_nonnull.
4414 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
4415 special case VR_ANTI_RANGE.
4416 (vrp_finalize): Same.
4417 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
4419 (range_is_nonnull): Remove.
4420 * vr-values.c (vrp_stmt_computes_nonzero): Use
4421 range_includes_zero_p instead of range_is_nonnull.
4422 (extract_range_basic): Pass value_range to range_includes_zero_p
4423 instead of range_is_nonnull.
4425 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
4427 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
4428 * emit-rtl.h (rtl_data): Remove return_bnd.
4429 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
4430 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
4431 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
4432 (POINTER_BOUNDS_MODE): Remove definition.
4433 (make_pointer_bounds_mode): Remove.
4434 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
4435 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
4436 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
4437 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
4438 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
4439 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
4440 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
4442 * config/i386/i386-modes.def (BND32, BND64): Remove.
4443 * config/i386/i386.c (dbx_register_map): Remove bound registers.
4444 (dbx64_register_map): Ditto.
4445 (svr4_dbx_register_map): Ditto.
4446 (indirect_thunk_bnd_needed): Remove.
4447 (indirect_thunks_bnd_used): Ditto.
4448 (indirect_return_bnd_needed): Ditto.
4449 (indirect_return_via_cx_bnd): Ditto.
4450 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
4451 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
4452 (output_indirect_thunk): Ditto. Remove need_prefix argument.
4453 (output_indirect_thunk_function): Remove handling of
4454 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
4455 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
4456 (ix86_save_reg): Remove handling of crtl->return_bnd.
4457 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
4458 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
4459 and UNSPEC_BNDLX_ADDR.
4460 (ix86_output_indirect_branch_via_reg): Remove handling of
4461 indirect_thunk_prefix_bnd.
4462 (ix86_output_indirect_branch_via_push): Ditto.
4463 (ix86_output_function_return): Ditto.
4464 (ix86_output_indirect_function_return): Ditto.
4465 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
4466 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
4467 (CALL_USED_REGISTERS): Ditto.
4468 (REG_ALLOC_ORDER): Update for removal of bound registers.
4469 (HI_REGISTER_NAMES): Ditto.
4470 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
4471 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
4472 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
4473 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
4474 (FIRST_PSEUDO_REG): Update.
4475 (BND): Remove mode iterator.
4476 * config/i386/predicates.md (bnd_mem_operator): Remove.
4478 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
4480 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
4483 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
4485 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
4486 the case in which the permute needs only a single element and
4487 repeats for every vector of the result. Extend that case to
4488 handle variable-length vectors.
4489 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
4491 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
4494 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
4495 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
4497 2018-08-24 Richard Biener <rguenther@suse.de>
4499 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
4500 bb_flags_allocated members.
4501 (auto_flag): New RAII class for allocating flags.
4502 (auto_edge_flag): New RAII class for allocating edge flags.
4503 (auto_bb_flag): New RAII class for allocating bb flags.
4504 * cfgloop.c (verify_loop_structure): Allocate temporary edge
4506 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
4507 in favor of temporarily allocated BB flag.
4508 * hsa-brig.c: Re-order includes.
4509 * hsa-dump.c: Likewise.
4510 * hsa-regalloc.c: Likewise.
4511 * print-rtl.c: Likewise.
4512 * profile-count.c: Likewise.
4514 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
4517 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
4520 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
4523 * wide-int-range.cc (wide_int_range_div): Do not ignore result
4524 from wide_int_range_multiplicative_op.
4526 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4528 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
4531 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
4533 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
4536 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
4538 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
4539 full GENERIC expression used for replacement.
4541 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
4543 * tree-vrp.c (abs_extent_range): Remove.
4544 (extract_range_into_wide_ints): Pass wide ints by reference.
4545 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
4546 Pass wide ints by reference in all calls to
4547 extract_range_into_wide_ints.
4548 * wide-int-range.cc (wide_int_range_div): New.
4549 * wide-int-range.h (wide_int_range_div): New.
4550 (wide_int_range_includes_zero_p): New.
4551 (wide_int_range_zero_p): New.
4553 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
4555 * config/aarch64/aarch64.md (arches): New enum.
4556 (arch): New enum attr.
4557 (arch_enabled): New attr.
4558 (enabled): Now uses arch_enabled only.
4559 (simd, sve, fp16): Removed attribute.
4560 (fp): Attr now defined in terms of 'arch'.
4561 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
4562 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
4563 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
4564 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
4565 attributes into 'arch'.
4566 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
4567 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
4568 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
4569 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
4570 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
4571 'simd' attribute into 'arch'.
4572 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
4573 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
4574 Convert use of 'fp' attribute to 'arch'.
4575 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
4576 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
4578 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
4579 (different modes) Merge 'fp' and 'simd' into 'arch'.
4580 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
4583 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
4585 PR rtl-optimization/87026
4586 * expmed.c (canonicalize_comparison): If we can no longer create
4587 pseudoregisters, don't.
4589 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
4592 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
4594 * config/arm/arm.c (speculation_barrier_libfunc): New static
4596 (arm_init_libfuncs): Initialize it.
4597 (arm_emit_speculation_barrier): New function.
4598 * config/arm/arm.md (speculation_barrier): Call
4599 arm_emit_speculation_barrier for architectures that do not have
4601 (speculation_barrier_insn): Only match on Armv7 or later.
4603 2018-08-23 Richard Biener <rguenther@suse.de>
4606 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
4609 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
4611 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
4612 of single-vector TBLs.
4613 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
4616 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
4619 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
4620 aarch64_evpc_tbl guard.
4622 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
4624 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
4627 2018-08-22 Martin Sebor <msebor@redhat.com>
4630 * tree-pretty-print.c (pretty_print_string): Add argument.
4631 (dump_generic_node): Call to pretty_print_string with string size.
4633 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
4635 PR rtl-optimization/86771
4636 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
4637 of two SETs into those two SETs, one to be placed at i2, if that SETs
4638 destination is modified between i2 and i3.
4640 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
4642 PR tree-optimization/86725
4643 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
4645 (vect_analyze_scalar_cycles_1): Check it.
4647 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
4649 PR tree-optimization/86725
4650 * tree-vect-loop.c (vect_is_simple_reduction): When treating
4651 an outer loop phi as a double reduction, make sure that the
4652 single user of the phi result is an inner loop phi.
4654 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
4656 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
4657 grouped stores with gaps to a strided group.
4659 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
4661 * tree-vect-stmts.c (get_group_load_store_type)
4662 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
4663 first statement in a group.
4665 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
4667 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
4668 the sequence used in gcc/gcc.c.
4670 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
4673 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
4676 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
4678 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
4679 Darwin10-specific unwinder-shim.
4680 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
4681 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
4682 New to cater for Darwin10 Rosetta.
4684 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
4686 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
4687 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
4689 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
4694 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
4695 (dwarf2out_switch_text_section): Generate a local label for the
4696 second function sub-section and apply it as the second FDE start
4698 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
4699 second sub-section start.
4701 2018-08-22 Richard Biener <rguenther@suse.de>
4703 PR tree-optimization/86988
4704 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
4706 2018-08-22 Richard Biener <rguenther@suse.de>
4708 PR tree-optimization/86945
4709 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
4711 2018-08-22 Alexandre Oliva <oliva@adacore.com>
4713 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
4714 a comment about how uses of r2 for .sdata2 come about.
4716 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
4718 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
4720 2018-08-21 Marek Polacek <polacek@redhat.com>
4722 PR c++/86981, Implement -Wpessimizing-move.
4723 * doc/invoke.texi: Document -Wpessimizing-move.
4725 2018-08-21 Jan Hubicka <jh@suse.cz>
4727 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
4728 * tree.h (is_redundant_typedef): Remove.
4729 * dwarf2out.c (is_redundant_typedef): Turn into static function.
4731 2018-08-21 Jan Hubicka <jh@suse.cz>
4733 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
4736 2018-08-21 Tamar Christina <tamar.christina@arm.com>
4738 * expmed.c (extract_low_bits): Reject invalid subregs early.
4740 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
4743 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
4746 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
4748 * config/vxworks.h: Guard vxworks_asm_out_constructor and
4749 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
4750 * config/vxworks.c: Likewise.
4752 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
4754 * config/vxworks.c: Set targetm.have_ctors_dtors
4755 if HAVE_INITFINI_ARRAY_SUPPORT.
4756 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
4757 if HAVE_INITFINI_ARRAY_SUPPORT.
4759 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
4761 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
4762 default search path for VxWorks < 7.
4764 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
4766 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
4767 (get_format_string): Refer to c_getstr.
4769 2018-08-21 Tom de Vries <tdevries@suse.de>
4771 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
4772 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
4773 (debuginfo_early_stop): Declare.
4774 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
4775 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
4776 (debuginfo_early_stop): New function.
4777 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
4778 and debuginfo_early_stop.
4779 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
4780 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
4781 (general_init): Call debuginfo_early_init.
4782 (finalize): Call debuginfo_fini.
4783 (do_compile): Call debuginfo_init.
4784 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
4786 (@item -fdump-debug, @item -fdump-earlydebug): Add.
4788 2018-08-21 Tom de Vries <tdevries@suse.de>
4790 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
4791 flag_dump_noaddr and flag_dump_unnumbered.
4793 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
4795 * wide-int-range.cc (wide_int_range_abs): New.
4796 (wide_int_range_order_set): Rename from wide_int_range_min_max.
4797 * wide-int-range.h (wide_int_range_abs): New.
4798 (wide_int_range_min_max): New.
4799 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
4800 case to call wide_int_range_abs.
4801 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
4802 (extract_range_from_abs_expr): Delete.
4804 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
4807 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
4808 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
4811 2018-08-20 Nathan Sidwell <nathan@acm.org>
4812 Jeff Law <law@redhat.com>
4814 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
4815 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
4817 2018-08-20 David Malcolm <dmalcolm@redhat.com>
4820 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
4821 (decl_attributes): Likewise.
4822 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
4824 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
4825 * diagnostic-core.h (class auto_diagnostic_group): New class.
4826 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
4827 (diagnostic_report_diagnostic): Handle the first diagnostics within
4829 (emit_diagnostic): Add auto_diagnostic_group instance.
4831 (inform_n): Likewise.
4832 (warning): Likewise.
4833 (warning_at): Likewise.
4834 (warning_n): Likewise.
4835 (pedwarn): Likewise.
4836 (permerror): Likewise.
4838 (error_n): Likewise.
4839 (error_at): Likewise.
4841 (fatal_error): Likewise.
4842 (internal_error): Likewise.
4843 (internal_error_no_backtrace): Likewise.
4844 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
4845 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
4846 * diagnostic.h (struct diagnostic_context): Add fields
4847 "diagnostic_group_nesting_depth",
4848 "diagnostic_group_emission_count", "begin_group_cb",
4850 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
4851 Add auto_diagnostic_group instance(s).
4852 (find_explicit_erroneous_behavior): Likewise.
4853 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
4854 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
4855 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
4856 (gimplify_va_arg_expr): Likewise.
4857 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
4858 (HSA_SORRY_AT): Likewise.
4859 * ipa-devirt.c (compare_virtual_tables): Likewise.
4860 (warn_odr): Likewise.
4861 * multiple_target.c (expand_target_clones): Likewise.
4862 * opts-common.c (cmdline_handle_error): Likewise.
4863 * reginfo.c (globalize_reg): Likewise.
4864 * substring-locations.c (format_warning_n_va): Likewise.
4865 * tree-inline.c (expand_call_inline): Likewise.
4866 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
4867 * tree-ssa-loop-niter.c
4868 (do_warn_aggressive_loop_optimizations): Likewise.
4869 * tree-ssa-uninit.c (warn_uninit): Likewise.
4870 * tree.c (warn_deprecated_use): Likewise.
4872 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
4875 * config/i386/i386.md (eh_return): Always update EH return
4876 address in word_mode.
4878 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
4880 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
4881 TARGET_SPLIT_COMPLEX_ARG is defined.
4883 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
4885 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
4887 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
4890 * expr.c (expand_assignment): Assert that bitpos is positive.
4891 (store_field): Likewise
4892 (expand_expr_real_1): Make sure that bitpos is positive.
4893 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
4896 2018-08-20 Nathan Sidwell <nathan@acm.org>
4898 * Makefile.in (CPP_ID_DATA_H): Delete.
4899 (CPP_INTERNAL_H): Don't add it.
4900 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
4901 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
4903 2018-08-20 Richard Biener <rguenther@suse.de>
4905 PR tree-optimization/78655
4906 * tree-vrp.c (extract_range_from_binary_expr_1): Make
4907 pointer + offset nonnull if either operand is nonnull work.
4909 2018-08-20 Tom de Vries <tdevries@suse.de>
4911 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
4912 unless the referenced die describes the added property using
4913 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
4914 Otherwise, add a DW_AT_location to the referenced die.
4916 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
4919 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
4920 register_operand when calling ix86_set_reg_reg_cost.
4921 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
4922 Set *total to 0 for operands that satisfy x86_64_immediate_operand
4923 predicate and to 1 otherwise.
4925 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
4927 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
4928 emit a diagnostic that it is not supported and reset the option.
4929 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
4930 supported and consume the option. (ASM_FINAL_SPEC): New.
4932 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
4934 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
4937 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
4939 C-SKY port: Documentation
4941 * doc/extend.texi (C-SKY Function Attributes): New section.
4942 * doc/invoke.texi (Option Summary): Add C-SKY options.
4943 (C-SKY Options): New section.
4944 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
4946 2018-08-17 Jojo <jijie_rong@c-sky.com>
4947 Huibin Wang <huibin_wang@c-sky.com>
4948 Sandra Loosemore <sandra@codesourcery.com>
4949 Chung-Lin Tang <cltang@codesourcery.com>
4951 C-SKY port: Backend implementation
4953 * config/csky/*: New.
4954 * common/config/csky/*: New.
4956 2018-08-17 Jojo <jijie_rong@c-sky.com>
4957 Huibin Wang <huibin_wang@c-sky.com>
4958 Sandra Loosemore <sandra@codesourcery.com>
4959 Chung-Lin Tang <cltang@codesourcery.com>
4960 Andrew Jenner <andrew@codesourcery.com>
4962 C-SKY port: Configury
4964 * config.gcc (csky-*-*): New.
4965 * configure.ac: Add csky to targets for dwarf2 debug_line support.
4966 * configure: Regenerated.
4968 2018-08-17 David Malcolm <dmalcolm@redhat.com>
4970 * dump-context.h: Include "dumpfile.h".
4971 (dump_context::dump_printf_va): Convert final param from va_list
4972 to va_list *. Convert from ATTRIBUTE_PRINTF to
4973 ATTRIBUTE_GCC_DUMP_PRINTF.
4974 (dump_context::dump_printf_loc_va): Likewise.
4975 * dumpfile.c: Include "stringpool.h".
4976 (make_item_for_dump_printf_va): Delete.
4977 (make_item_for_dump_printf): Delete.
4978 (class dump_pretty_printer): New class.
4979 (dump_pretty_printer::dump_pretty_printer): New ctor.
4980 (dump_pretty_printer::emit_items): New member function.
4981 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
4983 (dump_pretty_printer::emit_item): New member function.
4984 (dump_pretty_printer::stash_item): New member function.
4985 (dump_pretty_printer::format_decoder_cb): New member function.
4986 (dump_pretty_printer::decode_format): New member function.
4987 (dump_context::dump_printf_va): Reimplement in terms of
4988 dump_pretty_printer.
4989 (dump_context::dump_printf_loc_va): Convert final param from va_list
4991 (dump_context::begin_scope): Reimplement call to
4992 make_item_for_dump_printf.
4993 (dump_printf): Update for change to dump_printf_va.
4994 (dump_printf_loc): Likewise.
4995 (selftest::test_capture_of_dump_calls): Convert "stmt" from
4996 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
4997 with %T, %E, and %G.
4998 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
4999 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
5000 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
5001 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
5002 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
5003 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
5004 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
5005 within a dump_printf_loc call to "%wu".
5006 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
5007 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
5008 missing space after "=".
5009 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
5010 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
5011 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
5012 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
5013 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
5014 duplicate "vectorized" from message.
5016 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
5018 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
5019 polyNxK_t element's TYPE_STRING_FLAG.
5021 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
5023 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
5024 (they were unnamed before). Fix comments.
5026 2018-08-17 Nathan Sidwell <nathan@acm.org>
5028 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
5030 2018-08-17 Richard Biener <rguenther@suse.de>
5032 PR tree-optimization/86841
5033 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
5035 2018-08-17 Martin Liska <mliska@suse.cz>
5037 * common.opt: Remove Warn, Init and Report for options with
5038 Ignore/Deprecated flag. Warning is done automatically for
5040 * config/i386/i386.opt: Likewise.
5041 * config/ia64/ia64.opt: Likewise.
5042 * config/rs6000/rs6000.opt: Likewise.
5043 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
5044 Remove usage of flag_check_pointer_bounds.
5045 * lto-wrapper.c (merge_and_complain): Do not handle
5046 OPT_fcheck_pointer_bounds.
5047 (append_compiler_options): Likewise.
5048 * opt-functions.awk: Do not handle Deprecated.
5049 * optc-gen.awk: Check that Var, Report and Init are not
5050 used for an option with Ignore/Deprecated flag.
5051 * opts-common.c (decode_cmdline_option): Do not report
5053 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
5055 * opts.h (struct cl_option): Remove cl_deprecated flag.
5056 (CL_ERR_DEPRECATED): Remove error enum value.
5058 2018-08-17 Richard Biener <rguenther@suse.de>
5061 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
5062 across a va-arg-pack using call adjust its return value accordingly.
5064 2018-08-16 Martin Sebor <msebor@redhat.com>
5066 PR tree-optimization/86853
5067 * gimple-ssa-sprintf.c (struct format_result): Rename member.
5068 (struct fmtresult): Add member and initialize it in ctors.
5069 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
5070 (format_string): Handle %S the same as %ls. Set MAYFAIL.
5071 (format_directive): Set POSUNDER4K when MAYFAIL is set.
5072 (parse_directive): Handle %C same as %c.
5073 (sprintf_dom_walker::compute_format_length): Adjust.
5074 (is_call_safe): Adjust.
5076 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
5078 * builtins.c (c_strlen): Add new parameter eltsize. Use it
5079 for determining how to count the elements.
5080 * builtins.h (c_strlen): Adjust prototype.
5081 * expr.c (string_constant): Add new parameter mem_size.
5082 Set *mem_size appropriately.
5083 * expr.h (string_constant): Adjust protoype.
5084 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
5085 * gimple-fold.h (get_range_strlen): Adjust prototype.
5086 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
5087 (format_string): Call get_string_length with eltsize.
5089 2018-08-16 David Malcolm <dmalcolm@redhat.com>
5091 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
5092 to emit the span, rather than setting it as the prefix.
5094 2018-08-16 David Malcolm <dmalcolm@redhat.com>
5096 * diagnostic-show-locus.c (layout::start_annotation_line): Add
5097 "margin_char" parameter, defaulting to space. Use it in place
5098 of pp_space for the initial part of the margin.
5099 (layout::print_leading_fixits): Use '+' when filling the margin
5100 of line-insertion fix-it hints.
5102 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
5104 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
5107 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
5109 * config/rs6000/altivec.md: Don't set length attribute to the default
5111 * config/rs6000/darwin.md: Ditto.
5112 * config/rs6000/dfp.md: Ditto.
5113 * config/rs6000/htm.md: Ditto.
5114 * config/rs6000/rs6000.md: Ditto.
5115 * config/rs6000/sync.md: Ditto.
5116 * config/rs6000/vsx.md: Ditto.
5118 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
5120 * config/rs6000/altivec.md: Don't set length attribute to the default
5121 value, for branch instructions.
5122 * config/rs6000/darwin.md: Ditto.
5123 * config/rs6000/rs6000.md: Ditto.
5125 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
5127 * config/rs6000/rs6000.md (length): Always define as const_int 4.
5128 (unnamed conditional branch define_insn): Set length to 4 or 8
5129 depending on offset.
5130 (<bd>_<mode>): Similar, for alternative 0.
5131 (<bd>tf_<mode>): Ditto.
5133 2018-08-16 Tamar Christina <tamar.christina@arm.com>
5135 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
5137 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
5139 * doc/rtl.texi: Replace old RTX class names with new names.
5142 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
5144 * expmed.h (canonicalize_comparison): New declaration.
5145 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
5146 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
5147 * optabs.c (prepare_cmp_insn): Likewise.
5148 * rtl.h (unsigned_condition_p): New function which checks if a
5149 comparison operator is unsigned.
5151 2018-08-16 Nathan Sidwell <nathan@acm.org>
5153 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
5154 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
5156 2018-08-16 Tamar Christina <tamar.christina@arm.com>
5159 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
5160 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
5161 (mov<mov>): ..this and enable unconditionally.
5163 2018-08-16 Tamar Christina <tamar.christina@arm.com>
5165 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
5167 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
5169 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
5170 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
5171 "Common" with "Target".
5173 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
5175 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
5176 * doc/invoke.texi (mmitigate-rop): Remove.
5177 * config/i386/i386.c: Do not include "regrename.h".
5178 (ix86_rop_should_change_byte_p, reg_encoded_number)
5179 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
5181 (ix86_reorg): Remove call to ix86_mitigate_rop.
5182 * config/i386/i386.md (attr "modrm_class"): Remove.
5183 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
5184 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
5185 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
5187 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
5189 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
5190 allow folding of mergeh() and mergel() for the float and double types.
5191 (fold_mergehl_helper): Rework to handle building a permute tree
5194 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
5196 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
5199 2018-08-15 David Malcolm <dmalcolm@redhat.com>
5201 * common.opt (fdiagnostics-show-labels): New option.
5202 * diagnostic-show-locus.c (class layout_range): Add field
5204 (class layout): Add field "m_show_labels_p".
5205 (layout_range::layout_range): Add param "label" and use it to
5207 (make_range): Pass in NULL for new "label" param of layout_range's
5209 (layout::layout): Initialize m_show_labels_p.
5210 (layout::maybe_add_location_range): Pass in loc_range->m_label
5211 when constructing layout_range instances.
5212 (struct line_label): New struct.
5213 (layout::print_any_labels): New member function.
5214 (layout::print_line): Call it if label-printing is enabled.
5215 (selftest::test_one_liner_labels): New test.
5216 (selftest::test_diagnostic_show_locus_one_liner): Call it.
5217 * diagnostic.c (diagnostic_initialize): Initialize
5218 context->show_labels_p.
5219 * diagnostic.h (struct diagnostic_context): Add field
5221 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
5222 -fno-diagnostics-show-labels.
5223 * dwarf2out.c (gen_producer_string): Add
5224 OPT_fdiagnostics_show_labels to the ignored options.
5225 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
5227 (gcc_rich_location::maybe_add_expr): Likewise.
5228 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
5229 label" param, defaulting to NULL.
5230 (gcc_rich_location::add_expr): Add "label" param.
5231 (gcc_rich_location::maybe_add_expr): Likewise.
5232 (class text_range_label): New class.
5233 (class range_label_for_type_mismatch): New class.
5234 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
5235 of format_warning_va.
5236 (fmtwarn_n): Likewise for new params of format_warning_n_va.
5237 * lto-wrapper.c (merge_and_complain): Add
5238 OPT_fdiagnostics_show_labels to the "pick one setting" options.
5239 (append_compiler_options): Likewise to the dropped options.
5240 (append_diag_options): Likewise to the passed-on options.
5241 * opts.c (common_handle_option): Handle the new option.
5242 * selftest-diagnostic.c
5243 (test_diagnostic_context::test_diagnostic_context): Enable
5245 * substring-locations.c: Include "gcc-rich-location.h".
5246 (format_warning_n_va): Add "fmt_label" and "param_label" params
5247 and use them as appropriate.
5248 (format_warning_va): Add "fmt_label" and "param_label" params,
5249 passing them on to format_warning_n_va.
5250 (format_warning_at_substring): Likewise.
5251 (format_warning_at_substring_n): Likewise.
5252 * substring-locations.h (format_warning_va): Add "fmt_label" and
5253 "param_label" params.
5254 (format_warning_n_va): Likewise.
5255 (format_warning_at_substring): Likewise.
5256 (format_warning_at_substring_n): Likewise.
5257 * toplev.c (general_init): Initialize global_dc->show_labels_p.
5259 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
5262 * builtins.c (expand_builtin_memcmp): Do not expand the call
5263 when overflow is detected.
5265 2018-08-15 Martin Sebor <msebor@redhat.com>
5267 PR tree-optimization/71625
5268 * config/aarch64/aarch64-builtins.c
5269 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
5271 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
5273 * config/s390/s390.c (s390_reorg): Remove loop.
5275 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
5278 (darwin_function_switched_text_sections): Delete.
5279 * gcc/config/darwin.h
5280 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
5282 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
5285 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
5286 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
5287 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
5289 2018-08-15 Martin Liska <mliska@suse.cz>
5291 PR tree-optimization/86925
5292 * predict.c (expr_expected_value_1): When taking
5293 later predictor, assign also probability.
5294 Use fold_build2_initializer_loc in order to fold
5295 the expression in -frounding-math.
5297 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
5299 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
5301 (expand_vec_perm_1): Try the new method.
5303 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
5306 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
5307 Check whether lra_live_max_point is 0 before dividing.
5309 2018-08-14 Martin Sebor <msebor@redhat.com>
5311 PR tree-optimization/86650
5312 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
5313 (vrp_prop::check_mem_ref): Same.
5315 2018-08-13 Liu Hao <lh_mouse@126.com>
5317 * pretty-print.c (eat_esc_sequence): Swap the foreground and
5318 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
5319 and clear it thereafter, as it only works for DBCS.
5321 2018-08-13 Liu Hao <lh_mouse@126.com>
5323 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
5324 handle returned by _get_osf_handle().
5326 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
5328 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
5329 for folding vec_perm.
5331 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
5333 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
5334 Add support for gimple-folding of vec_pack() and vec_unpack()
5337 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
5339 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
5340 vec_xst variants to the list.
5341 (rs6000_gimple_fold_builtin): Add support for folding unaligned
5342 vector loads and stores.
5344 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
5346 * config.gcc (rs6000-ibm-aix4.x): Delete.
5347 (rs6000-ibm-aix5.1): Delete.
5348 (rs6000-ibm-aix5.2): Delete.
5349 (rs6000-ibm-aix5.3): Delete.
5350 * config/rs6000/aix43.h: Delete.
5351 * config/rs6000/aix51.h: Delete.
5352 * config/rs6000/aix52.h: Delete.
5353 * config/rs6000/t-aix43: Delete.
5355 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
5357 * config/s390/s390.c (s390_decompose_constant_pool_ref):
5359 (s390_decompose_address): Factor out constant pool ref
5362 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
5364 * config/nds32/nds32-predicates.c
5365 (nds32_can_use_bclr_p): Change return type as bool.
5366 (nds32_can_use_bset_p): Ditto.
5367 (nds32_can_use_btgl_p): Ditto.
5368 (nds32_can_use_bitci_p): Ditto.
5369 * config/nds32/nds32-protos.h
5370 (nds32_can_use_bclr_p): Change declaration.
5371 (nds32_can_use_bset_p): Ditto.
5372 (nds32_can_use_btgl_p): Ditto.
5373 (nds32_can_use_bitci_p): Ditto.
5375 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
5377 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
5378 Support -msched-prolog-epilog option.
5379 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
5381 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
5383 * common/config/nds32/nds32-common.c
5384 (nds32_option_optimization_table): Enalbe -malways-align.
5386 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
5388 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
5390 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
5391 OPT_misr_secure_ case.
5392 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
5393 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
5394 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
5396 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
5397 (nds32_isr_info): New field security_level.
5398 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
5399 * config/nds32/nds32.md (return_internal): Consider critical attribute.
5400 * config/nds32/nds32.opt (misr-secure): New option.
5401 * config/nds32/nds32_init.inc: New file.
5402 * config/nds32/nds32_isr.h: New file.
5404 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
5406 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
5407 Update comment for atomic instructions.
5408 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
5409 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
5411 (atomic_loaddi): Revise fence expansion to only emit fence prior to
5412 load for __ATOMIC_SEQ_CST model.
5413 (atomic_loaddi_1): Remove float register target.
5414 (atomic_storedi): Handle CONST_INT values.
5415 (atomic_storedi_1): Remove float register source. Add special case
5417 (memory_barrier): New expander and insn.
5419 2018-08-11 Jakub Jelinek <jakub@redhat.com>
5421 PR tree-optimization/86835
5422 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
5423 new_stmt after def_gsi, make sure to insert new_square_stmt after
5424 that stmt, not 2 stmts before it.
5426 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
5429 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
5432 2018-08-10 Martin Liska <mliska@suse.cz>
5435 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
5437 * builtins.c (expand_builtin_expect_with_probability):
5439 (expand_builtin_expect_with_probability): New function.
5440 (build_builtin_expect_predicate): Add new argumnet probability
5441 for BUILT_IN_EXPECT_WITH_PROBABILITY.
5442 (fold_builtin_expect):
5445 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
5446 * builtins.h (fold_builtin_expect): Set new argument.
5447 * doc/extend.texi: Document __builtin_expect_with_probability.
5448 * doc/invoke.texi: Likewise.
5449 * gimple-fold.c (gimple_fold_call): Pass new argument.
5450 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
5451 also BUILT_IN_EXPECT_WITH_PROBABILITY.
5452 * predict.c (get_predictor_value): New function.
5453 (expr_expected_value): Add new argument probability. Assume
5454 that predictor and probability are always non-null.
5455 (expr_expected_value_1): Likewise. For __builtin_expect and
5456 __builtin_expect_with_probability set probability. Handle
5457 combination in binary expressions.
5458 (tree_predict_by_opcode): Simplify code by simply calling
5459 get_predictor_value.
5460 (pass_strip_predict_hints::execute): Add handling of
5461 BUILT_IN_EXPECT_WITH_PROBABILITY.
5462 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
5464 * tree.h (DECL_BUILT_IN_P): New function.
5466 2018-08-10 Martin Liska <mliska@suse.cz>
5468 PR tree-optimization/85799
5469 * passes.def: Add argument for pass_strip_predict_hints.
5470 * predict.c (class pass_strip_predict_hints): Add new argument
5472 (strip_predictor_early): New function.
5473 (pass_strip_predict_hints::execute): Call the function to
5475 (strip_predict_hints): New function.
5476 * predict.def: Fix comment.
5478 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
5480 * Makefile.in: Clarify which tm.texi to copy over to assert the
5481 right to grant a GFDL license for all.
5483 2018-08-09 Jeff Law <law@redhat.com>
5485 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
5488 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
5490 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
5493 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
5495 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
5496 reductions for variable-length vectors.
5498 2018-08-09 David Malcolm <dmalcolm@redhat.com>
5501 * common.opt (fdiagnostics-show-line-numbers): New option.
5502 * diagnostic-show-locus.c (class layout): Add fields
5503 "m_show_line_numbers_p" and "m_linenum_width";
5504 (num_digits): New function.
5505 (test_num_digits): New function.
5506 (layout::layout): Initialize new fields. Update m_x_offset
5507 logic to handle any left margin.
5508 (layout::print_source_line): Print line number when requested.
5509 (layout::start_annotation_line): New member function.
5510 (layout::print_annotation_line): Call it.
5511 (layout::print_leading_fixits): Likewise.
5512 (layout::print_trailing_fixits): Likewise. Update calls to
5513 move_to_column for new parameter.
5514 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
5515 it to potentially call start_annotation_line.
5516 (layout::show_ruler): Call start_annotation_line.
5517 (selftest::test_line_numbers_multiline_range): New selftest.
5518 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
5519 and selftest::test_line_numbers_multiline_range.
5520 * diagnostic.c (diagnostic_initialize): Initialize
5521 show_line_numbers_p.
5522 * diagnostic.h (struct diagnostic_context): Add field
5523 "show_line_numbers_p".
5524 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
5525 -fno-diagnostics-show-line-numbers.
5526 * dwarf2out.c (gen_producer_string): Add
5527 OPT_fdiagnostics_show_line_numbers to the ignored options.
5528 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
5529 one setting" options.
5530 (append_compiler_options): Likewise to the dropped options.
5531 (append_diag_options): Likewise to the passed-on options.
5532 * opts.c (common_handle_option): Handle the new option.
5533 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
5535 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
5537 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
5538 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
5539 third argument of type "const signed char" to descriptions of
5540 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
5541 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
5542 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
5543 __builtin_bcdsub_ov functions.
5545 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
5547 PR tree-optimization/86858
5548 * tree-vect-loop.c (vect_is_simple_reduction): Restore
5549 flow_bb_inside_loop_p calls.
5551 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
5553 PR tree-optimization/86871
5554 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
5555 instead of gimple_assign_lhs.
5557 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
5560 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
5561 register constraint to operand 0.
5562 (add<mode>3_carryinC): Likewise.
5563 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
5565 2018-08-09 Martin Liska <mliska@suse.cz>
5568 * common.opt: Remove extra line.
5570 2018-08-09 Martin Liska <mliska@suse.cz>
5572 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
5573 at the end of a line, make first letter capital and end up
5574 a sentence with a dot.
5575 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
5576 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
5577 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
5578 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
5579 (PARAM_MAX_ISL_OPERATIONS): Likewise.
5580 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
5581 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
5582 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
5583 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
5584 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
5585 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
5586 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
5587 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
5588 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
5589 (PARAM_TREE_REASSOC_WIDTH): Likewise.
5590 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
5591 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
5592 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
5594 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
5597 * config/s390/s390.c (s390_option_override_internal): Reduce the
5598 stack-clash-protection-probe-interval param if it would be too big
5601 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
5604 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
5605 * config/m68k/m68k.c (handle_move_double): Don't call
5606 m68k_final_prescan_insn.
5607 (m68k_adjust_decorated_operand): Renamed from
5608 m68k_final_prescan_insn, remove first and third operand and
5610 (print_operand): Call it.
5611 (print_operand_address): Call it.
5613 2018-08-08 Nathan Sidwell <nathan@acm.org>
5615 * diagnostic.c (diagnostic_report_current_module): Use
5616 linemap_included_from & linemap_included_from_linemap.
5618 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
5620 * config/aarch64/aarch64-cores.def: Add phecda core.
5621 * config/aarch64/aarch64-tune.md: Regenerate.
5622 * doc/invoke.texi: Add phecda core.
5624 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
5627 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
5629 * config/s390/s390.md ("movti"): Add more alternatives for
5630 constant to GPR copies.
5632 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
5634 * config/s390/s390.c: Fix whitespace damage throughout the file.
5635 * config/s390/s390.h: Likewise.
5636 * config/s390/tpf.h: Likewise.
5638 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
5640 * config/s390/s390.c (s390_loadrelative_operand_p):
5641 Remove TARGET_CPU_ZARCH usages.
5642 (s390_rtx_costs): Likewise.
5643 (s390_legitimate_constant_p): Likewise.
5644 (s390_cannot_force_const_mem): Likewise.
5645 (legitimate_reload_constant_p): Likewise.
5646 (s390_preferred_reload_class): Likewise.
5647 (legitimize_pic_address): Likewise.
5648 (legitimize_tls_address): Likewise.
5649 (s390_split_branches): Removed.
5650 (s390_add_execute): Removed.
5651 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
5652 (s390_mainpool_start): Likewise.
5653 (s390_mainpool_finish): Likewise.
5654 (s390_mainpool_cancel): Removed.
5655 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
5656 (s390_chunkify_cancel): Likewise.
5657 (s390_return_addr_rtx): Likewise.
5658 (s390_register_info): Remove split_branches_pending_p uages.
5659 (s390_optimize_register_info): Likewise.
5660 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
5661 split_branches_pending_p usages.
5662 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
5663 (s390_load_got): Likewise.
5664 (s390_expand_split_stack_prologue): Likewise.
5665 (output_asm_nops): Likewise.
5666 (s390_function_profiler): Likewise.
5667 (s390_emit_call): Likewise.
5668 (s390_conditional_register_usage): Likewise.
5669 (s390_optimize_prologue): Likewise.
5670 (s390_reorg): Remove TARGET_CPU_ZARCH and
5671 split_branches_pending_p usages.
5672 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
5674 (s390_output_indirect_thunk_function): Likewise.
5675 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
5676 (TARGET_CPU_ZARCH_P): Removed.
5677 (struct machine_function): Remove split_branches_pending_p.
5678 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
5680 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
5682 * common/config/s390/s390-common.c (processor_flags_table):
5684 * config.gcc: Remove with_arch/with_tune support.
5685 * config/s390/2064.md: Remove cpu attribute comparisons.
5686 * config/s390/driver-native.c (s390_host_detect_local_cpu):
5688 * config/s390/linux.h (ASM_SPEC):
5689 Remove -march support.
5690 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
5691 Use a table to get an arch level.
5692 * config/s390/s390-opts.h (enum processor_type):
5694 * config/s390/s390.c
5695 (processor_table): Remove entries, add arch_level values.
5696 (s390_issue_rate): Remove cases.
5697 (s390_option_override): Adjust
5698 s390_option_override_internal() call.
5699 (s390_option_override_internal): Remove deprecation warning.
5700 (s390_valid_target_attribute_tree): Adjust
5701 s390_option_override_internal() call.
5702 * config/s390/s390.h (struct s390_processor):
5703 Share with s390-c.c, add arch_level field.
5704 * config/s390/s390.md:
5705 Remove occurrences in cpu attribute.
5706 * config/s390/s390.opt: Remove -march/-mtune support.
5707 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
5708 * doc/invoke.texi: Remove deprecation warning.
5710 2018-08-08 Luis Machado <luis.machado@linaro.org>
5712 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
5714 (qdf24xx_tunings): Set vector cost structure to
5715 qdf24xx_vector_cost.
5717 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
5718 <register_sextend>: Set to 3.
5720 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
5723 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
5724 * config/aarch64/aarch64-simd.md
5725 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
5726 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
5727 (aarch64_frecpx<mode>): ...this new pattern.
5728 * config/aarch64/aarch64-simd-builtins.def: Remove comment
5729 about aarch64_frecp<FRECP:frecp_suffix><mode>.
5731 2018-08-07 Martin Liska <mliska@suse.cz>
5734 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
5735 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
5736 * predict.def (PRED_MALLOC_NONNULL): New predictor.
5737 * doc/extend.texi: Document that malloc attribute adds
5740 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
5743 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
5744 Define to speculation_safe_value_not_needed.
5746 2018-08-06 Jeff Law <law@redhat.com>
5748 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
5749 the vr_values instance to cprop_into_stmt.
5750 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
5751 (cprop_operand): Also query EVRP to determine if OP is a constant.
5753 2018-08-06 Nathan Sidwell <nathan@acm.org>
5755 * diagnostic.c (diagnostic_report_current_module): Reroll
5756 included-at loop. Translate text.
5758 2018-08-06 David Malcolm <dmalcolm@redhat.com>
5760 * function-tests.c (selftest::test_expansion_to_rtl): Call
5761 free_after_compilation.
5763 2018-08-06 Alan Hayward <alan.hayward@arm.com>
5765 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
5767 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
5769 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
5770 loops with memory block operations from getting unrolled.
5772 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
5775 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
5776 Define to speculation_safe_value_not_needed.
5778 2018-08-06 Jeff Law <law@redhat.com>
5780 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
5783 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
5786 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
5787 with all enabled __intN types.
5789 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
5791 2018-08-06 Alan Hayward <alan.hayward@arm.com>
5793 * alias.c (record_set): Check for clobber high.
5794 * cfgexpand.c (expand_gimple_stmt): Likewise.
5795 * combine-stack-adj.c (single_set_for_csa): Likewise.
5796 * combine.c (find_single_use_1): Likewise.
5797 (set_nonzero_bits_and_sign_copies): Likewise.
5798 (get_combine_src_dest): Likewise.
5799 (is_parallel_of_n_reg_sets): Likewise.
5800 (try_combine): Likewise.
5801 (record_dead_and_set_regs_1): Likewise.
5802 (reg_dead_at_p_1): Likewise.
5803 (reg_dead_at_p): Likewise.
5804 * dce.c (deletable_insn_p): Likewise.
5805 (mark_nonreg_stores_1): Likewise.
5806 (mark_nonreg_stores_2): Likewise.
5807 * df-scan.c (df_find_hard_reg_defs): Likewise.
5808 (df_uses_record): Likewise.
5809 (df_get_call_refs): Likewise.
5810 * dwarf2out.c (mem_loc_descriptor): Likewise.
5811 * haifa-sched.c (haifa_classify_rtx): Likewise.
5812 * ira-build.c (create_insn_allocnos): Likewise.
5813 * ira-costs.c (scan_one_insn): Likewise.
5814 * ira.c (equiv_init_movable_p): Likewise.
5815 (rtx_moveable_p): Likewise.
5816 (interesting_dest_for_shprep): Likewise.
5817 * jump.c (mark_jump_label_1): Likewise.
5818 * postreload-gcse.c (record_opr_changes): Likewise.
5819 * postreload.c (reload_cse_simplify): Likewise.
5820 (struct reg_use): Add source expr.
5821 (reload_combine): Check for clobber high.
5822 (reload_combine_note_use): Likewise.
5823 (reload_cse_move2add): Likewise.
5824 (move2add_note_store): Likewise.
5825 * print-rtl.c (print_pattern): Likewise.
5826 * recog.c (decode_asm_operands): Likewise.
5827 (store_data_bypass_p): Likewise.
5828 (if_test_bypass_p): Likewise.
5829 * regcprop.c (kill_clobbered_value): Likewise.
5830 (kill_set_value): Likewise.
5831 * reginfo.c (reg_scan_mark_refs): Likewise.
5832 * reload1.c (maybe_fix_stack_asms): Likewise.
5833 (eliminate_regs_1): Likewise.
5834 (elimination_effects): Likewise.
5835 (mark_not_eliminable): Likewise.
5836 (scan_paradoxical_subregs): Likewise.
5837 (forget_old_reloads_1): Likewise.
5838 * reorg.c (find_end_label): Likewise.
5839 (try_merge_delay_insns): Likewise.
5840 (redundant_insn): Likewise.
5841 (own_thread_p): Likewise.
5842 (fill_simple_delay_slots): Likewise.
5843 (fill_slots_from_thread): Likewise.
5844 (dbr_schedule): Likewise.
5845 * resource.c (update_live_status): Likewise.
5846 (mark_referenced_resources): Likewise.
5847 (mark_set_resources): Likewise.
5848 * rtl.c (copy_rtx): Likewise.
5849 * rtlanal.c (reg_referenced_p): Likewise.
5850 (single_set_2): Likewise.
5851 (noop_move_p): Likewise.
5852 (note_stores): Likewise.
5853 * sched-deps.c (sched_analyze_reg): Likewise.
5854 (sched_analyze_insn): Likewise.
5856 2018-08-06 Alan Hayward <alan.hayward@arm.com>
5858 * cse.c (invalidate_reg): New function extracted from...
5859 (invalidate): ...here.
5860 (canonicalize_insn): Check for clobber high.
5861 (invalidate_from_clobbers): invalidate clobber highs.
5862 (invalidate_from_sets_and_clobbers): Likewise.
5863 (count_reg_usage): Check for clobber high.
5864 (insn_live_p): Likewise.
5865 * cselib.c (cselib_expand_value_rtx_1):Likewise.
5866 (cselib_invalidate_regno): Check for clobber in setter.
5867 (cselib_invalidate_rtx): Pass through setter.
5868 (cselib_invalidate_rtx_note_stores):
5869 (cselib_process_insn): Check for clobber high.
5870 * cselib.h (cselib_invalidate_rtx): Add operand.
5872 2018-08-06 Alan Hayward <alan.hayward@arm.com>
5874 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
5875 (mark_not_eliminable): Likewise.
5876 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
5877 * lra-lives.c (process_bb_lives): Check for clobber high.
5878 * lra.c (new_insn_reg): Remember clobber highs.
5879 (collect_non_operand_hard_regs): Check for clobber high.
5880 (lra_set_insn_recog_data): Likewise.
5881 (add_regs_to_insn_regno_info): Likewise.
5882 (lra_update_insn_regno_info): Likewise.
5884 2018-08-06 Alan Hayward <alan.hayward@arm.com>
5886 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
5887 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
5889 2018-08-06 Alan Hayward <alan.hayward@arm.com>
5891 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
5892 (copy_insn_1): Likewise.
5893 (gen_hard_reg_clobber_high): New gen function.
5894 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
5895 * genemit.c (gen_exp): Likewise.
5896 (gen_emit_seq): Pass through info.
5897 (gen_insn): Check for CLOBBER_HIGH.
5898 (gen_expand): Pass through info.
5899 (gen_split): Likewise.
5900 (output_add_clobbers): Likewise.
5901 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
5902 (remove_clobbers): Likewise.
5903 * rtl.h (gen_hard_reg_clobber_high): New declaration.
5905 2018-08-06 Alan Hayward <alan.hayward@arm.com>
5907 * doc/rtl.texi (clobber_high): Add.
5908 (parallel): Add in clobber high
5909 * rtl.c (rtl_check_failed_code3): Add function.
5910 * rtl.def (CLOBBER_HIGH): Add expression.
5911 * rtl.h (RTL_CHECKC3): Add macro.
5912 (rtl_check_failed_code3): Add declaration.
5913 (XC3EXP): Add macro.
5915 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
5918 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
5919 cfun->machine->max_used_stack_alignment if needed.
5921 2018-08-04 Martin Sebor <msebor@redhat.com>
5923 PR tree-optimization/86571
5924 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
5927 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
5930 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
5933 2018-08-03 Jeff Law <law@redhat.com>
5936 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
5937 Define to speculation_safe_value_not_needed.
5939 2018-08-03 David Malcolm <dmalcolm@redhat.com>
5941 * doc/gcov.texi (-x): Remove duplicate "to".
5942 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
5943 (-Wif-not-aligned): Remove duplicate "is".
5944 (-flto): Remove duplicate "the".
5945 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
5946 duplicate "v5.00.b".
5947 (MSP430 Options): Remove duplicate "and" from the description
5948 of "-mgprel-sec=regexp".
5949 (x86 Options): Remove duplicate copies of "vmldLog102" and
5950 vmlsLog104 from description of "-mveclibabi=type".
5952 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
5954 * internal-fn.h (first_commutative_argument): Declare.
5955 * internal-fn.c (first_commutative_argument): New function.
5956 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
5957 restrictions for pattern statements. Use first_commutative_argument
5958 to look for commutative operands in calls to internal functions.
5960 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
5962 * Makefile.in (wide-int-range.o): New.
5963 * tree-vrp.c: Move all the wide_int_* functions to...
5964 * wide-int-range.cc: ...here.
5965 * tree-vrp.h: Move all the wide_int_* prototypes to...
5966 * wide-int-range.h: ...here.
5968 2018-08-03 Tom de Vries <tdevries@suse.de>
5970 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
5972 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
5973 * except.c (output_function_exception_table): Do early exit if
5974 targetm_common.except_unwind_info (&global_options) == UI_NONE.
5976 2018-08-03 Martin Liska <mliska@suse.cz>
5978 * predict.c (dump_prediction): Change to 2 digits
5981 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
5983 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
5985 (falkor_am_1_gtov_gtov): ... a new insn reservation.
5987 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
5989 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
5990 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
5991 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
5992 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
5993 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
5995 2018-08-02 David Malcolm <dmalcolm@redhat.com>
5997 * diagnostic-show-locus.c (diagnostic_show_locus): Use
5998 pp_take_prefix when saving the existing prefix.
5999 * diagnostic.c (diagnostic_append_note): Likewise.
6000 * langhooks.c (lhd_print_error_function): Likewise.
6001 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
6002 param's type. Free the existing prefix.
6003 (pp_take_prefix): New function.
6004 (pretty_printer::pretty_printer): Drop the prefix parameter.
6005 Rename the length parameter to match the comment.
6006 (pretty_printer::~pretty_printer): Free the prefix.
6007 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
6009 (struct pretty_printer): Drop the "const" from "prefix" field's
6010 type and clarify memory management.
6011 (pp_set_prefix): Drop the "const" from the 2nd param.
6012 (pp_take_prefix): New decl.
6014 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
6016 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
6017 for word_mode_ok here instead of passing as argument.
6018 (expand_block_compare): Change select_block_compare_mode() call.
6019 (expand_strncmp_gpr_sequence): New function.
6020 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
6022 2018-08-02 Jeff Law <law@redhat.com>
6025 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
6026 Define to speculation_safe_value_not_needed.
6029 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
6030 Define to speculation_safe_value_not_needed.
6032 2018-08-02 Tom de Vries <tdevries@suse.de>
6035 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
6036 function. Return UI_TARGET unconditionally.
6037 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
6038 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
6040 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
6042 * genemit.c (print_overload_test): Fix typo.
6044 2018-08-02 Richard Biener <rguenther@suse.de>
6046 PR tree-optimization/86816
6047 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
6048 which checks for value availability before querying it.
6049 (gvn_uses_equal): Use it.
6050 (same_succ_hash): Likewise.
6051 (gimple_equal_p): Likewise.
6053 2018-08-02 Nick Clifton <nickc@redhat.com>
6056 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
6057 Define to speculation_safe_value_not_needed.
6060 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
6061 Define to speculation_safe_value_not_needed.
6064 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
6065 Define to speculation_safe_value_not_needed.
6068 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
6069 Define to speculation_safe_value_not_needed.
6072 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
6073 Define to speculation_safe_value_not_needed.
6076 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
6077 Define to speculation_safe_value_not_needed.
6080 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
6081 Define to speculation_safe_value_not_needed.
6084 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
6085 Define to speculation_safe_value_not_needed.
6088 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
6089 speculation_safe_value_not_needed.
6092 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
6093 to speculation_safe_value_not_needed.
6095 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
6097 * doc/md.texi: Expand the documentation of instruction names
6098 to mention port-local uses. Document '@' in pattern names.
6099 * read-md.h (overloaded_instance, overloaded_name): New structs.
6101 (md_reader::handle_overloaded_name): New member function.
6102 (md_reader::get_overloads): Likewise.
6103 (md_reader::m_first_overload): New member variable.
6104 (md_reader::m_next_overload_ptr): Likewise.
6105 (md_reader::m_overloads_htab): Likewise.
6106 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
6107 m_next_overload_ptr and m_overloads_htab.
6108 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
6109 (get_mode_token, get_code_token, get_int_token): New functions.
6110 (map_attr_string): Add an optional argument that passes back
6111 the associated iterator.
6112 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
6113 (md_reader::handle_overloaded_name, add_overload_instance): New
6115 (apply_iterators): Handle '@' names. Report an error if '@'
6116 is used without iterators.
6117 (initialize_iterators): Initialize the new iterator_group fields.
6118 * genopinit.c (handle_overloaded_code_for)
6119 (handle_overloaded_gen): New functions.
6120 (main): Use them to print declarations of maybe_code_for_* and
6121 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
6122 * genemit.c (print_overload_arguments, print_overload_test)
6123 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
6124 (main): Use it to print definitions of maybe_code_for_* and
6125 maybe_gen_* functions.
6126 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
6127 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
6128 instead of explicit mode checks.
6129 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
6130 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
6131 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
6132 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
6133 (aarch64_expand_compare_and_swap): Likewise
6134 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
6135 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
6136 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
6137 (aarch64_constant_pool_reload_icode): Delete.
6138 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
6139 instead of aarch64_constant_pool_reload_icode. Use
6140 code_for_aarch64_reload_mov instead of explicit mode checks.
6141 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
6142 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
6143 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
6144 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
6145 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
6146 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
6147 (aarch64_atomic_load_op_code): Delete.
6148 (aarch64_emit_atomic_load_op): Likewise.
6149 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
6150 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
6151 instead of aarch64_emit_atomic_load_op.
6152 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
6153 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
6154 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
6155 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
6156 character before the pattern name.
6157 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
6158 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
6159 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
6160 (aarch64_frecps<mode>): Likewise.
6161 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
6162 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
6163 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
6164 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
6165 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
6167 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
6169 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
6170 Allow HFmode constants if TARGET_FP_F16INST.
6172 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
6175 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
6176 No longer check last store for clobber of address register.
6178 2018-08-02 Martin Liska <mliska@suse.cz>
6180 PR gcov-profile/86817
6181 * gcov.c (process_all_functions): New function.
6183 (process_file): Move functions processing to
6184 process_all_functions.
6186 2018-08-02 David Malcolm <dmalcolm@redhat.com>
6188 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
6189 "const" to the "gimple *" and "rtx_insn *" parameters.
6190 * dumpfile.h (dump_user_location_t::dump_user_location_t):
6192 (dump_location_t::dump_location_t): Likewise.
6194 2018-08-01 Martin Sebor <msebor@redhat.com>
6196 PR tree-optimization/86650
6197 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
6198 rather than a "gcall *". Directly pass the data of interest
6199 to percent_K_format, rather than building a temporary CALL_EXPR
6201 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
6202 (gimple_fold_builtin_strncat): Adjust.
6203 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
6204 gcall* argument with gimple*.
6205 * gimple-ssa-warn-restrict.c (check_call): Same.
6206 (wrestrict_dom_walker::before_dom_children): Same.
6207 (builtin_access::builtin_access): Same.
6208 (check_bounds_or_overlap): Same
6209 (maybe_diag_overlap): Same.
6210 (maybe_diag_offset_bounds): Same.
6211 * tree-diagnostic.c (default_tree_printer): Move usage of
6212 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
6214 * tree-pretty-print.c (percent_K_format): Add argument.
6215 * tree-pretty-print.h: Add argument.
6216 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
6217 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
6218 (maybe_diag_stxncpy_trunc): Same.
6219 (handle_builtin_stxncpy): Same.
6220 (handle_builtin_strcat): Same.
6222 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
6224 * match.pd: Optimise pointer range checks.
6226 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
6228 PR tree-optimization/86758
6229 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
6230 to remove pattern statements.
6232 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
6234 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
6235 result of dfs_enumerate_from when constructing stmt_vec_infos,
6236 instead of additionally calling get_loop_body.
6238 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
6240 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
6242 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
6243 When creating an iv, assert that the step is not known to be zero.
6244 (vect_setup_realignment): Update call accordingly.
6245 * tree-vect-stmts.c (vectorizable_store): Likewise.
6246 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
6248 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
6250 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
6251 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
6252 (vectorizable_reduction): Likewise.
6253 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
6254 (vect_detect_hybrid_slp_stmts): Likewise.
6255 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
6257 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
6259 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
6260 (wide_int_set_zero_nonzero_bits): ...this.
6261 (zero_nonzero_bits_from_vr): Rename to...
6262 (vrp_set_zero_nonzero_bits): ...this.
6263 (extract_range_from_multiplicative_op_1): Abstract wide int
6265 (wide_int_range_multiplicative_op): ...here.
6266 (extract_range_from_binary_expr_1): Extract wide int binary
6267 operations into their own functions.
6268 (wide_int_range_lshift): New.
6269 (wide_int_range_can_optimize_bit_op): New.
6270 (wide_int_range_shift_undefined_p): New.
6271 (wide_int_range_bit_xor): New.
6272 (wide_int_range_bit_ior): New.
6273 (wide_int_range_bit_and): New.
6274 (wide_int_range_trunc_mod): New.
6275 (extract_range_into_wide_ints): New.
6276 (vrp_shift_undefined_p): New.
6277 (extract_range_from_multiplicative_op): New.
6278 (vrp_can_optimize_bit_op): New.
6279 * tree-vrp.h (value_range::dump): New.
6280 (wide_int_range_multiplicative_op): New.
6281 (wide_int_range_lshift):New.
6282 (wide_int_range_shift_undefined_p): New.
6283 (wide_int_range_bit_xor): New.
6284 (wide_int_range_bit_ior): New.
6285 (wide_int_range_bit_and): New.
6286 (wide_int_range_trunc_mod): New.
6287 (zero_nonzero_bits_from_bounds): Rename to...
6288 (wide_int_set_zero_nonzero_bits): ...this.
6289 (zero_nonzero_bits_from_vr): Rename to...
6290 (vrp_set_zero_nonzero_bits): ...this.
6291 (range_easy_mask_min_max): Rename to...
6292 (wide_int_range_can_optimize_bit_op): this.
6293 * vr-values.c (simplify_bit_ops_using_ranges): Rename
6294 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
6296 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
6298 * tree-vectorizer.h (vect_orig_stmt): New function.
6299 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
6300 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
6301 (vect_create_epilog_for_reduction): Likewise.
6302 (vectorizable_live_operation): Likewise.
6303 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
6304 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
6305 * tree-vect-stmts.c (vectorizable_call): Likewise.
6306 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
6308 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
6310 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
6312 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
6313 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
6314 (vect_transform_loop): Likewise.
6315 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
6317 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
6319 * tree-vectorizer.h (vect_schedule_slp): Return void.
6320 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
6321 (vect_schedule_slp): Likewise.
6323 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
6325 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
6327 (vect_transform_loop): Update calls accordingly. Schedule SLP
6328 instances before the main loop, if any exist.
6330 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
6332 PR tree-optimization/86749
6333 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
6334 If the lhs is used in a COND_EXPR, check that it is being used
6335 as the "then" or "else" value.
6337 2018-08-01 Tom de Vries <tdevries@suse.de>
6340 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
6341 speculation_safe_value_not_needed.
6343 2018-08-01 Richard Biener <rguenther@suse.de>
6345 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
6348 2018-08-01 Martin Liska <mliska@suse.cz>
6350 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
6351 format how successful transformation is dumped.
6352 (gimple_mod_pow2_value_transform): Likewise.
6353 (gimple_mod_subtract_transform): Likewise.
6354 (gimple_stringops_transform): Likewise.
6356 2018-08-01 Martin Liska <mliska@suse.cz>
6359 * value-prof.c (interesting_stringop_to_profile_p):
6360 Simplify the code and add BUILT_IN_MEMMOVE.
6361 (gimple_stringops_transform): Likewise.
6363 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
6365 * config/aarch64/aarch64-simd.md
6366 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
6367 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
6368 use GPI iterator instead of SI mode.
6370 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
6372 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
6373 rs6000_speculation_barrier.
6374 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
6375 new barrier pattern name.
6377 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
6379 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
6380 (speculation_barrier): New insn.
6382 2018-08-01 Richard Biener <rguenther@suse.de>
6385 * graphite.h: Include isl/id.h and isl/space.h to allow build
6388 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
6391 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
6392 mode for COFF targets.
6393 * defaults.h (TARGET_COFF): Define.
6394 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
6395 TARGET_COFF): Define.
6396 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
6397 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
6398 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
6400 2018-07-31 Alexandre Oliva <oliva@adacore.com>
6402 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
6403 * gimple-streamer-out.c (output_bb): Save it.
6404 * lto-streamer-in.c (input_struct_function_base): Restore
6405 instance discriminator if available. Create map on demand.
6406 * lto-streamer-out.c (output_struct_function_base): Save it if
6408 * final.c (decl_to_instance_map): Document LTO strategy.
6410 2018-07-31 Alexandre Oliva <oliva@adacore.com>
6411 Olivier Hainque <hainque@adacore.com>
6413 * debug.h (decl_to_instance_map_t): New type.
6414 (decl_to_instance_map): Declare.
6415 (maybe_create_decl_to_instance_map): New inline function.
6416 * final.c (bb_discriminator, last_bb_discriminator): New statics,
6417 to track basic block discriminators.
6418 (final_start_function_1): Initialize them.
6419 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
6421 (decl_to_instance_map): New variable.
6422 (map_decl_to_instance, maybe_set_discriminator): New functions.
6423 (notice_source_line): Set discriminator.
6425 2018-07-31 Ian Lance Taylor <iant@golang.org>
6427 * targhooks.c (default_have_speculation_safe_value): Add
6430 2018-07-31 David Malcolm <dmalcolm@redhat.com>
6432 * dump-context.h: Include "pretty-print.h".
6433 (dump_context::refresh_dumps_are_enabled): New decl.
6434 (dump_context::emit_item): New decl.
6435 (class dump_context): Add fields "m_test_pp" and
6437 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
6438 (temp_dump_context::get_dumped_text): New decl.
6439 (class temp_dump_context): Add field "m_pp".
6440 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
6441 (dump_context::refresh_dumps_are_enabled): ...and add a test for
6443 (set_dump_file): Update for above change.
6444 (set_alt_dump_file): Likewise.
6445 (dump_loc): New overload, taking a pretty_printer *.
6446 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
6447 to any test pretty-printer.
6448 (make_item_for_dump_gimple_stmt): New function, adapted from
6449 optinfo::add_gimple_stmt.
6450 (dump_context::dump_gimple_stmt): Call it, and use the result,
6451 eliminating the direct usage of dump_file and alt_dump_file in
6452 favor of indirectly using them via emit_item.
6453 (make_item_for_dump_gimple_expr): New function, adapted from
6454 optinfo::add_gimple_expr.
6455 (dump_context::dump_gimple_expr): Call it, and use the result,
6456 eliminating the direct usage of dump_file and alt_dump_file in
6457 favor of indirectly using them via emit_item.
6458 (make_item_for_dump_generic_expr): New function, adapted from
6460 (dump_context::dump_generic_expr): Call it, and use the result,
6461 eliminating the direct usage of dump_file and alt_dump_file in
6462 favor of indirectly using them via emit_item.
6463 (make_item_for_dump_printf_va): New function, adapted from
6464 optinfo::add_printf_va.
6465 (make_item_for_dump_printf): New function.
6466 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
6467 and use the result, eliminating the direct usage of dump_file and
6468 alt_dump_file in favor of indirectly using them via emit_item.
6469 (make_item_for_dump_dec): New function.
6470 (dump_context::dump_dec): Call it, and use the result,
6471 eliminating the direct usage of dump_file and alt_dump_file in
6472 favor of indirectly using them via emit_item.
6473 (make_item_for_dump_symtab_node): New function, adapted from
6474 optinfo::add_symtab_node.
6475 (dump_context::dump_symtab_node): Call it, and use the result,
6476 eliminating the direct usage of dump_file and alt_dump_file in
6477 favor of indirectly using them via emit_item.
6478 (dump_context::begin_scope): Reimplement, avoiding direct usage
6479 of dump_file and alt_dump_file in favor of indirectly using them
6481 (dump_context::emit_item): New member function.
6482 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
6483 Set up test pretty-printer on the underlying context. Call
6484 refresh_dumps_are_enabled.
6485 (temp_dump_context::~temp_dump_context): Call
6486 refresh_dumps_are_enabled.
6487 (temp_dump_context::get_dumped_text): New member function.
6488 (selftest::verify_dumped_text): New function.
6489 (ASSERT_DUMPED_TEXT_EQ): New macro.
6490 (selftest::test_capture_of_dump_calls): Run all tests twice, with
6491 and then without optinfo enabled. Add uses of
6492 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
6493 * dumpfile.h: Update comment for the dump_* API.
6494 * optinfo-emit-json.cc
6495 (selftest::test_building_json_from_dump_calls): Update for new
6496 param for temp_dump_context ctor.
6497 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
6498 and "m_owned" field.
6499 (optinfo_item::~optinfo_item): Likewise.
6500 (optinfo::add_item): New member function.
6501 (optinfo::emit): Update comment.
6502 (optinfo::add_string): Delete.
6503 (optinfo::add_printf): Delete.
6504 (optinfo::add_printf_va): Delete.
6505 (optinfo::add_gimple_stmt): Delete.
6506 (optinfo::add_gimple_expr): Delete.
6507 (optinfo::add_tree): Delete.
6508 (optinfo::add_symtab_node): Delete.
6509 (optinfo::add_dec): Delete.
6510 * optinfo.h (class dump_context): New forward decl.
6511 (optinfo::add_item): New decl.
6512 (optinfo::add_string): Delete.
6513 (optinfo::add_printf): Delete.
6514 (optinfo::add_printf_va): Delete.
6515 (optinfo::add_gimple_stmt): Delete.
6516 (optinfo::add_gimple_expr): Delete.
6517 (optinfo::add_tree): Delete.
6518 (optinfo::add_symtab_node): Delete.
6519 (optinfo::add_dec): Delete.
6520 (optinfo::add_poly_int): Delete.
6521 (optinfo_item::optinfo_item): Remove "owned" param.
6522 (class optinfo_item): Remove field "m_owned".
6524 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
6527 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
6528 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
6529 requested variable alignment.
6530 (expand_one_ssa_partition): Likewise.
6531 (expand_one_var): Likewise.
6533 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
6535 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
6536 to speculation_safe_value_not_needed.
6538 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
6540 * targhooks.h (speculation_safe_value_not_needed): New prototype.
6541 * targhooks.c (speculation_safe_value_not_needed): New function.
6542 * target.def (have_speculation_safe_value): Update documentation.
6543 * doc/tm.texi: Regenerated.
6545 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
6547 * config/aarch64/iterators.md (ALLI_TI): New iterator.
6548 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
6550 (despeculate_copy<ALLI:mode>_insn): New insn.
6551 (despeculate_copyti_insn): New insn.
6552 (despeculate_simple<ALLI:mode>): New insn
6553 (despeculate_simpleti): New insn.
6554 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
6556 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
6557 aarch64_speculation_safe_value.
6558 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
6560 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
6562 * config/aarch64/aarch64-speculation.cc: New file.
6563 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
6564 before pass_reorder_blocks.
6565 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
6567 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
6568 X14 and X15 when tracking speculation.
6569 * config/aarch64/aarch64.md (register name constants): Add
6570 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
6571 (unspec): Add UNSPEC_SPECULATION_TRACKER.
6572 (speculation_barrier): New insn attribute.
6573 (cmp<mode>): Allow SP in comparisons.
6574 (speculation_tracker): New insn.
6575 (speculation_barrier): Add speculation_barrier attribute.
6576 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
6577 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
6578 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
6580 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
6582 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
6583 aarch64_track_speculation is true.
6584 (tb<optab><mode>1): Likewise.
6585 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
6586 generate CB[N]Z when tracking speculation.
6587 (aarch64_split_compare_and_swap): Likewise.
6588 (aarch64_split_atomic_op): Likewise.
6590 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
6592 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
6594 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
6596 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
6597 (speculation_barrier): New insn.
6599 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
6601 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
6602 * config/arm/arm.md (speculation_barrier): New expand.
6603 (speculation_barrier_insn): New pattern.
6605 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
6607 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
6608 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
6609 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
6610 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
6612 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
6613 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
6614 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
6615 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
6616 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
6617 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
6618 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
6619 * builtins.c (expand_speculation_safe_value): New function.
6620 (expand_builtin): Call it.
6621 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
6622 * doc/extend.texi: Document __builtin_speculation_safe_value.
6623 * doc/md.texi: Document "speculation_barrier" pattern.
6624 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
6625 TARGET_HAVE_SPECULATION_SAFE_VALUE.
6626 * doc/tm.texi: Regenerated.
6627 * target.def (have_speculation_safe_value, speculation_safe_value): New
6629 * targhooks.c (default_have_speculation_safe_value): New function.
6630 (default_speculation_safe_value): New function.
6631 * targhooks.h (default_have_speculation_safe_value): Add prototype.
6632 (default_speculation_safe_value): Add prototype.
6634 2018-07-31 David Malcolm <dmalcolm@redhat.com>
6636 * dump-context.h (dump_context::dump_loc): New decl.
6637 * dumpfile.c (dump_context::dump_loc): New member function.
6638 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
6639 and dump_gimple_stmt.
6640 (dump_context::dump_gimple_expr_loc): Likewise, using
6642 (dump_context::dump_generic_expr_loc): Likewise, using
6644 (dump_context::dump_printf_loc_va): Likewise, using
6646 (dump_context::begin_scope): Explicitly using the global function
6647 "dump_loc", rather than the member function.
6649 2018-07-31 Martin Sebor <msebor@redhat.com>
6651 PR tree-optimization/86741
6652 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
6654 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
6656 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
6657 depend on whether prefetch instructions will be emitted or not.
6658 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
6659 will be emitted or not.
6660 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
6661 (TARGET_SETMEM_PFD): New macros.
6663 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6665 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
6666 (NULL_STMT_VEC_INFO): Delete.
6667 (stmt_vec_info::operator*): Likewise.
6668 (stmt_vec_info::operator gimple *): Likewise.
6669 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
6670 of NULL_STMT_VEC_INFO.
6671 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
6672 (vect_reassociating_reduction_p): Likewise.
6673 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
6674 (vectorizable_store): Likewise.
6675 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
6676 (vec_info::free_stmt_vec_infos): Likewise.
6678 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6680 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
6681 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
6683 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6685 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
6686 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
6687 (vec_info::free_stmt_vec_info): New private member functions.
6688 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
6689 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
6690 * tree-parloops.c (gather_scalar_reductions): Remove calls to
6691 set_stmt_vec_info_vec and free_stmt_vec_infos.
6692 * tree-vect-loop.c (_loop_vec_info): Remove call to
6693 set_stmt_vec_info_vec.
6694 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
6695 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
6696 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
6697 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
6698 (vec_info::free_stmt_vec_info): ...these new functions. Remove
6699 assignments in {vec_info::,}new_stmt_vec_info that are redundant
6700 with the clearing in the xcalloc.
6701 (stmt_vec_info_vec): Delete.
6702 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
6703 (vectorize_loops): Likewise.
6704 (vec_info::~vec_info): Remove argument from call to
6705 free_stmt_vec_infos.
6706 (vec_info::add_stmt): Remove vinfo argument from call to
6709 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6711 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
6712 rather than a gimple stmt.
6713 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
6714 information for pattern statements when passed the original
6715 statement; instead wait to be passed the pattern statement itself.
6716 Don't call set_vinfo_for_stmt here.
6717 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
6718 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
6719 stmt_vec_infos here.
6720 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
6721 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
6722 stmt_vec_infos entry.
6724 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6726 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
6727 * tree-vectorizer.c (vec_info::replace_stmt): New function.
6728 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
6729 * tree-vect-stmts.c (vectorizable_call): Likewise.
6730 (vectorizable_simd_clone_call): Likewise.
6732 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6734 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
6735 * tree-vectorizer.c (vec_info::remove_stmt): New function.
6736 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
6737 * tree-vect-loop.c (vect_transform_loop): Likewise.
6738 * tree-vect-slp.c (vect_schedule_slp): Likewise.
6739 * tree-vect-stmts.c (vect_remove_stores): Likewise.
6741 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6743 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
6744 (vect_dr_stmt): Delete.
6745 * tree-vectorizer.c (vec_info::lookup_dr): New function.
6746 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
6748 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
6749 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
6750 (vect_verify_datarefs_alignment, vect_peeling_supportable)
6751 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
6752 (vect_analyze_data_refs): Likewise.
6753 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
6755 (vect_find_same_alignment_drs): Likewise.
6756 (vect_slp_analyze_node_dependences): Update calls accordingly.
6757 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
6758 instead of DR_VECT_AUX.
6759 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
6760 of a vector data references. Use vec_info::lookup_dr instead of
6762 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
6763 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
6764 instead of DR_VECT_AUX.
6766 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6768 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
6770 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
6772 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
6773 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
6774 (vect_gen_prolog_loop_niters): Likewise.
6776 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6778 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
6779 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
6780 (vect_known_alignment_in_bytes, vect_dr_behavior)
6781 (vect_get_scalar_dr_size): Take references as dr_vec_infos
6782 instead of data_references. Update calls to other routines for
6783 which the same change has been made.
6784 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
6785 dr_vec_infos instead of stmt_vec_infos.
6786 (vect_analyze_data_ref_dependence): Update call accordingly.
6787 (vect_slp_analyze_data_ref_dependence)
6788 (vect_record_base_alignments): Use DR_VECT_AUX.
6789 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
6790 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
6791 (vector_alignment_reachable_p, vect_get_data_access_cost)
6792 (vect_peeling_supportable, vect_analyze_group_access_1)
6793 (vect_analyze_group_access, vect_analyze_data_ref_access)
6794 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
6795 (vect_compile_time_alias, vect_small_gap_p)
6796 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
6797 (vect_supportable_dr_alignment): Take references as dr_vec_infos
6798 instead of data_references. Update calls to other routines for
6799 which the same change has been made.
6800 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
6801 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
6802 (vect_slp_analyze_and_verify_node_alignment)
6803 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
6804 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
6805 (vect_setup_realignment): Use dr_vec_infos. Update calls after
6807 (_vect_peel_info::dr): Replace with...
6808 (_vect_peel_info::dr_info): ...this new field.
6809 (vect_peeling_hash_get_most_frequent)
6810 (vect_peeling_hash_choose_best_peeling): Update accordingly.
6811 (vect_peeling_hash_get_lowest_cost):
6812 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
6813 routines for which the same change has been made.
6814 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
6816 * tree-vect-loop-manip.c (get_misalign_in_elems)
6817 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
6819 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
6820 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
6821 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
6822 (get_group_load_store_type, get_negative_load_store_type)
6823 (vect_get_data_ptr_increment, vectorizable_store)
6824 (vectorizable_load): Likewise.
6825 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
6826 Update calls to other routines for which the same change has been made.
6828 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6830 * tree-vectorizer.h (vec_info::move_dr): New member function.
6831 (dataref_aux): Rename to...
6832 (dr_vec_info): ...this and add "dr" and "stmt" fields.
6833 (_stmt_vec_info::dr_aux): Update accordingly.
6834 (_stmt_vec_info::data_ref_info): Delete.
6835 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
6836 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
6837 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
6838 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
6840 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
6841 (STMT_VINFO_DR_INFO): New macro.
6842 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
6843 (set_dr_misalignment): Update after rename of dataref_aux.
6844 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
6845 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
6846 initialization of STMT_VINFO_DATA_REF.
6847 * tree-vectorizer.c (vec_info::move_dr): New function.
6848 * tree-vect-patterns.c (vect_recog_bool_pattern)
6849 (vect_recog_mask_conversion_pattern)
6850 (vect_recog_gather_scatter_pattern): Use it.
6851 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
6852 the "dr" and "stmt" fields of dr_vec_info instead of
6853 STMT_VINFO_DATA_REF.
6855 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6857 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
6858 (is_pattern_stmt_p): Use it.
6859 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
6860 on pattern statements.
6862 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6864 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
6865 original stmt as a stmt_vec_info rather than a gimple stmt.
6866 (vect_pattern_recog_1): Take the statement directly as a
6867 stmt_vec_info, rather than via a gimple_stmt_iterator.
6868 Update call to vect_mark_pattern_stmts.
6869 (vect_pattern_recog): Update calls accordingly.
6871 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6873 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
6874 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
6875 a vect_def_type for the first argument.
6876 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
6877 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
6878 operand if it isn't defined by a vectorized statement.
6879 (vect_build_gather_load_calls): Remove the mask_dt argument and
6880 update calls to vect_get_vec_def_for_stmt_copy.
6881 (vectorizable_bswap): Likewise the dt argument.
6882 (vectorizable_call): Update calls to vectorizable_bswap and
6883 vect_get_vec_def_for_stmt_copy.
6884 (vectorizable_simd_clone_call, vectorizable_assignment)
6885 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
6886 (vectorizable_comparison): Update calls to
6887 vect_get_vec_def_for_stmt_copy.
6888 (vectorizable_store): Likewise. Remove now-unnecessary calls to
6890 (vect_get_loop_based_defs): Remove dt argument and update call
6891 to vect_get_vec_def_for_stmt_copy.
6892 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
6893 and vect_get_vec_def_for_stmt_copy.
6894 (vectorizable_load): Update calls to vect_build_gather_load_calls
6895 and vect_get_vec_def_for_stmt_copy.
6896 * tree-vect-loop.c (vect_create_epilog_for_reduction)
6897 (vectorizable_reduction, vectorizable_live_operation): Update calls
6898 to vect_get_vec_def_for_stmt_copy.
6900 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6902 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
6903 and gimple stmt arguments with a stmt_vec_info.
6904 (vect_record_base_alignments): Update calls accordingly.
6905 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
6906 and gimple stmt arguments with a stmt_vec_info.
6907 (vect_build_slp_tree_1): Remove vinfo argument and update call
6908 to vect_record_max_nunits.
6909 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
6910 and vect_record_max_nunits.
6912 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6914 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
6915 file and take a stmt_vec_info instead of a gimple stmt.
6916 (supportable_widening_operation, vect_finish_replace_stmt)
6917 (vect_finish_stmt_generation, vect_get_store_rhs)
6918 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
6919 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
6920 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
6921 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
6922 (vect_create_data_ref_ptr, bump_vector_ptr)
6923 (vect_permute_store_chain, vect_setup_realignment)
6924 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
6925 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
6926 (vectorizable_reduction, vectorizable_induction)
6927 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
6928 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
6929 than gimple stmts as arguments.
6930 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
6931 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
6932 (can_group_stmts_p, vect_check_gather_scatter)
6933 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
6934 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
6935 (vect_permute_load_chain, vect_shift_permute_load_chain)
6936 (vect_transform_grouped_load)
6937 (vect_record_grouped_load_vectors): Likewise.
6938 * tree-vect-loop.c (vect_fixup_reduc_chain)
6939 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
6940 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
6941 (vectorizable_reduction, vectorizable_induction)
6942 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
6943 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
6944 (vect_get_load_store_mask): Likewise.
6945 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
6946 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
6947 * tree-vect-stmts.c (vect_mark_relevant)
6948 (is_simple_and_all_uses_invariant)
6949 (exist_non_indexing_operands_for_use_p, process_use)
6950 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
6951 (vect_get_vec_def_for_operand, vect_get_vec_defs)
6952 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
6953 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
6954 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
6955 (get_negative_load_store_type, get_load_store_type)
6956 (vect_check_load_store_mask, vect_check_store_rhs)
6957 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
6958 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
6959 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
6960 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
6961 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
6962 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
6963 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
6964 (supportable_widening_operation): Likewise.
6966 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6968 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
6969 a stmt_vec_info instead of a gcall.
6970 (vect_check_gather_scatter): Update call accordingly.
6971 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
6973 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
6974 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
6975 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
6976 instead of a gimple stmt.
6977 (vect_transform_loop): Update calls accordingly.
6978 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
6979 stmt_vec_infos instead of gimple stmts.
6980 (vect_analyze_slp_instance): Update use accordingly.
6981 * tree-vect-stmts.c (read_vector_array, write_vector_array)
6982 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
6983 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
6984 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
6985 (vect_gen_widened_results_half, vect_get_loop_based_defs)
6986 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
6987 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
6988 down to subroutines.
6990 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6992 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
6993 of the worklist from a vector of gimple stmts to a vector of
6995 * tree-vect-stmts.c (vect_mark_relevant, process_use)
6996 (vect_mark_stmts_to_be_vectorized): Likewise
6998 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7000 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
7001 statement before passing it to vect_analyze_stmt.
7002 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
7003 the chain of phi vector definitions. Track the exit phi via its
7005 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
7006 STMT_VINFO_REDUC_DEF.
7007 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
7008 stmt_vec_infos to handle the statement chains.
7009 (vect_get_slp_defs): Record the first statement in the node
7010 using a stmt_vec_info.
7011 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
7012 statements here and pass their stmt_vec_info down to subroutines.
7013 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
7014 down to vect_finish_stmt_generation.
7015 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
7016 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
7017 stmt_vec_infos to subroutines.
7018 (vect_remove_stores): Use stmt_vec_infos to handle the statement
7021 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7023 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
7024 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
7025 (vect_permute_store_chain, vect_setup_realignment)
7026 (vect_permute_load_chain, vect_shift_permute_load_chain)
7027 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
7028 stmts internally, and when passing values to other vectorizer routines.
7029 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
7030 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
7031 (vect_analyze_loop_operations, get_initial_def_for_reduction)
7032 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
7033 (vectorizable_reduction, vectorizable_induction)
7034 (vectorizable_live_operation, vect_transform_loop_stmt)
7035 (vect_transform_loop): Likewise.
7036 * tree-vect-patterns.c (vect_reassociating_reduction_p)
7037 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
7038 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
7039 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
7040 (vect_slp_analyze_node_operations_1): Likewise.
7041 * tree-vect-stmts.c (vect_mark_relevant, process_use)
7042 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
7043 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
7044 (vect_finish_stmt_generation_1, get_group_load_store_type)
7045 (get_load_store_type, vect_build_gather_load_calls)
7046 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
7047 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
7048 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
7049 (vectorizable_store, vectorizable_load, vectorizable_condition)
7050 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
7051 (supportable_widening_operation): Likewise.
7052 (vect_get_vector_types_for_stmt): Likewise.
7053 * tree-vectorizer.h (vect_dr_behavior): Likewise.
7055 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7057 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
7058 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
7059 (vect_permute_store_chain, vect_permute_load_chain)
7060 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
7061 repeated stmt_vec_info lookups.
7062 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
7063 (vect_update_ivs_after_vectorizer): Likewise.
7064 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
7065 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
7066 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
7067 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
7068 (vect_bb_slp_scalar_cost): Likewise.
7069 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
7071 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7073 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
7074 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
7075 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
7076 to get gassigns and gcalls, rather than operating on generc gimple
7078 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
7079 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
7080 (vectorizable_load, vect_analyze_stmt): Likewise.
7081 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
7083 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7085 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
7086 return stmt_vec_infos rather than gimple stmts. Do not accept
7088 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
7090 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
7091 Update use of get_later_stmt.
7092 (vect_get_constant_vectors): Update call accordingly.
7093 (vect_schedule_slp_instance): Likewise
7094 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
7095 (vect_slp_analyze_instance_dependence): Likewise.
7096 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
7098 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7100 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
7101 (stmt_info_for_cost::stmt_info): ...this new field.
7102 (add_stmt_costs): Update accordingly.
7103 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
7104 (vect_get_known_peeling_cost): Likewise.
7105 (vect_estimate_min_profitable_iters): Likewise.
7106 * tree-vect-stmts.c (record_stmt_cost): Likewise.
7108 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7110 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
7111 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
7112 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
7114 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
7116 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7118 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
7119 a gimple stmt to a stmt_vec_info.
7120 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
7122 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7124 * tree-vectorizer.h (vec_info::grouped_stores): Change from
7125 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
7126 (_loop_vec_info::reduction_chains): Likewise.
7127 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
7129 * tree-vect-slp.c (vect_analyze_slp): Likewise.
7131 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7133 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
7134 a gimple stmt to a stmt_vec_info.
7135 (_stmt_vec_info::next_element): Likewise.
7136 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
7137 (vect_slp_analyze_and_verify_node_alignment)
7138 (vect_analyze_group_access_1, vect_analyze_group_access)
7139 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
7140 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
7141 (vect_supportable_dr_alignment): Update accordingly.
7142 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
7143 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
7144 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
7145 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
7146 * tree-vect-slp.c (vect_build_slp_tree_1)
7147 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
7148 (vect_split_slp_store_group, vect_analyze_slp_instance)
7149 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
7150 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
7151 (get_group_load_store_type, get_load_store_type)
7152 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
7153 (vect_transform_stmt, vect_remove_stores): Likewise.
7155 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7157 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
7159 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
7160 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
7161 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
7162 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
7163 (vector_alignment_reachable_p, vect_get_data_access_cost)
7164 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
7165 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
7166 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
7167 (vect_analyze_group_access_1, vect_analyze_group_access)
7168 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
7169 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
7170 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
7171 result of vect_dr_stmt and use the stmt_vec_info instead of
7172 the associated gimple stmt.
7173 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
7174 (vect_gen_prolog_loop_niters): Likewise.
7175 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
7177 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7179 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
7180 to a vec<stmt_vec_info>.
7181 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
7182 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
7184 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
7185 to a vec<stmt_vec_info>.
7186 (bst_traits::value_type, bst_traits::value_type): Likewise.
7187 (bst_traits::hash): Update accordingly.
7188 (vect_get_and_check_slp_defs): Change the stmts parameter from
7189 a vec<gimple *> to a vec<stmt_vec_info>.
7190 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
7191 (vect_build_slp_tree): Likewise.
7192 (vect_build_slp_tree_2): Likewise. Update uses of
7193 SLP_TREE_SCALAR_STMTS.
7194 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
7195 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
7196 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
7197 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
7198 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
7199 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
7200 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
7201 (vect_get_constant_vectors, vect_get_slp_defs)
7202 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
7203 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
7204 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
7205 instead of gimple stmts.
7206 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
7207 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
7208 (vect_slp_analyze_instance_dependence): Update uses of
7209 SLP_TREE_SCALAR_STMTS.
7210 (vect_slp_analyze_and_verify_node_alignment): Likewise.
7211 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
7212 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
7213 (get_initial_defs_for_reduction): Likewise.
7214 (vect_create_epilog_for_reduction): Likewise.
7215 (vectorize_fold_left_reduction): Likewise.
7216 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
7217 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
7218 (can_vectorize_live_stmts): Likewise.
7220 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7222 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
7223 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
7224 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
7226 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
7228 (gather_scalar_reductions): Update after above interface changes.
7229 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
7230 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
7232 (vect_force_simple_reduction): Likewise.
7233 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
7234 LOOP_VINFO_REDUCTIONS.
7235 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
7237 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7239 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
7240 a gimple stmt to a stmt_vec_info.
7241 * tree-vect-loop.c (vect_active_double_reduction_p)
7242 (vect_force_simple_reduction, vectorizable_reduction): Update
7245 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7247 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
7248 vec<gimple *> to a vec<stmt_vec_info>.
7249 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
7250 the reduction_phis argument from a vec<gimple *> to a
7252 (vectorizable_reduction): Likewise the phis local variable that
7253 is passed to vect_create_epilog_for_reduction. Update for new type
7254 of SLP_TREE_VEC_STMTS.
7255 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
7256 (vectorizable_live_operation): Likewise.
7257 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
7258 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
7260 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7262 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
7263 a gimple stmt to a stmt_vec_info.
7264 (vectorizable_condition, vectorizable_live_operation)
7265 (vectorizable_reduction, vectorizable_induction): Pass back the
7266 vectorized statement as a stmt_vec_info.
7267 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
7268 use of STMT_VINFO_VEC_STMT.
7269 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
7270 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
7271 as stmt_vec_infos rather than gimple stmts.
7272 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
7274 (vectorizable_live_operation): Likewise.
7275 (vectorizable_reduction, vectorizable_induction): Likewise,
7276 updating use of STMT_VINFO_VEC_STMT.
7277 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
7278 of STMT_VINFO_VEC_STMT.
7279 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
7280 (vectorizable_simd_clone_call, vectorizable_conversion)
7281 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
7282 (vectorizable_store, vectorizable_load, vectorizable_condition)
7283 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
7284 from a gimple stmt to a stmt_vec_info.
7285 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
7286 pointer to a stmt_vec_info to the vectorizable_* routines.
7288 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7290 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
7291 a gimple stmt to a stmt_vec_info.
7292 (is_pattern_stmt_p): Update accordingly.
7293 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
7294 (vect_record_grouped_load_vectors): Likewise.
7295 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
7296 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
7297 (vect_model_reduction_cost): Likewise.
7298 (vect_create_epilog_for_reduction): Likewise.
7299 (vectorizable_reduction, vectorizable_induction): Likewise.
7300 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
7301 Return the stmt_vec_info for the pattern statement.
7302 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
7303 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
7304 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
7305 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
7306 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
7307 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
7308 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
7309 (free_stmt_vec_info, vect_is_simple_use): Likewise.
7311 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7313 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
7314 (vect_finish_stmt_generation): Likewise.
7315 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
7316 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
7317 (vect_build_gather_load_calls): Use the return value of the above
7318 functions instead of a separate call to vinfo_for_stmt. Use narrow
7319 scopes for the input gimple stmt and wider scopes for the associated
7320 stmt_vec_info. Use vec_info::lookup_def when setting these
7321 stmt_vec_infos from an SSA_NAME definition.
7322 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
7323 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
7324 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
7325 (vectorizable_store, vectorizable_load, vectorizable_condition)
7326 (vectorizable_comparison): Likewise.
7327 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
7328 (vectorizable_reduction): Likewise.
7330 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7332 * tree-vectorizer.h (vect_is_simple_use): Add an optional
7333 stmt_vec_info * parameter before the optional gimple **.
7334 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
7335 (process_use, vect_get_vec_def_for_operand_1): Update callers.
7336 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
7337 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7338 (vectorizable_live_operation): Likewise.
7339 * tree-vect-patterns.c (type_conversion_p): Likewise.
7340 (vect_look_through_possible_promotion): Likewise.
7341 (vect_recog_rotate_pattern): Likewise.
7342 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
7344 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7346 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
7347 a typedef to a wrapper class.
7348 (NULL_STMT_VEC_INFO): New macro.
7349 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
7350 (stmt_vec_info::operator*): New function.
7351 (stmt_vec_info::operator gimple *): Likewise.
7352 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
7353 (add_stmt_costs): Likewise.
7354 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
7355 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
7356 (vect_get_known_peeling_cost): Likewise.
7357 (vect_estimate_min_profitable_iters): Likewise.
7358 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
7359 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
7360 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
7361 (vectorizable_store, free_stmt_vec_infos): Likewise.
7362 (new_stmt_vec_info): Change return type of xcalloc to
7365 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7367 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
7368 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
7369 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
7370 a single_imm_use-based sequence.
7371 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
7373 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7375 * tree-vectorizer.h (vec_info::lookup_def): Declare.
7376 * tree-vectorizer.c (vec_info::lookup_def): New function.
7377 * tree-vect-patterns.c (vect_get_internal_def): Use it.
7378 (vect_widened_op_tree): Likewise.
7379 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
7380 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
7381 (vectorizable_reduction): Likewise.
7382 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
7384 (vect_is_slp_reduction): Update calls accordingly. Use
7385 vec_info::lookup_def.
7386 (vect_is_simple_reduction): Likewise
7387 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
7389 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7391 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
7392 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
7393 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
7395 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
7396 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
7397 (vect_update_vf_for_slp, vect_analyze_loop_operations)
7398 (vect_is_slp_reduction, vectorizable_induction)
7399 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
7400 * tree-vect-patterns.c (vect_init_pattern_stmt):
7401 (vect_determine_min_output_precision_1, vect_determine_precisions)
7402 (vect_pattern_recog): Likewise.
7403 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
7404 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
7405 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
7406 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
7407 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
7408 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
7409 info field from a loop to a loop_vec_info.
7411 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7413 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
7414 (vec_info::add_stmt): Declare.
7415 * tree-vectorizer.c (vec_info::add_stmt): New function.
7416 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
7417 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
7418 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
7419 (vectorizable_induction): Likewise.
7420 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
7421 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
7422 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
7423 (vectorizable_load): Likewise.
7424 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
7425 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
7426 (vect_recog_gather_scatter_pattern): Likewise.
7427 (append_pattern_def_seq): Likewise. Remove a check that is
7428 performed by add_stmt itself.
7430 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7432 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
7433 which make_ssa_name was called with new_stmt before new_stmt
7436 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7438 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
7440 (vect_is_slp_reduction): ...here...
7441 (vect_is_simple_reduction): ...and here. Remove repetition of tests
7442 that are already known to be false.
7444 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7446 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
7447 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
7448 STMT_VINFO_NUM_SLP_USES when it's true.
7449 (vect_free_slp_instance): Add a final_p parameter and pass it to
7451 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
7452 (vect_analyze_slp_instance): Likewise.
7453 (vect_slp_analyze_operations): Likewise.
7454 (vect_slp_analyze_bb_1): Likewise.
7455 * tree-vectorizer.c (vec_info): Likewise.
7456 * tree-vect-loop.c (vect_transform_loop): Likewise.
7458 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7460 * tree-vect-loop.c (vectorizable_reduction): Assert that the
7461 function is not called for second and subsequent members of
7464 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7466 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
7467 cases for nested loops from here to ...
7468 (vect_create_epilog_for_reduction): ...here. Only call
7469 vect_is_simple_use for inner-loop reductions.
7471 2018-07-31 Martin Liska <mliska@suse.cz>
7473 PR gcov-profile/85338
7474 PR gcov-profile/85350
7475 PR gcov-profile/85372
7476 * profile.c (struct location_triplet): New.
7477 (struct location_triplet_hash): Likewise.
7478 (output_location): Do not output a BB that
7479 is already recorded for a line.
7480 (branch_prob): Use streamed_locations.
7482 2018-07-31 Martin Liska <mliska@suse.cz>
7484 PR gcov-profile/85370
7485 * coverage.c (coverage_begin_function): Do not mark target
7486 clones as artificial functions.
7488 2018-07-31 Martin Liska <mliska@suse.cz>
7490 PR gcov-profile/83813
7491 PR gcov-profile/84758
7492 PR gcov-profile/85217
7493 PR gcov-profile/85332
7494 * profile.c (branch_prob): Do not record GOTO expressions
7495 for GIMPLE statements which locations are already streamed.
7497 2018-07-31 Olivier Hainque <hainque@adacore.com>
7499 * gcc.c (handle_spec_function): Accept a soft_matched_part
7500 argument, as do_spec_1. Pass it down to ...
7501 (eval_spec_function): Accept a soft_matched_part argument,
7502 and pass it down to ...
7503 (do_spec_2): Accept a soft_matched_part argument, and pass
7504 it down to do_spec_1.
7505 (do_spec_1): Pass soft_matched_part to handle_spec_function.
7506 (handle_braces): Update call to handle_spec_function.
7507 (driver::set_up_specs): Update calls to do_spec_2.
7508 (compare_debug_dump_opt_spec_function): Likewise.
7509 (compare_debug_self_opt_spec_function): Likewise.
7511 2018-07-31 Olivier Hainque <hainque@adacore.com>
7513 * common.opt (nolibc): New option.
7514 * doc/invoke.texi (Link Options): Document it.
7515 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
7516 * config/alpha/linux.h: Likewise.
7517 * config/arc/elf.h: Likewise.
7518 * config/arm/uclinux-elf.h: Likewise.
7519 * config/arm/unknown-elf.h: Likewise.
7520 * config/avr/avrlibc.h: Likewise.
7521 * config/bfin/bfin.h: Likewise.
7522 * config/bfin/linux.h: Likewise.
7523 * config/bfin/uclinux.h: Likewise.
7524 * config/darwin.h: Likewise.
7525 * config/darwin10.h: Likewise.
7526 * config/darwin12.h: Likewise.
7527 * config/gnu-user.h: Likewise.
7528 * config/lm32/uclinux-elf.h: Likewise.
7529 * config/pa/pa-hpux11.h: Likewise.
7530 * config/pa/pa64-hpux.h: Likewise.
7531 * config/sparc/sparc.h: Likewise.
7533 2018-07-31 Olivier Hainque <hainque@adacore.com>
7535 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
7536 undefined variables.
7538 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
7541 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
7544 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
7546 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
7547 terminated string literal.
7549 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
7551 PR rtl-optimization/85160
7552 * combine.c (is_just_move): New function.
7553 (try_combine): Allow combining two instructions into two if neither of
7554 the original instructions was a move.
7556 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
7559 * doc/extend.texi (Global Register Variables): Discourage use of type
7561 (Local Register Variables): Likewise.
7563 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
7565 PR tree-optimization/86506
7566 * hwint.h (ceil_log2): Resync with hwint.c implementation.
7568 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
7571 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
7572 hard_regno, make sure no insn between `from` and `to` clobbers it.
7574 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
7575 Tom de Vries <tdevries@suse.de>
7577 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
7578 (PTX_DEFAULT_RUNTIME_DIM): ... this.
7579 (nvptx_goacc_validate_dims): Set default worker and gang dims to
7580 PTX_DEFAULT_RUNTIME_DIM.
7581 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
7583 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
7585 * config/pa/pa.c (pa_output_addr_vec): Align address table.
7586 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
7587 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
7589 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
7591 * config/rs6000/constraints.md (wG constraint): Delete, no longer
7593 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
7594 predicate to reflect toc fusion has been deleted.
7595 (toc_fusion_mem_raw): Delete, no longer used.
7596 (toc_fusion_mem_wrapped): Likewise.
7597 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
7599 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
7600 Delete, no longer used.
7601 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
7602 meant to be used for toc fusion.
7603 (rs6000_debug_print_mode): Delete toc fusion debugging.
7604 (rs6000_debug_reg_global): Likewise.
7605 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
7606 fusion and secondary reload support that were never used.
7607 (rs6000_option_override_internal): Delete TOC fusion, that was only
7608 partially defined, and it did not work unless you also used the
7610 (rs6000_legitimate_address_p): Delete TOC fusion support.
7611 (rs6000_opt_masks): Likewise.
7612 (fusion_wrap_memory_address): Delete function, no longer used.
7613 (fusion_split_address); Delete TOC fusion support.
7614 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
7615 longer used with toc fusion being deleted.
7616 (TARGET_TOC_FUSION_FP): Likewise.
7617 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
7619 (toc fusion spliter): Delete TOC fusion support.
7620 (toc_fusionload_<mode>): Likewise.
7621 (toc_fusionload_di): Likewise.
7622 (fusion_gpr_load_<mode>): Delete generator function, this insn no
7623 longer needs to be named. Rename predicate to delete TOC fusion.
7624 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
7625 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
7626 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
7627 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
7628 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
7630 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
7632 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
7633 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
7634 __int128 in built-in function prototypes.
7635 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
7636 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
7638 2018-07-27 Martin Sebor <msebor@redhat.com>
7640 PR tree-optimization/86696
7641 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
7642 types, including enums.
7643 (handle_char_store): Be prepared for the above function to fail.
7645 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
7647 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
7648 when optimization level is lower than 2 or optimize for size.
7650 2018-07-26 Martin Sebor <msebor@redhat.com>
7652 PR tree-optimization/86043
7653 PR tree-optimization/86042
7654 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
7655 (get_string_cst_length): Rename...
7656 (get_min_string_length): ...to this. Add argument.
7657 (handle_char_store): Extend to handle multi-character stores by
7659 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
7660 * tree.h (initializer_zerop): Add argument.
7662 2018-07-26 Jakub Jelinek <jakub@redhat.com>
7665 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
7666 declare target to variables if they have always,{to,from,tofrom} map
7669 2018-07-26 Martin Liska <mliska@suse.cz>
7672 * lto-wrapper.c: Add linker_output as prefix
7673 for ltrans_output_file.
7675 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
7677 PR rtl-optimization/85805
7678 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
7679 value for hard registers if that was written in the same mode.
7681 2018-07-26 Martin Liska <mliska@suse.cz>
7683 PR gcov-profile/86536
7684 * gcov.c (format_gcov): Use printf format %.*f directly
7685 and do not handle special values.
7687 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
7689 * common/config/arc/arc-common.c (arc_option_optimization_table):
7690 Update default optimizations for size.
7692 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
7694 * config/arc/arc.md (movsf_insn): Add short instruction selection.
7695 * config/arc/constraints.md (CfZ): New constraint.
7696 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
7697 (subsf3_fpu): Likewise.
7698 (cmpsf_fpu): Likewise.
7699 (cmpsf_fpu_uneq): Likewise.
7701 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
7703 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
7705 (arc_is_uncached_mem_p): uncached applies to both the variable and
7708 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
7710 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
7713 2018-07-25 David Malcolm <dmalcolm@redhat.com>
7715 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
7716 field "m_scopes" from vec to auto_vec.
7718 2018-07-25 Martin Liska <mliska@suse.cz>
7720 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
7723 2018-07-25 Richard Biener <rguenther@suse.de>
7726 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
7727 special wrt context_die late.
7728 (gen_subprogram_die): Re-use DIEs in local scope.
7730 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
7732 PR tree-optimization/86644
7733 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
7735 2018-07-25 Martin Liska <mliska@suse.cz>
7738 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
7739 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
7741 2018-07-25 Martin Liska <mliska@suse.cz>
7744 * params.def: Explain ASan abbreviation and provide
7745 a documentation link.
7747 2018-07-24 Martin Sebor <msebor@redhat.com>
7749 PR tree-optimization/86622
7750 PR tree-optimization/86532
7751 * builtins.h (string_length): Declare.
7752 * builtins.c (c_strlen): Correct handling of non-constant offsets.
7753 (check_access): Be prepared for non-constant length ranges.
7754 (string_length): Make extern.
7755 * expr.c (string_constant): Only handle the minor non-constant
7756 array index. Use string_constant to compute the length of
7757 a generic string constant.
7759 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
7761 PR tree-optimization/86618
7762 * tree-vect-stmts.c (vectorizable_call): Don't take the address
7763 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
7765 2018-07-24 David Malcolm <dmalcolm@redhat.com>
7767 PR tree-optimization/86636
7768 * json.cc (json::object::set): Fix comment. Add assertions.
7769 (json::array::append): Move here from json.h. Add comment and an
7771 (json::string::string): Likewise.
7772 * json.h (json::array::append): Move to json.cc.
7773 (json::string::string): Likewise.
7774 * optinfo-emit-json.cc
7775 (optrecord_json_writer::impl_location_to_json): Assert that we
7776 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
7777 wrapper around it. Expand the location once, rather than three
7779 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
7780 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
7782 (optrecord_json_writer::optinfo_to_json): Likewise, in four
7783 places. Fix some overlong lines.
7785 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
7787 * config/aarch64/aarch64-simd.md
7788 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
7789 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
7790 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
7791 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
7792 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
7793 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
7794 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
7795 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
7796 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
7798 2018-07-24 Jakub Jelinek <jakub@redhat.com>
7801 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
7802 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
7803 and abs_d == d, do the power of two handling if profitable.
7805 2018-07-24 Richard Biener <rguenther@suse.de>
7807 * match.pd: Add BIT_FIELD_REF canonicalizations.
7809 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
7812 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
7814 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
7816 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
7817 terminated STRING_CST object.
7819 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
7821 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
7823 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
7825 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
7826 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
7827 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
7828 the elements into a register.
7829 (rs6000_split_v4si_init_di_reg): Delete.
7830 (rs6000_split_v4si_init): Delete.
7831 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
7832 (vsx_init_v4si): Rewrite as a define_expand.
7834 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
7836 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
7837 zero_extend argument from memory): New.
7839 2018-07-22 Martin Sebor <msebor@redhat.com>
7842 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
7843 diagnosing calls with unknown arguments unless -Walloca-larger-than
7844 is restricted to less than PTRDIFF_MAX bytes.
7846 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
7848 * doc/gcov.texi (Invoking Gcov): Editorial changes.
7850 2018-07-20 David Malcolm <dmalcolm@redhat.com>
7852 * pretty-print.c (text_info::set_location): Remove redundant
7853 "line_table" parameter from call to rich_location::set_range.
7855 2018-07-20 Martin Sebor <msebor@redhat.com>
7858 * builtins.c (expand_builtin_alloca): Adjust.
7859 * calls.c (alloc_max_size): Simplify.
7860 * cgraphunit.c (cgraph_node::expand): Adjust.
7861 * common.opt (larger_than_size, warn_frame_larger_than): Remove
7863 (frame_larger_than_size): Same.
7864 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
7865 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
7867 * doc/invoke.texi (GCC Command Options): Document option arguments.
7868 Explain byte-size arguments and suffixes.
7869 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
7870 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
7871 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
7872 * doc/options.texi (UInteger): Expand.
7873 (Host_Wide_Int, ByteSize): Document new properties.
7874 * final.c (final_start_function_1): Include sizes in an error message.
7875 * function.c (frame_offset_overflow): Same.
7876 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
7877 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
7878 Diagnose unbounded alloca calls only for limits of less than
7880 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
7881 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
7883 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
7884 only for limits of less than PTRDIFF_MAX.
7885 * langhooks-def.h (lhd_handle_option): Change function argument
7887 * langhooks.c (lhd_handle_option): Same.
7888 * langhooks.h (handle_option): Same.
7889 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
7891 (var_type, var_type_struct): Same.
7892 (var_set): Handle ByteSize flag.
7893 * optc-gen.awk: Add comments to output to ease debugging. Make
7894 use of HOST_WIDE_INT where appropriate.
7895 * opts-gen-save.awk: Use %lx to format unsigned long.
7896 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
7897 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
7898 arguments. Parse bytes-size suffixes.
7899 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
7900 (enum_value_to_arg): Same.
7901 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
7902 (handle_option): Adjust.
7903 (generate_option): Change function argument to HOST_WIDE_INT.
7904 (cmdline_handle_error): Adjust.
7905 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
7906 (set_option): Change function argument to HOST_WIDE_INT.
7907 (option_enabled): Handle cl_host_wide_int.
7908 (get_option_state): Handle CLVC_SIZE.
7909 (control_warning_option): Same.
7910 * opts.c (common_handle_option): Change function argument to
7911 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
7912 OPT_Wvla_larger_than_.
7913 * opts.h (enum cl_var_type): Add an enumerator.
7914 * stor-layout.c (layout_decl): Print a more meaningful warning.
7915 * toplev.c (output_stack_usage): Adjust.
7917 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
7919 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
7920 call to inline_expand_builtin_string_cmp.
7921 (expand_builtin_strcmp): Likewise.
7922 (expand_builtin_strncmp): Likewise.
7923 (inline_string_cmp): Delete the last parameter, change char_type_node
7924 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
7926 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
7927 the inlining expansion on target where the type of the call has same or
7928 narrower precision than unsigned char.
7930 2018-07-20 David Malcolm <dmalcolm@redhat.com>
7932 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
7933 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
7934 * common.opt (fsave-optimization-record): New option.
7935 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
7936 * doc/invoke.texi (-fsave-optimization-record): New option.
7937 * dumpfile.c: Include "optinfo-emit-json.h".
7938 (struct kv_pair): Move to coretypes.h.
7939 (optgroup_options): Make non-static.
7940 (dump_context::end_scope): Call
7941 optimization_records_maybe_pop_dump_scope.
7942 * dumpfile.h (optgroup_options): New decl.
7943 * json.cc: New file.
7945 * optinfo-emit-json.cc: New file.
7946 * optinfo-emit-json.h: New file.
7947 * optinfo.cc: Include "optinfo-emit-json.h".
7948 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
7949 (optinfo_enabled_p): Check optimization_records_enabled_p.
7950 (optinfo_wants_inlining_info_p): Likewise.
7951 * optinfo.h: Update comment.
7952 * profile-count.c (profile_quality_as_string): New function.
7953 * profile-count.h (profile_quality_as_string): New decl.
7954 (profile_count::quality): New accessor.
7955 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
7956 and optinfo_emit_json_cc_tests.
7957 * selftest.h (selftest::json_cc_tests): New decl.
7958 (selftest::optinfo_emit_json_cc_tests): New decl.
7959 * toplev.c: Include "optinfo-emit-json.h".
7960 (compile_file): Call optimization_records_finish.
7961 (do_compile): Call optimization_records_start.
7962 * tree-ssa-live.c: Include optinfo.h.
7963 (remove_unused_scope_block_p): Retain inlining information if
7964 optinfo_wants_inlining_info_p returns true.
7966 2018-07-20 Richard Biener <rguenther@suse.de>
7969 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
7970 to cover -flto-partition=none.
7972 2018-07-20 Martin Liska <mliska@suse.cz>
7974 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
7975 (get_decl_source_range): Remove unused function.
7977 2018-07-20 Richard Biener <rguenther@suse.de>
7979 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
7980 (struct vn_phi_s): Likewise.
7981 (struct vn_reference_s): Likewise.
7982 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
7983 for searching the slot of an entry known to be in the hash itself.
7984 (vn_phi_hasher::equal): Likewise.
7985 (vn_reference_hasher::equal): Likewise.
7986 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
7988 (optimistic_info, current_info): Remove, keeping only valid_info.
7989 (vn_reference_lookup_1): Remove fallback lookup.
7990 (vn_reference_lookup_2): Likewise.
7991 (vn_nary_op_lookup_1): Likewise.
7992 (vn_phi_lookup): Likewise.
7993 (vn_nary_build_or_lookup_1): Make sure to not chain the built
7995 (vn_reference_insert): Adjust, chain the inserted hash element
7996 at last_inserted_ref.
7997 (vn_reference_insert_pieces): Likewise.
7998 (visit_reference_op_call): Likewise.
7999 (vn_nary_op_insert_into): Chain the inserted hash element at
8001 (vn_nary_op_insert_pieces): Adjust.
8002 (vn_nary_op_insert): Likewise.
8003 (vn_nary_op_insert_stmt): Likewise.
8004 (vn_phi_insert): Adjust, chain the inserted hash element at
8006 (process_scc): Remove clearing and copying the optimistic
8007 table. Instead remove elements inserted during an optimistic
8008 iteration from the single table we maintain.
8009 (init_scc_vn): Adjust.
8010 (free_scc_vn): Likewise.
8011 (sccvn_dom_walker::record_cond): Likewise.
8012 (sccvn_dom_walker::after_dom_children): Likewise.
8014 2018-07-19 Martin Sebor <msebor@redhat.com>
8016 PR tree-optimization/84047
8017 PR tree-optimization/83776
8018 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
8019 (check_array_bounds): Call it.
8021 2018-07-19 Martin Sebor <msebor@redhat.com>
8023 * align.h (align_flags): Use member initialization.
8025 2018-07-19 David Malcolm <dmalcolm@redhat.com>
8027 * Makefile.in (OBJS): Add optinfo.o.
8028 * coretypes.h (class symtab_node): New forward decl.
8029 (struct cgraph_node): New forward decl.
8030 (class varpool_node): New forward decl.
8031 * dump-context.h: New file.
8032 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
8034 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
8035 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
8036 (set_alt_dump_file): Likewise.
8037 (dump_context::~dump_context): New dtor.
8038 (dump_gimple_stmt): Move implementation to...
8039 (dump_context::dump_gimple_stmt): ...this new member function.
8040 Add the stmt to any pending optinfo, creating one if need be.
8041 (dump_gimple_stmt_loc): Move implementation to...
8042 (dump_context::dump_gimple_stmt_loc): ...this new member function.
8043 Start a new optinfo and add the stmt to it.
8044 (dump_gimple_expr): Move implementation to...
8045 (dump_context::dump_gimple_expr): ...this new member function.
8046 Add the stmt to any pending optinfo, creating one if need be.
8047 (dump_gimple_expr_loc): Move implementation to...
8048 (dump_context::dump_gimple_expr_loc): ...this new member function.
8049 Start a new optinfo and add the stmt to it.
8050 (dump_generic_expr): Move implementation to...
8051 (dump_context::dump_generic_expr): ...this new member function.
8052 Add the tree to any pending optinfo, creating one if need be.
8053 (dump_generic_expr_loc): Move implementation to...
8054 (dump_context::dump_generic_expr_loc): ...this new member
8055 function. Add the tree to any pending optinfo, creating one if
8057 (dump_printf): Move implementation to...
8058 (dump_context::dump_printf_va): ...this new member function. Add
8059 the text to any pending optinfo, creating one if need be.
8060 (dump_printf_loc): Move implementation to...
8061 (dump_context::dump_printf_loc_va): ...this new member function.
8062 Start a new optinfo and add the stmt to it.
8063 (dump_dec): Move implementation to...
8064 (dump_context::dump_dec): ...this new member function. Add the
8065 value to any pending optinfo, creating one if need be.
8066 (dump_context::dump_symtab_node): New member function.
8067 (dump_context::get_scope_depth): New member function.
8068 (dump_context::begin_scope): New member function.
8069 (dump_context::end_scope): New member function.
8070 (dump_context::ensure_pending_optinfo): New member function.
8071 (dump_context::begin_next_optinfo): New member function.
8072 (dump_context::end_any_optinfo): New member function.
8073 (dump_context::s_current): New global.
8074 (dump_context::s_default): New global.
8075 (dump_scope_depth): Delete global.
8076 (dumpfile_ensure_any_optinfo_are_flushed): New function.
8077 (dump_symtab_node): New function.
8078 (get_dump_scope_depth): Reimplement in terms of dump_context.
8079 (dump_begin_scope): Likewise.
8080 (dump_end_scope): Likewise.
8081 (selftest::temp_dump_context::temp_dump_context): New ctor.
8082 (selftest::temp_dump_context::~temp_dump_context): New dtor.
8083 (selftest::verify_item): New function.
8084 (ASSERT_IS_TEXT): New macro.
8085 (ASSERT_IS_TREE): New macro.
8086 (ASSERT_IS_GIMPLE): New macro.
8087 (selftest::test_capture_of_dump_calls): New test.
8088 (selftest::dumpfile_c_tests): Call it.
8089 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
8090 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
8091 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
8092 descriptive comment.
8093 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
8094 (dump_node, dump_bb): Move these unrelated decls.
8095 (class dump_manager): Add leading comment.
8096 * optinfo.cc: New file.
8097 * optinfo.h: New file.
8099 2018-07-19 Michael Collison <michael.collison@arm.com>
8100 Richard Henderson <rth@redhat.com>
8102 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
8103 (subti): Handle op1 zero.
8104 (subvti4, usub4ti4): New.
8105 (*sub<GPI>3_compare1_imm): New.
8106 (sub<GPI>3_carryinCV): New.
8107 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
8108 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
8110 2018-07-19 Michael Collison <michael.collison@arm.com>
8111 Richard Henderson <rth@redhat.com>
8113 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
8114 (addti3): Create simpler code if low part is already known to be 0.
8115 (addvti4, uaddvti4): New.
8116 (*add<GPI>3_compareC_cconly_imm): New.
8117 (*add<GPI>3_compareC_cconly): New.
8118 (*add<GPI>3_compareC_imm): New.
8119 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
8120 handle constants within this pattern..
8121 (*add<GPI>3_compareV_cconly_imm): New.
8122 (*add<GPI>3_compareV_cconly): New.
8123 (*add<GPI>3_compareV_imm): New.
8124 (add<GPI>3_compareV): New.
8125 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
8126 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
8127 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
8128 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
8129 with 'comparison' operator.
8130 (*add<GPI>3_compareV_cconly_imm): Ditto.
8131 (*add<GPI>3_compareV_cconly): Ditto.
8132 (*add<GPI>3_compareV_imm): Ditto.
8133 (add<GPI>3_compareV): Ditto.
8134 (add<mode>3_carryinC): Ditto.
8135 (*add<mode>3_carryinC_zero): Ditto.
8136 (*add<mode>3_carryinC): Ditto.
8137 (add<mode>3_carryinV): Ditto.
8138 (*add<mode>3_carryinV_zero): Ditto.
8139 (*add<mode>3_carryinV): Ditto.
8141 2018-07-19 Michael Collison <michael.collison@arm.com>
8142 Richard Henderson <rth@redhat.com>
8144 * config/aarch64/aarch64-modes.def (CC_V): New.
8145 * config/aarch64/aarch64-protos.h
8146 (aarch64_addti_scratch_regs): Declare
8147 (aarch64_subvti_scratch_regs): Declare.
8148 (aarch64_expand_subvti): Declare.
8149 (aarch64_gen_unlikely_cbranch): Declare
8150 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
8151 for signed overflow using CC_Vmode.
8152 (aarch64_get_condition_code_1): Handle CC_Vmode.
8153 (aarch64_gen_unlikely_cbranch): New function.
8154 (aarch64_addti_scratch_regs): New function.
8155 (aarch64_subvti_scratch_regs): New function.
8156 (aarch64_expand_subvti): New function.
8158 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
8160 * config/aarch64/aarch64-option-extensions.def: New entry for profile
8162 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
8163 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
8166 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
8169 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
8170 address check not strict.
8172 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
8174 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
8176 (store_pair_lanes<mode>): Likewise.
8177 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
8178 enum value 'ADDR_QUERY_LDP_STP_N'.
8179 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
8180 (aarch64_print_address_internal): Add declaration.
8181 (aarch64_print_ldpstp_address): Remove.
8182 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
8183 (aarch64_print_operand): Change printing of 'y'.
8184 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
8185 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
8186 'true' rather than '1'.
8187 * config/aarch64/constraints.md (Uml): Likewise.
8188 (Uml): Rename to Umn.
8191 2018-07-19 Richard Biener <rguenther@suse.de>
8193 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
8195 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
8196 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
8197 (vn_reference_hasher): Likewise.
8198 (struct vn_tables_s): Remove obstack and alloc-pool members.
8199 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
8200 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
8201 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
8202 (vn_reference_insert_pieces): Likewise.
8203 (alloc_vn_nary_op_noinit): Adjust.
8204 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
8205 (vn_phi_eq): Adjust.
8206 (shared_lookup_phiargs): Remove.
8207 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
8208 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
8209 (visit_reference_op_call): Likewise.
8210 (copy_nary, copy_phi, copy_reference): Remove.
8211 (process_scc): Rewind the obstack when iterating. Do not
8212 copy the elements to valid_info but just move them from one
8213 hashtable to the other.
8214 (allocate_vn_table): Adjust.
8215 (free_vn_table): Likewise.
8216 (init_scc_vn): Likewise.
8217 (free_scc_vn): Likewise.
8219 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
8222 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
8223 indirect_return as function type attribute.
8224 (ix86_attribute_table): Change indirect_return to function
8226 * doc/extend.texi: Update indirect_return attribute.
8228 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
8230 * wide-int.h (widest2_int): New.
8231 * gimple-fold.c (arith_overflowed_p): Use it.
8232 * tree.h (widest2_int_cst): New.
8233 * tree-vrp.c (wide_int_binop_overflow): Rename from
8234 vrp_int_const_binop.
8235 Rewrite to work on trees.
8236 (extract_range_from_multiplicative_op_1): Abstract code to...
8237 (wide_int_range_min_max): ...here.
8238 (wide_int_range_cross_product): ...and here.
8239 (extract_range_from_binary_expr_1): Abstract overflow code to...
8240 (wide_int_range_mult_wrapping): ...here.
8241 * tree-vrp.h (wide_int_range_cross_product): New.
8242 (wide_int_range_mult_wrapping): New.
8244 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
8245 Julia Koval <julia.koval@intel.com>
8247 * config/i386/x86-tune-costs.h (skylake_memcpy,
8248 skylake_memset): Replace rep_prefix with unrolling for size 512.
8250 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
8253 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
8254 comparision with EQ_EXPR in last stmt.
8256 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
8258 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
8259 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
8260 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
8261 previously known as "PowerPC AltiVec Built-in Functions". Move
8262 some material to new subsubsections "PowerPC AltiVec Built-in
8263 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
8265 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
8266 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
8267 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
8268 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
8270 2018-07-18 Richard Biener <rguenther@suse.de>
8272 PR tree-optimization/86557
8273 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
8276 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
8278 * config/s390/s390.c (s390_function_profiler): Generate CFI.
8280 2018-07-17 Jeff Law <law@redhat.com>
8282 * config/arm/arm.c (get_label_padding): Update for recent
8283 changes to label_to_alignment.
8285 PR tree-optimization/86010
8286 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
8288 * config/mips/mips.c (vr4130_align_insns): Update for recent
8289 changes to label_to_alignment.
8291 * config/frv/frv.c (frv_label_align): Update for recent changes
8292 to label_to_alignment.
8294 * config/nios2/nios2.c (nios2_label_align): Update for recent
8295 changes which dropped ALIGN_LABELS_LOG.
8297 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
8299 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
8300 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
8302 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
8304 * config/arc/arc.c (arc_label_align): Use align_labels instead of
8305 deprecated align_labels_log.
8307 2018-07-17 Richard Biener <rguenther@suse.de>
8310 * dwarf2out.c (init_sections_and_labels): Always generate
8311 a debug_line_str_section for early LTO debug.
8312 (dwarf2out_finish): Reset debug_line_str_hash output early.
8313 Bump counter for extra dwarf5 .debug_loc labels to not conflict
8315 (dwarf2out_early_finish): Output debug_line_str.
8317 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
8319 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
8320 index register on z196 or later.
8322 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
8324 * config/s390/s390.c (s390_default_align): Set default function
8326 (s390_override_options_after_change): Call s390_default align.
8327 (s390_option_override_internal): Call s390_default_align.
8328 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
8330 2018-07-17 Jakub Jelinek <jakub@redhat.com>
8333 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
8334 to _looptemp_ clauses, other than the first two.
8336 2018-07-17 Martin Liska <mliska@suse.cz>
8338 * opts.c: Do not enable OPT_falign_* for -Os.
8340 2018-07-17 Martin Liska <mliska@suse.cz>
8342 * align.h (MAX_CODE_ALIGN): New.
8343 (MAX_CODE_ALIGN_VALUE): New.
8344 * common/config/i386/i386-common.c (ix86_handle_option):
8345 (MAX_CODE_ALIGN): Moved to align.h.
8346 * final.c (MAX_CODE_ALIGN): Likewise.
8347 * opts.c (parse_and_check_align_values):
8348 (MAX_CODE_ALIGN): Likewise.
8349 (MAX_CODE_ALIGN_VALUE): Likewise.
8351 2018-07-17 Martin Liska <mliska@suse.cz>
8353 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
8354 in order to fulfil coding style.
8355 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
8356 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
8357 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
8358 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
8359 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
8360 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
8361 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
8362 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
8363 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
8365 2018-07-17 Martin Liska <mliska@suse.cz>
8367 * align.h: New file.
8368 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
8370 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
8371 align_flags of label_to_alignment.
8372 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
8374 * config/m68k/m68k.c: Do not use removed align_labels_value and
8376 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
8377 (LOOP_ALIGN): Likewise.
8378 (LABEL_ALIGN): Likewise.
8379 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
8380 Remove not used macro.
8381 (rs6000_loop_align): Change return type to align_flags.
8382 (rs6000_loop_align_max_skip): Remove.
8383 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
8384 Change return type to align_flags.
8385 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
8386 Remove not used macro.
8387 (rs6000_loop_align): Change return type to align_flags.
8388 (rs6000_loop_align_max_skip): Remove.
8389 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
8390 * config/rx/rx-protos.h (rx_align_for_label): Make it
8392 * config/rx/rx.c (rx_align_for_label): Change return type
8394 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
8396 into align_flags class.
8397 (LABEL_ALIGN): Likewise.
8398 (LOOP_ALIGN): Likewise.
8399 * config/s390/s390.c (s390_label_align): Use align_flags
8401 (s390_asm_output_function_label): Likewise.
8402 * config/sh/sh.c (sh_override_options_after_change):
8403 Use align_flags class directly without macros.
8404 (find_barrier): Likewise.
8405 (barrier_align): Likewise.
8406 (sh_loop_align): Likewise.
8407 * config/spu/spu.c (spu_option_override):
8408 Use align_flags_tuple::get_value instead of removed macros.
8409 (spu_sched_init): Likewise.
8410 * config/spu/spu.h (GTY): Likewise.
8411 * config/visium/visium.c (visium_option_override):
8412 Set "8" as default secondary alignment.
8413 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
8414 in order to guarantee secondary alignment of 8.
8415 * coretypes.h: Include align.h header file.
8416 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
8417 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
8418 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
8419 * doc/tm.texi.in: Likewise.
8420 * final.c (struct label_alignment): Remove not used structure.
8421 (LABEL_ALIGN): Change type to align_flags.
8422 (LOOP_ALIGN): Likewise.
8423 (JUMP_ALIGN): Likewise.
8424 (default_loop_align_max_skip): Remove.
8425 (default_label_align_max_skip): Likewise.
8426 (default_jump_align_max_skip): Likewise.
8427 (default_label_align_after_barrier_max_skip):
8428 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
8429 (LABEL_TO_MAX_SKIP): Remove.
8430 (label_to_alignment): Return align_flags type instead of integer.
8431 (label_to_max_skip): Remove.
8432 (align_fuzz): Use align_flags type.
8433 (compute_alignments): Use align_flags type and use align_flags::max
8434 to combine multiple alignments.
8435 (grow_label_align): Grow vec instead of C array.
8436 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
8437 (shorten_branches): Use align_flags type and use align_flags::max
8438 to combine multiple alignments.
8439 (final_scan_insn_1): Remove usage of secondary alignment that comes
8440 from label alignment, but instead use proper secondary alignment
8441 which is computed in grow_label_align.
8442 * flags.h (struct align_flags_tuple): Move to align.h.
8443 (struct align_flags): Likewise.
8444 (state_align_loops): Rename to align_loops.
8445 (state_align_jumps): Rename to align_jumps.
8446 (state_align_labels): Rename to align_labels.
8447 (state_align_functions): Rename to align_functions.
8448 (align_loops_log): Remove.
8449 (align_jumps_log): Remove.
8450 (align_labels_log): Remove.
8451 (align_functions_log): Remove.
8452 (align_loops_max_skip): Remove.
8453 (align_jumps_max_skip): Remove.
8454 (align_labels_max_skip): Remove.
8455 (align_functions_max_skip): Remove.
8456 (align_loops_value): Remove.
8457 (align_jumps_value): Remove.
8458 (align_labels_value): Remove.
8459 (align_functions_value): Remove.
8460 * output.h (label_to_alignment): Change return type to align_flags.
8461 (label_to_max_skip): Remove.
8462 * target.def: Remove loop_align_max_skip, label_align_max_skip,
8463 jump_align_max_skip macros.
8464 * targhooks.h (default_loop_align_max_skip): Remove.
8465 (default_label_align_max_skip): Likewise.
8466 (default_jump_align_max_skip): Likewise.
8467 (default_label_align_after_barrier_max_skip): Remove.
8468 * toplev.c (read_log_maxskip): Use ::normalize function.
8469 (parse_N_M): Remove not used argument and also call ::normalize.
8470 (parse_alignment_opts): Do not pass unused arguments.
8471 * varasm.c (assemble_start_function): Use directly align_functions
8472 instead of removed macros.
8473 * system.h: Do not poison removed macros.
8475 2018-07-17 Jakub Jelinek <jakub@redhat.com>
8478 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
8479 and cond temporaries don't have reference type if iterator has
8480 pointer type. For init use &for_pre_body instead of pre_p if
8481 for_pre_body is non-empty.
8483 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
8485 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
8486 double-double modes to SFmode directly directly.
8487 (trunc<mode>sf2_fprs): Delete.
8489 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
8491 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
8492 for conversions between IFmode and the decimal floating point modes.
8493 (init_float128_ieee): Use the correct names for conversions between
8494 KFmode and the decimal floating point modes.
8496 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
8498 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
8499 for the conversions between TDmode and IFmode.
8500 (init_float128_ieee): Use more correct names for the conversions
8501 between TDmode and KFmode.
8503 2018-07-16 Jakub Jelinek <jakub@redhat.com>
8505 PR tree-optimization/86526
8506 * builtins.c (expand_builtin_memcmp): Formatting fixes.
8507 (inline_expand_builtin_string_cmp): Likewise.
8508 (inline_string_cmp): Likewise. Use c_readstr instead of
8509 builtin_memcpy_read_str. Add unit_mode temporary.
8511 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
8514 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
8515 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
8517 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
8519 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
8520 Alphabetize prototypes of built-in functions, separating out
8521 built-in functions that are listed in this section but should be
8522 described elsewhere.
8524 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
8527 * expmed.c (emit_store_flag): Do not emit setcc followed by a
8528 conditional move when trapping comparison was split to a
8529 non-trapping one (and vice versa).
8531 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
8533 * config/s390/s390.c (s390_function_profiler): Generate nops
8534 instead of profiler call sequences.
8535 * config/s390/s390.opt: Add the new option.
8537 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
8539 * config/s390/s390.c (s390_function_profiler): Generate
8540 __mcount_loc section.
8541 * config/s390/s390.opt: Add the new option.
8543 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
8545 * common.opt: Add the new warning.
8546 * config/s390/s390.c (s390_function_profiler): Emit "brasl
8547 %r0,__fentry__" when -mfentry is specified.
8548 (s390_option_override_internal): Disallow -mfentry for 31-bit
8550 * config/s390/s390.opt: Add the new option.
8552 2018-07-16 Richard Biener <rguenther@suse.de>
8555 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
8556 for function-local FUNCTION_DECL and RESULT_DECL immediately.
8558 2018-07-16 Martin Liska <mliska@suse.cz>
8561 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
8564 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
8566 * config/arc/arcHS.md: Update ARCHS scheduling rules.
8568 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
8570 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
8572 * config/arc/arc-cpus.def (hs4x): New cpu.
8574 * config/arc/arc-tables.opt: Regenerate.
8575 * config/arc/arc.c (arc_sched_issue_rate): New function.
8576 (TARGET_SCHED_ISSUE_RATE): Define.
8577 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
8578 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
8580 (attr tune): Add ARCHS4x tune values.
8581 (attr tune_dspmpy): Define.
8582 (*tst): Correct instruction type.
8583 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
8584 * config/arc/arcHS4x.md: New file.
8585 * config/arc/fpu.md: Update instruction type attributes.
8586 * config/arc/t-multilib: Regenerate.
8588 2018-07-16 Tom de Vries <tdevries@suse.de>
8591 * var-tracking.c (vt_initialize): Fix pre_dec handling.
8593 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8595 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
8598 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
8600 PR tree-optimization/86514
8601 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
8602 conversion to a boolean type from a type with greater precision.
8604 2018-07-16 Tom de Vries <tdevries@suse.de>
8606 * var-tracking.c (vt_initialize): Print adjusted insn slim if
8607 dump_flags request TDF_SLIM.
8609 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
8611 * fold-const.c (int_const_binop_1): Abstract...
8612 (wide_int_binop): ...wide int code here.
8613 (poly_int_binop): ...poly int code here.
8614 Abstract the rest of int_const_binop_1 into int_const_binop.
8615 * fold-const.h (wide_int_binop): New.
8616 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
8617 Remove useless PLUS/MINUS_EXPR case.
8618 (zero_nonzero_bits_from_vr): Move wide int code...
8619 (zero_nonzero_bits_from_bounds): ...here.
8620 (extract_range_from_binary_expr_1): Move mask optimization code...
8621 (range_easy_mask_min_max): ...here.
8622 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
8623 (range_easy_mask_min_max): New.
8625 2018-07-15 Jeff Law <law@redhat.com>
8628 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
8631 2018-07-14 Jim Wilson <jimw@sifive.com>
8633 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
8635 2018-07-14 Paul Koning <ni1d@arrl.net>
8637 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
8639 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
8641 * lto-streamer-out.c (copy_function_or_variable): Dump info about
8644 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
8645 Steve Munroe <munroesj52@gmail.com>
8647 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
8648 (_mm_andnot_si128): Likewise.
8649 (_mm_or_si128): Likewise.
8650 (_mm_xor_si128): Likewise.
8652 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
8655 * builtins.c (expand_builtin_memcmp): Inline the calls first
8656 when result_eq is false.
8657 (expand_builtin_strcmp): Inline the calls first.
8658 (expand_builtin_strncmp): Likewise.
8659 (inline_string_cmp): New routine. Expand a string compare
8660 call by using a sequence of char comparison.
8661 (inline_expand_builtin_string_cmp): New routine. Inline expansion
8662 a call to str(n)cmp/memcmp.
8663 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
8665 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
8667 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
8669 * config/arm/driver-arm.c: Include arm-native.h.
8670 (host_detect_local_cpu): Use auto-generated data tables.
8671 (vendors, arm_cpu_table): Delete. Move part information to ...
8672 * config/arm/arm-cpus.in: ... here.
8673 * config/arm/parsecpu.awk (gen_native): New function.
8674 (vendor, part): New CPU fields.
8675 (END): Add support for building the native CPU detection tables.
8676 * config/arm/t-arm (arm-native.h): Add build rule.
8677 (driver-arm.o): Add dependency on arm-native.h.
8679 2018-07-13 Richard Biener <rguenther@suse.de>
8682 * match.pd (addr1 - addr2): Allow either of the operand to
8685 2018-07-13 Tom de Vries <tdevries@suse.de>
8687 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
8690 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
8692 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
8693 arrays instead of numbered variables.
8695 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
8697 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
8698 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
8700 2018-07-13 Richard Biener <rguenther@suse.de>
8703 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
8704 instead of get_context_die.
8706 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
8707 Richard Biener <rguenther@suse.de>
8710 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
8711 that the loop latch destination where phi is defined.
8713 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
8715 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
8716 (riscv_expand_epilogue): Add assertion to check interrupt mode.
8717 (riscv_set_current_function): Extract getting interrupt type to new
8719 (riscv_get_interrupt_type): New function.
8720 (riscv_merge_decl_attributes): New function, checking interrupt type is
8722 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
8724 2018-07-12 Paul Koning <ni1d@arrl.net>
8726 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
8729 2018-07-12 Paul Koning <ni1d@arrl.net>
8731 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
8732 zero reference, add doloop_end instead.
8733 * doc/md.texi (decrement_and_branch_until_zero): Remove.
8734 (Looping patterns): Remove decrement_and_branch_until_zero. Add
8735 detail for doloop_end.
8737 2018-07-12 Martin Sebor <msebor@redhat.com>
8740 * attribs.c (decl_attributes): Reject conflicting attributes before
8741 calling attribute handlers.
8743 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
8745 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
8747 (gcc::dump_manager::get_dump_file_name): likewise.
8748 (dump_begin): Likewise.
8749 * dumpfile.h (dump_begin): Update prototype.
8750 (gcc::dump_manager::get_dump_file_name,
8751 gcc::dump_manager::get_dump_file_name): Update prototype.
8753 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
8755 * internal-fn.h (vectorizable_internal_fn_p): New function.
8756 * tree-vect-slp.c (compatible_calls_p): Likewise.
8757 (vect_build_slp_tree_1): Remove nops argument. Handle calls
8758 to internal functions.
8759 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
8761 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
8763 * fold-const.h (inverse_conditions_p): Declare.
8764 * fold-const.c (inverse_conditions_p): New function.
8765 * match.pd: Use inverse_conditions_p. Add folds of view_converts
8766 that test the inverse condition of a conditional internal function.
8767 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
8768 * internal-fn.c (internal_fn_mask_index): Handle conditional
8770 (vectorized_internal_fn_supported_p): New function.
8771 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
8772 (any_pred_load_store): Replace with...
8773 (need_to_predicate): ...this new variable.
8774 (redundant_ssa_names): New variable.
8775 (ifcvt_can_use_mask_load_store): Move initial checks to...
8776 (ifcvt_can_predicate): ...this new function. Handle tree codes
8777 for which a conditional internal function exists.
8778 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
8779 instead of ifcvt_can_use_mask_load_store. Update after variable
8781 (predicate_load_or_store): New function, split out from
8782 predicate_mem_writes.
8783 (check_redundant_cond_expr): New function.
8784 (value_available_p): Likewise.
8785 (predicate_rhs_code): Likewise.
8786 (predicate_mem_writes): Rename to...
8787 (predicate_statements): ...this. Use predicate_load_or_store
8788 and predicate_rhs_code.
8789 (combine_blocks, tree_if_conversion): Update after above name changes.
8790 (ifcvt_local_dce): Handle redundant_ssa_names.
8791 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
8792 general conditional functions.
8793 * tree-vect-stmts.c (vectorizable_call): Likewise.
8795 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
8796 Alan Hayward <alan.hayward@arm.com>
8797 David Sherwood <david.sherwood@arm.com>
8799 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
8800 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
8801 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
8802 plus and minus and convert them into IFN_COND_FMA-based sequences.
8803 (convert_mult_to_fma): Handle conditional plus and minus.
8805 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
8807 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
8808 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
8809 (cond_fnms_optab): New optabs.
8810 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
8812 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
8813 * internal-fn.h (get_conditional_internal_fn): Declare.
8814 (get_unconditional_internal_fn): Likewise.
8815 * internal-fn.c (cond_ternary_direct): New macro.
8816 (expand_cond_ternary_optab_fn): Likewise.
8817 (direct_cond_ternary_optab_supported_p): Likewise.
8818 (FOR_EACH_COND_FN_PAIR): Likewise.
8819 (get_conditional_internal_fn): New function.
8820 (get_unconditional_internal_fn): Likewise.
8821 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
8822 (gimple_match_op::gimple_match_op): Add a new overload for 5
8824 (gimple_match_op::set_op): Likewise.
8825 (gimple_resimplify5): Declare.
8826 * genmatch.c (decision_tree::gen): Generate simplifications for
8828 * gimple-match-head.c (gimple_simplify): Define an overload for
8829 5 operands. Handle calls with 5 arguments in the top-level overload.
8830 (convert_conditional_op): Handle conversions from unconditional
8831 internal functions to conditional ones.
8832 (gimple_resimplify5): New function.
8833 (build_call_internal): Pass a fifth operand.
8834 (maybe_push_res_to_seq): Likewise.
8835 (try_conditional_simplification): Try converting conditional
8836 internal functions to unconditional internal functions.
8837 Handle 3-operand unconditional forms.
8838 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
8839 Define ternary equivalents of the current rules for binary conditional
8841 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
8843 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
8844 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
8845 (optab): Handle them.
8846 (SVE_COND_FP_TERNARY): New int iterator.
8847 (sve_fmla_op, sve_fmad_op): New int attributes.
8848 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
8849 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
8850 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
8852 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
8854 * target.def (preferred_else_value): New target hook.
8855 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
8856 * doc/tm.texi: Regenerate.
8857 * targhooks.h (default_preferred_else_value): Declare.
8858 * targhooks.c (default_preferred_else_value): New function.
8859 * internal-fn.h (conditional_internal_fn_code): Declare.
8860 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
8861 (get_conditional_internal_fn): Use it.
8862 (conditional_internal_fn_code): New function.
8863 * gimple-match.h (gimple_match_cond): New struct.
8864 (gimple_match_op): Add a cond member function.
8865 (gimple_match_op::gimple_match_op): Update all forms to take a
8867 * genmatch.c (expr::gen_transform): Use the same condition as res_op
8868 for the suboperation, but don't specify a particular else_value.
8869 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
8870 (visit_nary_op, visit_reference_op_load): Pass
8871 gimple_match_cond::UNCOND to the gimple_match_op constructor.
8872 * gimple-match-head.c: Include tree-eh.h
8873 (convert_conditional_op): New function.
8874 (maybe_resimplify_conditional_op): Likewise.
8875 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
8876 (gimple_resimplify2): Likewise.
8877 (gimple_resimplify3): Likewise.
8878 (gimple_resimplify4): Likewise.
8879 (maybe_push_res_to_seq): Return null for conditional operations.
8880 (try_conditional_simplification): New function.
8881 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
8883 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
8885 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
8887 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
8889 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
8891 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
8893 (hash_tree): Do not hash DECL_FCONTEXT
8894 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
8895 Do not stream DECL_FCONTEXT.
8896 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
8897 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
8899 2018-07-12 Richard Biener <rguenther@suse.de>
8902 * dwarf2out.c (gen_block_die): Only output blocks when they have
8903 at least one !DECL_IGNORED_P variable.
8905 2018-07-12 Richard Biener <rguenther@suse.de>
8908 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
8909 Remove -mieee-fp handling.
8911 2018-07-12 Richard Biener <rguenther@suse.de>
8913 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
8914 left-over from last patch.
8916 2018-07-12 Jakub Jelinek <jakub@redhat.com>
8918 PR tree-optimization/86492
8919 * gimple-ssa-store-merging.c
8920 (imm_store_chain_info::coalesce_immediate_stores): Call
8921 check_no_overlap even for the merge_overlapping case. Formatting fix.
8923 2018-07-12 Richard Biener <rguenther@suse.de>
8926 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
8927 move possibly trapping operations into the conditional.
8929 2018-07-12 Richard Biener <rguenther@suse.de>
8931 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
8932 (vn_lookup_simplify_result): Remove recursion limit applied
8934 (vn_nary_build_or_lookup_1): Adjust.
8935 (try_to_simplify): Likewise.
8936 * gimple-match-head.c (gimple_resimplify1): Instead apply one
8938 (gimple_resimplify2): Likewise.
8939 (gimple_resimplify3): Likewise.
8940 (gimple_resimplify4): Likewise.
8942 2018-07-11 Jakub Jelinek <jakub@redhat.com>
8944 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
8945 Use __mmask64 type instead of __mmask8 for __M argument.
8946 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
8947 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
8949 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
8950 __mmask16 for __M argument.
8951 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
8952 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
8953 to __mmask16 instead of __mmask8.
8954 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
8955 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
8956 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
8957 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
8958 instead of __mmask16 for __U argument.
8959 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
8960 __mmask16 instead of __mmask8 for __U argument.
8961 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
8963 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
8965 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
8967 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
8969 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
8971 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
8973 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
8975 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
8977 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
8979 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
8981 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
8982 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
8983 return type as well as __M argument type and all casts from __mmask8
8985 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
8986 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
8987 return type as well as __M argument type and all casts from __mmask8
8989 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
8990 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
8991 return type as well as __M argument type and all casts from __mmask8
8993 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
8994 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
8995 return type as well as __M argument type and all casts from __mmask8
8997 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
8998 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
9001 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
9003 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
9004 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
9007 2018-07-11 Paul Koning <ni1d@arrl.net>
9009 * doc/md.texi (define_subst): Document how multiple occurrences of
9010 the same argument in the replacement pattern are handled.
9012 2018-07-11 Paul Koning <ni1d@arrl.net>
9014 * doc/extend.texi (Common Variable Attributes): Move "mode" into
9016 (Common Type Attributes): Add "mode" attribute.
9018 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
9020 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
9021 stream DECL_ORIGINAL_TYPE.
9022 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
9023 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
9024 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
9025 Do not walk original type.
9026 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
9028 (write_ts_decl_non_common_tree_pointers): Do not stream
9030 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
9031 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
9033 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
9035 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
9036 thread twice from the same starting edge.
9038 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
9040 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
9041 * gimple.c (gimple_call_nonnull_result_p): ...here...
9042 (gimple_call_nonnull_arg): ...and here.
9043 * gimple.h (gimple_call_nonnull_result_p): New.
9044 (gimple_call_nonnull_arg): New.
9046 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
9048 * config/arm/arm-cpus.in: Move information from fpu field of each
9049 cpu definition to the isa field.
9050 * config/arm/parsecpu.awk (fpu): Delete match rule.
9051 (gen_comm_data): Don't add bits from the CPU's FPU entry.
9053 2018-07-11 Richard Biener <rguenther@suse.de>
9056 * dwarf2out.c (init_sections_and_labels): Use
9057 output_asm_line_debug_info consistently.
9058 (dwarf2out_early_finish): Likewise.
9059 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
9062 2018-07-11 Richard Biener <rguenther@suse.de>
9064 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
9065 Rework father_bb setting in a way to avoid propagating constants
9066 multiple times on a loop body.
9068 2018-07-10 Mark Wielaard <mark@klomp.org>
9071 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
9073 2018-07-10 Richard Biener <rguenther@suse.de>
9075 * hash-map.h (hash_map::iterator::operator*): Return
9076 references to key and value.
9078 2018-07-10 Jakub Jelinek <jakub@redhat.com>
9081 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
9082 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
9083 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
9084 (gimplify_omp_for): For composite loops, move outer
9085 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
9086 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
9087 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
9088 TREE_LIST for both the original class iterator and the "last" helper
9089 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
9090 loop, remember has_decl_expr from outer composite loops for the
9091 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
9093 2018-07-09 Martin Sebor <msebor@redhat.com>
9097 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
9098 accessing implicitly initialized array elements.
9099 * expr.c (string_constant): Handle string initializers of
9100 character arrays within aggregates.
9101 * gimple-fold.c (fold_array_ctor_reference): Add argument.
9102 Store element offset. As a special case, handle zero size.
9103 (fold_nonarray_ctor_reference): Same.
9104 (fold_ctor_reference): Add argument. Store subobject offset.
9105 * gimple-fold.h (fold_ctor_reference): Add argument.
9107 2018-07-09 Paul Koning <ni1d@arrl.net>
9109 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
9110 (pdp11_insn_cost): New function.
9111 (pdp11_md_asm_adjust): New function.
9112 (TARGET_INVALID_WITHIN_DOLOOP): Define.
9113 (pdp11_rtx_costs): Update to match machine better.
9114 (output_addr_const_pdp11): Correct format mismatch warnings.
9115 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
9116 * config/pdp11/pdp11.md: General change to add base_cost and/or
9117 length attributes for use by new pdp11_insn_cost function.
9118 (MIN_BRANCH): Correct definition.
9120 (doloop_end): Use standard pattern name for looping pattern.
9121 (doloop_end_nocc): New.
9122 (movsf): Add another constraint alternative.
9123 (zero_extendqihi2): Add constraint alternatives for not in place
9125 (zero_extendhisi2): Remove.
9126 (shift patterns): Add CC handling variants.
9130 (define_peephole2): New peephole to recognize mov that sets CC for
9133 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9135 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
9136 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
9139 2018-07-09 Jakub Jelinek <jakub@redhat.com>
9142 * real.c (real_nextafter): Return true if result is denormal.
9144 2018-07-09 Martin Liska <mliska@suse.cz>
9146 * common.opt: Add back wrongly removed attribute.
9148 2018-07-09 Richard Biener <rguenther@suse.de>
9151 * dwarf2out.c (gen_block_die): For an early generated DIE
9152 always output high/low PC attributes.
9154 2018-07-09 Tom de Vries <tdevries@suse.de>
9156 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
9157 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
9158 onto VAR_DECL with abstract origin.
9160 2018-07-07 Jim Wilson <jimw@sifive.com>
9162 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
9164 2018-07-07 Tom de Vries <tdevries@suse.de>
9166 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
9168 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
9170 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
9172 (combine_bound): Use wide-int overflow calculation instead of
9174 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
9176 * fold-const.c (int_const_binop_2): Same.
9177 (extract_muldiv_1): Same.
9178 (fold_div_compare): Same.
9179 (fold_abs_const): Same.
9181 * poly-int.h (add): Same.
9185 * predict.c (predict_iv_comparison): Same.
9186 * profile-count.c (slow_safe_scale_64bit): Same.
9187 * simplify-rtx.c (simplify_const_binary_operation): Same.
9188 * tree-chrec.c (tree_fold_binomial): Same.
9189 * tree-data-ref.c (split_constant_offset_1): Same.
9190 * tree-if-conv.c (idx_within_array_bound): Same.
9191 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
9192 * tree-ssa-phiopt.c (minmax_replacement): Same.
9193 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
9194 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
9195 * vr-values.c (vr_values::adjust_range_with_scev): Same.
9196 * wide-int.cc (wi::add_large): Same.
9197 (wi::mul_internal): Same.
9198 (wi::sub_large): Same.
9199 (wi::divmod_internal): Same.
9200 * wide-int.h: Change overflow type to overflow_type for neg, add,
9201 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
9202 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
9203 mul_internal, divmod_internal.
9204 (overflow_type): New enum.
9205 (accumulate_overflow): New.
9207 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
9209 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
9210 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
9212 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
9214 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
9215 argument is checked for zero before entering loop, avoid checking again.
9217 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
9219 * gimplify.h (generic_expr_could_trap_p): Set as global function.
9220 * gimplify.h (generic_expr_could_trap_p): Likwise.
9221 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
9223 2018-07-06 Jakub Jelinek <jakub@redhat.com>
9225 PR tree-optimization/86401
9226 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
9227 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
9228 (fold_bit_and_mask): ... here. New helper function for match.pd.
9229 * fold-const.h (fold_bit_and_mask): Declare.
9230 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
9232 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
9235 * target.def (translate_mode_attribute): New hook.
9236 * targhooks.h (default_translate_mode_attribute): Declare.
9237 * targhooks.c (default_translate_mode_attribute): New function.
9238 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
9239 * doc/tm.texi: Regenerate.
9240 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
9241 (rs6000_translate_mode_attribute): New function.
9243 2018-07-06 Paul Koning <ni1d@arrl.net>
9245 * doc/md.texi (define_split): Document DONE and FAIL.
9246 (define_peephole2): Ditto.
9248 2018-07-05 Jeff Law <law@redhat.com>
9250 PR tree-optimization/86010
9251 * tree-ssa-dse.c (compute_trims): More aggressively trim at
9252 both the head and tail of mem* and str* calls.
9254 2018-07-05 Jim Wilson <jimw@sifive.com>
9256 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
9258 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
9260 * config/aarch64/aarch64-simd.md: correct flags text for
9261 MIN_EXPR replacement.
9263 2018-07-05 James Clarke <jrtc27@jrtc27.com>
9265 * configure: Regenerated.
9267 2018-07-05 Carl Love <cel@us.ibm.com>
9269 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
9270 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
9271 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
9272 VSX_BUILTIN_DOUBLEL_V4SF.
9274 2018-07-05 Martin Sebor <msebor@redhat.com>
9277 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
9278 than its domain to compute its the upper bound of a char array.
9280 2018-07-05 Nathan Sidwell <nathan@acm.org>
9282 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
9283 * doc/cpp.texi: Update comment.
9284 * doc/tm.texi: Rebuilt.
9285 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
9286 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
9287 * doc/extend.texi (Backwards Compatibility): Clarify it is system
9288 headers affected by extern "C".
9289 * system.h: Poison NO_IMPLICIT_EXTERN_C.
9290 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
9291 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
9292 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
9293 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
9294 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
9295 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
9296 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
9297 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
9298 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
9299 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
9300 config/sparc/sp64-elf.h, config/spu/spu.h,
9301 config/stormy16/stormy16.h, config/v850/v850.h,
9302 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
9303 define NO_IMPLICIT_EXTERN_C.
9304 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
9306 2018-07-05 Tamar Christina <tamar.christina@arm.com>
9309 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
9310 instead of GET_MODE_SIZE when comparing Units.
9312 2018-07-05 Tamar Christina <tamar.christina@arm.com>
9315 * rtlanal.c (set_noop_p): Constrain on mode change,
9316 include hard-reg-set.h
9318 2018-07-05 Tamar Christina <tamar.christina@arm.com>
9320 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
9322 2018-07-05 Jakub Jelinek <jakub@redhat.com>
9325 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
9328 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
9330 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
9332 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
9335 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
9337 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
9339 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
9342 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
9344 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
9346 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
9347 neg_*_op* variables.
9349 2018-07-04 Martin Liska <mliska@suse.cz>
9351 * tree-switch-conversion.c: Define
9352 max_ratio_for_speed and max_ratio_for_size constants.
9354 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
9355 Martin Liska <mliska@suse.cz>
9360 * common.opt: Rename align options with 'str_' prefix.
9361 * common/config/i386/i386-common.c (set_malign_value): New
9363 (ix86_handle_option): Use it to set -falign-* options/
9364 * config/aarch64/aarch64-protos.h (struct tune_params): Change
9365 type from int to string.
9366 * config/aarch64/aarch64.c: Update default values from int
9368 * config/alpha/alpha.c (alpha_override_options_after_change):
9370 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
9371 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
9372 max skip conditionally.
9373 * config/i386/freebsd.h (SUBALIGN_LOG): New.
9374 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
9375 max skip conditionally.
9376 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
9377 max skip conditionally.
9378 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
9379 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
9380 max skip conditionally.
9381 * config/i386/i386.c (struct ptt): Change type from int to
9383 (ix86_default_align): Set default values.
9384 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
9385 max skip conditionally.
9386 * config/i386/iamcu.h (SUBALIGN_LOG): New.
9387 (ASM_OUTPUT_MAX_SKIP_ALIGN):
9388 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
9389 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
9390 max skip conditionally.
9391 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
9392 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
9393 * config/i386/x86-64.h (SUBALIGN_LOG): New.
9394 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
9395 max skip conditionally.
9396 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
9397 * config/ia64/ia64.c (ia64_option_override): Set default values
9398 for alignment options.
9399 * config/m68k/m68k.c: Handle new str_align_* options.
9400 * config/mips/mips.c (mips_set_compression_mode): Change
9402 (mips_option_override): Set default values for options.
9403 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
9405 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9407 * config/rx/rx.c (rx_option_override): Likewise.
9408 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
9409 (LABEL_ALIGN): Use align_labels_log.
9410 (LOOP_ALIGN): Use align_loops_align.
9411 * config/s390/s390.c (s390_asm_output_function_label): Use new
9413 * config/sh/sh.c (sh_override_options_after_change):
9414 Change type of constants.
9415 * config/spu/spu.c (spu_sched_init): Likewise.
9416 * config/sparc/sparc.c (sparc_option_override): Set default
9418 * config/visium/visium.c (visium_option_override): Likewise.
9419 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
9420 emit p2align format with last argument if it's not needed.
9421 * doc/invoke.texi: Document extended format of -falign-*.
9422 * final.c: Use align_labels alignment.
9423 * flags.h (struct target_flag_state): Change type to use
9425 (struct align_flags_tuple): New.
9426 (struct align_flags): Likewise.
9427 (align_loops_log): Redefine macro to use new types.
9428 (align_loops_max_skip): Redefine macro to use new types.
9429 (align_jumps_log): Redefine macro to use new types.
9430 (align_jumps_max_skip): Redefine macro to use new types.
9431 (align_labels_log): Redefine macro to use new types.
9432 (align_labels_max_skip): Redefine macro to use new types.
9433 (align_functions_log): Redefine macro to use new types.
9434 (align_loops): Redefine macro to use new types.
9435 (align_jumps): Redefine macro to use new types.
9436 (align_labels): Redefine macro to use new types.
9437 (align_functions): Redefine macro to use new types.
9438 (align_functions_max_skip): Redefine macro to use new types.
9439 (align_loops_value): New macro.
9440 (align_jumps_value): New macro.
9441 (align_labels_value): New macro.
9442 (align_functions_value): New macro.
9443 * function.c (invoke_set_current_function_hook): Propagate
9444 alignment values from flags to global variables default in
9446 * ipa-icf.c (sem_function::equals_wpa): Use
9447 cl_optimization_option_eq instead of memcmp.
9448 * lto-streamer.h (cl_optimization_stream_out): Support streaming
9450 (cl_optimization_stream_in): Likewise.
9451 * optc-save-gen.awk: Support strings in cl_optimization.
9452 * opth-gen.awk: Likewise.
9453 * opts.c (finish_options): Remove error checking of invalid
9455 (MAX_CODE_ALIGN): Remove.
9456 (MAX_CODE_ALIGN_VALUE): Likewise.
9457 (parse_and_check_align_values): New function.
9458 (check_alignment_argument): Likewise.
9459 (common_handle_option): Use check_alignment_argument.
9460 * opts.h (parse_and_check_align_values): Declare.
9461 * toplev.c (init_alignments): Remove.
9462 (read_log_maxskip): New.
9463 (parse_N_M): Likewise.
9464 (parse_alignment_opts): Likewise.
9465 (backend_init_target): Remove usage of init_alignments.
9466 * toplev.h (parse_alignment_opts): Declare.
9467 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
9469 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
9470 * tree.c (cl_option_hasher::equal): New.
9471 * varasm.c: Use new global macros.
9473 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
9475 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
9476 Use a simpler align directive also if MAXSKIP = ALIGN-1.
9477 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
9478 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
9479 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
9480 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
9481 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
9482 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
9483 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
9484 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
9485 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
9486 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
9487 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
9489 2018-07-04 Martin Liska <mliska@suse.cz>
9490 Jonathan Wakely <jwakely@redhat.com>
9492 * coverage.c: Use correct type.
9493 * doc/invoke.texi: Language correction.
9495 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
9498 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
9499 ENDBRANCH for non-tail call which may return via indirect branch.
9500 * doc/extend.texi: Document indirect_return attribute.
9502 2018-07-03 Martin Sebor <msebor@redhat.com>
9504 PR tree-optimization/86274
9505 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
9507 (format_floating): Correct handling of infinities and NaNs.
9509 2018-07-03 Martin Sebor <msebor@redhat.com>
9511 * print-tree.c (print_real_cst): New function.
9512 (print_node_brief): Call it.
9513 (print_node): Ditto.
9515 2018-07-03 Jeff Law <law@redhat.com>
9517 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
9518 into a single pattern.
9520 * config/h8300/h8300.md (ors code_iterator): New.
9521 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
9522 a single pattern and single splitter.
9523 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
9524 (iorqi3_1, xorqi3_1): Likewise.
9525 (iorqi3, xorqi3 expanders): Similarly.
9527 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
9528 (movmd_internal) into a single pattern using the P mode iterator.
9529 (movmd splitters): Similarly.
9530 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
9531 (movsd splitters): Similarly.
9533 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
9534 ADDB, ADDW and ADDL into a single ADD attribute which selects the
9535 right table based on the size of the operand.
9536 * config/h8300/h8300.md (length_table): Corresponding changes. All
9537 references to "addb", "addw" and "addl" changed to "add".
9538 (btst patterns): Merge two variants into a single pattern.
9539 (tstqi, tsthi): Likewise.
9540 (addhi3_incdec, addsi3_incdec): Likewise.
9541 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
9542 (mulhi3, mulsi3): Likewise.
9543 (udivhi3, udivsi3): Likewise.
9544 (divhi3, divsi3): Likewise.
9545 (andorqi3, andorhi3, andorsi3): Likewise.
9547 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
9550 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
9551 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
9553 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
9555 PR tree-optimization/85694
9556 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
9557 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
9559 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
9560 (<u>avg<mode>3_ceil): New patterns.
9562 2018-07-03 David Malcolm <dmalcolm@redhat.com>
9564 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
9565 scan-tree-dump directive.
9566 * gcc.dg/vect/slp-perm-2.c: Likewise.
9567 * gcc.dg/vect/slp-perm-3.c: Likewise.
9568 * gcc.dg/vect/slp-perm-5.c: Likewise.
9569 * gcc.dg/vect/slp-perm-6.c: Likewise.
9570 * gcc.dg/vect/slp-perm-7.c: Likewise.
9571 * gcc.dg/vect/slp-perm-8.c: Likewise.
9573 2018-07-03 Marek Polacek <polacek@redhat.com>
9576 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
9578 2018-07-03 Richard Biener <rguenther@suse.de>
9581 * tree-ssa-structalias.c (find_func_clobbers): Properly
9582 handle indirect calls.
9584 2018-07-03 Jeff Law <law@redhat.com>
9586 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
9587 (shifts): New code iterator.
9588 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
9589 expander. Fix HImode handling on H8/SX.
9590 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
9591 (subqi3, subhi3, subsi3 expanders): Likewise.
9592 (andqi3, andhi3, andsi3 expanders): Likewise.
9593 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
9594 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
9595 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
9596 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
9597 (zero_extendqihi2, zero_extendqisi2): Likewise.
9598 (extendqihi2, extendqisi2): Likewise.
9599 (rotlqi3, rotlhi3, rotlsi3): Likewise.
9600 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
9601 (rotlqi3_1, rotlhi3_1): Likewise.
9602 (logicalhi3_sn, logicalsi3_sn): Likewise.
9603 (logicalhi3, logicalsi3): Likewise.
9605 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
9607 * tree-vect-patterns.c (vect_recog_rotate_pattern)
9608 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
9609 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
9610 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
9611 type to append_pattern_def_seq instead of creating a stmt_vec_info
9613 (build_mask_conversion): Likewise. Remove vinfo argument.
9614 (vect_add_conversion_to_patterm): Likewise, renaming to...
9615 (vect_add_conversion_to_pattern): ...this.
9616 (vect_recog_mask_conversion_pattern): Update call to
9617 build_mask_conversion. Pass the vector type to
9618 append_pattern_def_seq here too.
9619 (vect_recog_gather_scatter_pattern): Update call to
9620 vect_add_conversion_to_pattern.
9622 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
9624 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
9625 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
9626 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
9627 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
9628 STMT_VINFO_PATTERN_DEF_SEQ to null here.
9629 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
9630 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
9631 append_pattern_def_seq instead of new_pattern_def_seq.
9632 (vect_recog_divmod_pattern): Do both of the above.
9633 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
9636 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
9638 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
9639 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
9640 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
9641 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
9642 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
9643 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
9644 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
9645 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
9646 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
9647 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
9648 parameter with a single stmt_vec_info.
9649 (vect_recog_func_ptr): Likewise.
9650 (vect_recog_gather_scatter_pattern): Likewise, folding in...
9651 (vect_try_gather_scatter_pattern): ...this.
9652 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
9653 the stmt_vec_info of the statement to be matched. Don't clear
9654 STMT_VINFO_RELATED_STMT.
9655 (vect_pattern_recog): Update call accordingly.
9657 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
9659 PR tree-optimization/85694
9660 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
9661 (uavgM3_ceil): Document new optabs.
9662 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
9663 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
9665 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
9666 (savg_ceil_optab): New optabs.
9667 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
9668 (vect_vect_recog_func_ptrs): Add it.
9669 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
9670 constant directly from the associated lhs.
9672 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
9674 * tree-vect-patterns.c (vect_split_statement): New function.
9675 (vect_convert_input): Use it to try to split an existing cast.
9677 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
9679 * poly-int.h (print_hex): New function.
9680 * dumpfile.h (dump_dec, dump_hex): Declare.
9681 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
9682 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
9683 min_input_precision, operation_precision and operation_sign.
9684 * tree-vect-patterns.c (vect_get_range_info): New function.
9685 (vect_same_loop_or_bb_p, vect_single_imm_use)
9686 (vect_operation_fits_smaller_type): Delete.
9687 (vect_look_through_possible_promotion): Add an optional
9688 single_use_p parameter.
9689 (vect_recog_over_widening_pattern): Rewrite to use new
9690 stmt_vec_info infomration. Handle one operation at a time.
9691 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
9692 (vect_truncatable_operation_p, vect_set_operation_type)
9693 (vect_set_min_input_precision): New functions.
9694 (vect_determine_min_output_precision_1): Likewise.
9695 (vect_determine_min_output_precision): Likewise.
9696 (vect_determine_precisions_from_range): Likewise.
9697 (vect_determine_precisions_from_users): Likewise.
9698 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
9699 (vect_vect_recog_func_ptrs): Put over_widening first.
9701 (vect_pattern_recog): Call vect_determine_precisions.
9703 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
9705 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
9706 statements that have been replaced by further pattern statements.
9707 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
9709 2018-07-03 Richard Biener <rguenther@suse.de>
9711 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
9712 always set *dt. Dump vectype in vectype overload.
9713 * dumpfile.h (dump_gimple_expr): New function.
9714 (dump_gimple_expr_loc): Likewise.
9715 * dumpfile.c (dump_gimple_expr): New function.
9716 (dump_gimple_expr_loc): Likewise.
9718 2018-07-02 Jeff Law <law@redhat.com>
9720 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
9721 the H8/300, H8/300H and H8/S variants into a single pattern.
9722 (movhi_h8300, movqi_h8300hs): Similarly.
9723 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
9724 (QHI mode iterator): New.
9726 * config/h8300/h8300.md: Remove trailing whitespace.
9728 2018-07-02 Jim Wilson <jimw@sifive.com>
9730 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
9731 instead of emit_insn for interrupt returns.
9732 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
9733 (riscv_sret, riscv_uret): Likewise.
9735 2018-07-02 David Malcolm <dmalcolm@redhat.com>
9737 * pretty-print.c (selftest::test_pp_format): Move save and restore
9738 of quotes to class auto_fix_quotes, and add an instance.
9739 * selftest.c: Include "intl.h".
9740 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
9741 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
9742 * selftest.h (selftest::auto_fix_quotes): New class.
9744 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
9746 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
9747 (aarch64_sve_prepare_conditional_op): Remove.
9748 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
9749 Allow aarch64_simd_reg_or_zero as select operand; remove
9750 the aarch64_sve_prepare_conditional_op call.
9751 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
9752 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
9753 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
9754 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
9755 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
9756 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
9757 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
9758 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
9759 and a splitters to match all of the *_any patterns.
9760 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
9762 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
9763 (SVE_COND_FP_BINARY_REV): Remove.
9764 (sve_int_op_rev, sve_fp_op_rev): New.
9765 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
9766 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
9767 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
9768 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
9769 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
9770 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
9771 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
9772 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
9773 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
9775 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
9776 Remove match_dup 1 from the inner unspec.
9777 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
9779 * config/aarch64/aarch64.md (movprfx): New attr.
9780 (length): Default movprfx to 8.
9781 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
9782 (*madd<SVE_I>, *msub<SVE_I): Likewise.
9783 (*<su>mul<SVE_I>3_highpart): Likewise.
9784 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
9785 (*v<ASHIFT><SVE_I>3): Likewise.
9786 (*<su><MAXMIN><SVE_I>3): Likewise.
9787 (*<su><MAXMIN><SVE_F>3): Likewise.
9788 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
9789 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
9790 (*div<SVE_F>4): Likewise.
9792 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
9794 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
9797 2018-07-02 Richard Biener <rguenther@suse.de>
9799 PR tree-optimization/86363
9800 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
9801 memset argument refers to a non-variable address.
9803 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
9805 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
9806 {PLUS,MINUS}_EXPR code to...
9807 (adjust_symbolic_bound): ...here,
9808 (combine_bound): ...here,
9809 (set_value_range_with_overflow): ...and here.
9811 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
9813 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
9815 (extract_range_from_abs_expr): ...here.
9817 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
9819 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
9820 -fno-omit-frame-pointer when not optimizing.
9822 2018-07-02 Martin Liska <mliska@suse.cz>
9825 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
9826 (propagate_nothrow): Likewise.
9828 2018-07-02 Martin Liska <mliska@suse.cz>
9831 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
9833 2018-07-02 David Malcolm <dmalcolm@redhat.com>
9835 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
9836 function in r262149, changing "loc" param from source_location to
9837 const dump_location_t &.
9838 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
9839 declaration, as above.
9841 2018-07-01 Paul Koning <ni1d@arrl.net>
9843 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
9844 -munit-asm, -mgnu-asm, -mdec-asm.
9845 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
9846 (pdp11_output_labelref): New.
9847 (pdp11_output_def): New.
9848 (pdp11_output_addr_vec_elt): New.
9849 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
9850 %# and %@ format codes.
9851 (pdp11_option_override): New.
9852 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
9853 (pdp11_output_ident): New.
9854 (pdp11_asm_named_section): New.
9855 (pdp11_asm_init_sections): New.
9856 (pdp11_file_start): New.
9857 (pdp11_file_end): New.
9858 (output_ascii): Use .ascii/.asciz for -mdec-asm.
9859 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
9860 %o, like %c but octal.
9861 (pdp11_option_override): New.
9862 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
9864 (DATA_SECTION_ASM_OP): Ditto.
9865 (READONLY_DATA_SECTION_ASM_OP): New.
9866 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
9867 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
9868 (ASM_OUTPUT_LABELREF): Ditto.
9869 (ASM_OUTPUT_DEF): Ditto.
9870 (ASM_OUTPUT_EXTERNAL): New.
9871 (ASM_OUTPUT_SOURCE_FILENAME): New.
9872 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
9873 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
9874 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
9875 %# and %@ format codes.
9876 * config/pdp11/pdp11.opt (mgnu-asm): New.
9877 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
9878 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
9879 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
9881 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
9883 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
9884 dereferencing path[] beyond its length.
9886 (debug_all_paths): New.
9887 (rewire_first_differing_edge): New.
9888 (adjust_paths_after_duplication): New.
9889 (duplicate_thread_path): Call adjust_paths_after_duplication.
9891 (thread_through_all_blocks): Add new argument to
9892 duplicate_thread_path.
9894 2018-06-30 Jim Wilson <jimw@sifive.com>
9896 * config/riscv/predicates.md (p2m1_shift_operand): New.
9897 (high_mask_shift_operand): New.
9898 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
9899 pattern using p2m1_shift_operand.
9900 (lshsi3_zero_extend_3+2): New combiner pattern using
9901 high_mask_shift_operand.
9903 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
9905 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
9907 (vect_recog_rotate_pattern): ...here.
9908 (vect_convert_input): Try to insert casts of invariants in the
9910 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
9911 preheader to be empty.
9913 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
9915 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
9916 vector type. If given, install it in the new statement's
9918 (vect_element_precision): New function.
9919 (vect_unpromoted_value): New struct.
9920 (vect_unpromoted_value::vect_unpromoted_value): New function.
9921 (vect_unpromoted_value::set_op): Likewise.
9922 (vect_look_through_possible_promotion): Likewise.
9923 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
9924 (vect_widened_op_tree, vect_convert_input): Likewise.
9925 (vect_convert_inputs, vect_convert_output): Likewise.
9926 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
9927 to handle the optional cast of the multiplication result and
9928 vect_widened_op_tree to detect the widened multiplication itself.
9929 Do not require the input and output of promotion casts to have
9930 the same sign, but base the signedness of the operation on the
9931 input rather than the result. If the pattern includes two
9932 promotions, check that those promotions have the same sign.
9933 Do not restrict the MULT_EXPR handling to a double-width result;
9934 handle quadruple-width results and wider. Use vect_convert_inputs
9935 to convert the inputs to the common type.
9936 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
9937 to handle the optional cast of the ABS result. Also allow a sign
9938 change or a sign extension between the ABS and MINUS.
9939 Use vect_widened_op_tree to detect the widened subtraction and use
9940 vect_convert_inputs to convert the inputs to the common type.
9941 (vect_handle_widen_op_by_const): Delete.
9942 (vect_recog_widen_op_pattern): New function.
9943 (vect_recog_widen_mult_pattern): Use it.
9944 (vect_recog_widen_shift_pattern): Likewise.
9945 (vect_recog_widen_sum_pattern): Use
9946 vect_look_through_possible_promotion to handle the promoted
9949 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
9951 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
9952 the containing gimple_seq *.
9953 * gimple-iterator.h (gsi_for_stmt): Declare it.
9954 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
9955 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
9956 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
9957 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
9958 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
9960 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
9962 (vect_mark_pattern_stmts): ...here. Handle cases in which the
9963 statement being replaced is part of an existing pattern
9964 definition sequence, inserting the new pattern statements before
9966 (vect_pattern_recog_1): Don't return a bool. If the statement
9967 is already part of a pattern, instead apply pattern matching
9968 to the pattern definition statements. Don't clear the
9969 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
9970 (vect_pattern_recog): Don't break after the first match;
9971 continue processing the pattern definition statements instead.
9972 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
9974 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
9976 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
9977 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
9978 (vect_recog_widen_sum_pattern): Use it.
9980 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
9982 * tree-vect-loop.c (vectorizable_reduction): Assert that the
9983 phi is not a pattern statement and has not been replaced by
9984 a pattern statement.
9985 * tree-vect-patterns.c (type_conversion_p): Don't check
9986 STMT_VINFO_IN_PATTERN_P.
9987 (vect_recog_vector_vector_shift_pattern): Likewise.
9988 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
9989 the pattern statement rather than the original statement; check
9990 directly for a WIDEN_MULT_EXPR here.
9991 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
9992 vect_is_simple_use to return the pattern statement rather
9993 than the original statement; use is_pattern_stmt_p to check
9994 for such a pattern statement.
9995 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
9996 to return the pattern statement rather than the original statement;
9997 don't do the same transformation here.
9998 (vect_is_simple_use): If the defining statement has been replaced
9999 by a pattern statement, return the pattern statement instead.
10000 Remove the corresponding (local) transformation from the vectype
10003 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
10005 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
10006 end and default to null.
10007 * tree-vect-loop.c (vect_create_epilog_for_reduction)
10008 (vectorizable_reduction): Update calls accordingly, dropping the
10009 gimple ** argument if the passed-back statement isn't needed.
10010 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
10011 (vect_recog_rotate_pattern): Likewise.
10012 (vect_recog_mask_conversion_pattern): Likewise.
10013 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
10014 (vect_mask_constant_operand_p): Likewise.
10015 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
10016 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
10017 (get_group_load_store_type, get_load_store_type): Likewise.
10018 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
10019 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
10020 (vectorizable_conversion, vectorizable_assignment): Likewise.
10021 (vectorizable_shift, vectorizable_operation): Likewise.
10022 (vectorizable_store, vect_is_simple_cond): Likewise.
10023 (vectorizable_condition, vectorizable_comparison): Likewise.
10024 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
10025 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
10026 and move it to the end. Cope with null def_stmt_outs.
10028 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
10030 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
10032 2018-06-29 Jeff Law <law@redhat.com>
10034 * config/v850/v850.c (v850_legitimate_address_p): Handle large
10035 displacements for TARGET_V850E2V3 and newer.
10036 (TARGET_LRA_P): Remove. Defaults to LRA now.
10037 * config/v850/v850.md (sign23byte_load): Remove.
10038 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
10039 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
10041 2018-06-29 Martin Liska <mliska@suse.cz>
10044 * coverage.c (coverage_init): Mangle full path name.
10045 * doc/invoke.texi: Document the change.
10046 * gcov-io.c (mangle_path): New.
10047 * gcov-io.h (mangle_path): Likewise.
10048 * gcov.c (mangle_name): Use mangle_path for path mangling.
10050 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10052 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
10053 if starting source register is not even.
10055 2018-06-29 Martin Liska <mliska@suse.cz>
10057 PR tree-optimization/86263
10058 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
10059 Make edge redirection.
10061 2018-06-29 David Malcolm <dmalcolm@redhat.com>
10063 * dumpfile.c (dump_loc): Add indentation based on scope depth.
10064 (dump_scope_depth): New variable.
10065 (get_dump_scope_depth): New function.
10066 (dump_begin_scope): New function.
10067 (dump_end_scope): New function.
10068 * dumpfile.h (get_dump_scope_depth): New declaration.
10069 (dump_begin_scope): New declaration.
10070 (dump_end_scope): New declaration.
10071 (class auto_dump_scope): New class.
10072 (AUTO_DUMP_SCOPE): New macro.
10073 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
10076 2018-06-29 Richard Biener <rguenther@suse.de>
10078 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
10079 compute_all_dependences succeeds.
10080 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
10081 exceed --param loop-max-datarefs-for-datadeps.
10083 2018-06-29 Jakub Jelinek <jakub@redhat.com>
10085 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
10087 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
10090 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
10091 alternative 0 in preferred_for_speed attribute.
10093 2018-06-28 Paul Koning <ni1d@arrl.net>
10095 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
10096 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
10097 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
10098 * config/pdp11/pdp11.md: Correct "length" attribute calculation
10099 for shift insn patterns.
10101 2018-06-28 David Malcolm <dmalcolm@redhat.com>
10103 * cgraph.c (cgraph_node::get_body): Replace assignments to
10104 "dump_file" with calls to set_dump_file.
10105 * dumpfile.c (alt_dump_file): Make static, and group with...
10106 (alt_flags): ...this definition.
10107 (dumps_are_enabled): New variable.
10108 (refresh_dumps_are_enabled): New function.
10109 (set_dump_file): New function.
10110 (set_alt_dump_file): New function.
10111 (gcc::dump_manager::dump_start): Replace assignments to
10112 "dump_file" and "alt_dump_file" with calls to set_dump_file and
10114 (gcc::dump_manager::dump_finish): Likewise.
10115 * dumpfile.h (alt_dump_file): Delete decl.
10116 (dumps_are_enabled): New variable decl.
10117 (set_dump_file): New function decl.
10118 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
10120 * tree-nested.c (lower_nested_functions): Replace assignments to
10121 "dump_file" with calls to set_dump_file.
10123 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
10125 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
10126 goto_locus of each outgoing edge of each basic block.
10128 2018-06-28 Richard Biener <rguenther@suse.de>
10130 * dwarf2out.c (decl_scope_table): Remove.
10131 (push_decl_scope): Likewise.
10132 (pop_decl_scope): Likewise.
10133 (gen_type_die_for_member): Do not call push/pop_decl_scope.
10134 (gen_struct_or_union_type_die): Likewise.
10135 (gen_tagged_type_die): Likewise.
10136 (dwarf2out_init): Do not initialize decl_scope_table.
10137 (dwarf2out_c_finalize): Do not free it.
10139 2018-06-28 Richard Biener <rguenther@suse.de>
10141 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
10142 deciding whether to not re-use a DIE.
10144 2018-06-28 Richard Biener <rguenther@suse.de>
10146 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
10147 DW_AT_abstract_origin attribute.
10149 2018-06-28 Martin Liska <mliska@suse.cz>
10151 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
10152 Use newly introduced constants.
10153 * tree-switch-conversion.h (struct jump_table_cluster):
10154 Define max_ratio_for_size and max_ratio_for_speed.
10156 2018-06-28 Martin Liska <mliska@suse.cz>
10158 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
10159 Add new checking assert to catch invalid state.
10160 (jump_table_cluster::can_be_handled): Handle single case
10162 (jump_table_cluster::is_beneficial): Bail out for such case.
10163 (bit_test_cluster::find_bit_tests):
10164 Add new checking assert to catch invalid state.
10165 (bit_test_cluster::can_be_handled): Handle single case
10167 (bit_test_cluster::is_beneficial): Bail out for such case.
10168 (switch_decision_tree::analyze_switch_statement):
10171 2018-06-28 Martin Liska <mliska@suse.cz>
10173 * common.opt: Introduce -completion option.
10174 * gcc.c (driver_handle_option): Handle it.
10175 (driver::main): Print completions if completion
10177 * opt-suggestions.c (option_proposer::get_completions):
10179 (option_proposer::suggest_completion): Likewise.
10180 (option_proposer::find_param_completions): Likewise.
10181 (verify_autocompletions): Likewise.
10182 (test_completion_valid_options): Likewise.
10183 (test_completion_valid_params): Likewise.
10184 (in_completion_p): Likewise.
10185 (empty_completion_p): Likewise.
10186 (test_completion_partial_match): Likewise.
10187 (test_completion_garbage): Likewise.
10188 (opt_proposer_c_tests): Likewise.
10189 * opt-suggestions.h: Declare new functions.
10190 * opts.c (common_handle_option): Handle OPT__completion_.
10191 * selftest-run-tests.c (selftest::run_tests): Add
10192 opt_proposer_c_tests.
10193 * selftest.c (assert_str_startswith): New.
10194 * selftest.h (assert_str_startswith): Likewise.
10195 (opt_proposer_c_tests): New.
10196 (ASSERT_STR_STARTSWITH): Likewise.
10198 2018-06-28 Martin Liska <mliska@suse.cz>
10200 * Makefile.in: Add opt-suggestions.o.
10201 * gcc-main.c: Include opt-suggestions.h.
10202 * gcc.c (driver::driver): Likewise.
10203 (driver::~driver): Remove m_option_suggestions.
10204 (driver::build_option_suggestions): Moved to option_proposer.
10205 (driver::suggest_option): Likewise.
10206 (driver::handle_unrecognized_options): Use option_proposer.
10207 * gcc.h (class driver): Add new memver m_option_proposer.
10208 * opt-suggestions.c: New file.
10209 * opt-suggestions.h: New file.
10211 2018-06-28 Martin Liska <mliska@suse.cz>
10213 * vec.h (class auto_string_vec): New (moved from auto_argvec).
10214 (auto_string_vec::~auto_string_vec): Likewise.
10216 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
10218 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
10219 prevent_decl_creation_for_types fields up and add reset_location field.
10220 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
10221 statement if id->reset_location is true.
10222 (copy_edges_for_bb): Do not set goto_locus on the new edges if
10223 id->reset_location is true.
10224 (copy_phis_for_bb): Force input_location on the arguments if
10225 id->reset_location is true.
10226 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
10227 is set on the function to be inlined.
10229 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
10231 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
10233 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
10235 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
10236 registers for Pmode.
10237 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
10238 hard registers for the clobbered pseudo.
10240 2018-06-27 Paul Koning <ni1d@arrl.net>
10242 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
10243 mutually exclusive options.
10244 * config/pdp11/constraints.md (h): New constraint.
10245 (O): Update definition to match shift code generation.
10246 (D): New constraint.
10247 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
10249 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
10251 (output_jump): Change arguments.
10252 (pdp11_fixed_cc_regs): New function.
10253 (pdp11_cc_mode): Ditto.
10254 (pdp11_expand_shift): Ditto.
10255 (pdp11_assemble_shift): Ditto.
10256 (pdp11_small_shift): Ditto.
10257 (pdp11_branch_cost): Remove.
10258 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
10260 (pdp11_register_move_cost): Update for CC registers.
10261 (pdp11_rtx_costs): Add case for LSHIFTRT.
10262 (pdp11_output_jump): Add CCNZ mode conditional branches.
10263 (notice_update_cc_on_set): Remove.
10264 (pdp11_cc_mode): New function.
10265 (simple_memory_operand): Correct pre/post decrement case.
10266 (no_side_effect_operand): New function.
10267 (pdp11_regno_reg_class): Add CC_REGS class.
10268 (pdp11_fixed_cc_regs): New function.
10269 (pdp11_small_shift): New function.
10270 (pdp11_expand_shift): New function to expand shift insns.
10271 (pdp11_assemble_shift): New function to output shifts.
10272 (pdp11_branch_cost): Remove.
10273 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
10274 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
10275 (WCHAR_TYPE): Ditto.
10276 (PTRDIFF_TYPE): Ditto.
10277 (ADJUST_INSN_LENGTH): New macro.
10278 (FIXED_REGISTERS): Add CC registers.
10279 (CALL_USED_REGISTERS): Ditto.
10280 (reg_class): Ditto.
10281 (REG_CLASS_NAMES): Ditto.
10282 (REG_CLASS_CONTENTS): Ditto.
10283 (SELECT_CC_MODE): Use new function.
10284 (TARGET_FLAGS_REGNUM): New macro.
10285 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
10286 (cc0_reg_rtx): Remove.
10287 (CC_STATUS_MDEP): Remove.
10288 (CC_STATUS_MDEFP_INIT): Remove.
10289 (CC_IN_FPU): Remove.
10290 (NOTICE_UPDATE_CC): Remove.
10291 (REGISTER_NAMES): Add CC registers.
10292 (BRANCH_COST): Change to constant 1.
10293 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
10295 * config/pdp11/pdp11.opt (mbcopy): Remove.
10296 (mbcopy-builtin): Remove.
10297 (mbranch-cheap): Remove.
10298 (mbranch-expensive): Remove.
10299 * config/pdp11/predicates.md (expand_shift_operand): Update to
10300 match shift code generation.
10301 (ccnz_operator): New predicate.
10302 * doc/invoke.texi (PDP-11 Options): Remove deleted options
10303 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
10304 Remove non-existent option -mabshi, -mno-abshi. Document mutually
10306 * doc/md.texi (PDP-11): Document new D and h constraints. Update
10307 description of O constraint.
10309 2018-06-27 Jeff Law <law@redhat.com>
10310 Austin Law <austinklaw@gmail.com>
10312 * config/v850/v850.md (addsi3_set_flags): New pattern.
10313 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
10314 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
10315 (zero_extendhisi2_v850_set_flags): Likewise.
10316 (zero_extendqisi2_v850_set_flags): Likewise.
10317 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
10318 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
10319 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
10321 * config/v850/v850-protos.h (notice_update_cc): Remove.
10322 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
10323 (v850_print_operand): Handle 'D' and "d".
10324 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
10325 Add handling of arithmetic/logical operations compared against zero.
10326 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
10327 Do not look at v850_compare_op, instead get mode from last argument.
10328 (v850_gen_compare): Remove
10329 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
10330 after reload for prologue insns.
10331 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
10333 (construct_save_jarl): Likewise.
10334 (TARGET_FLAGS_REGNUM): Define.
10335 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
10336 (NOTICE_UPDATE_CC): Remove.
10337 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
10338 than cc0. Conditionalize on reload_completed.
10339 (cmpsi_insn, setfcc_insn): Likewise.
10340 (tst1 splitter): Turn into define_and_split which sets the flags
10342 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
10343 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
10344 (cstoresf4, cstoredf4): Clobber the flags.
10345 (cmpsi, cmpsf, cmpdf): Remove expanders.
10346 (setf_insn): Remove pattern.
10347 (addsi3): Turn into define_and_split which clobbers the flags after
10348 reload and a suitable pattern (addsi3_clobber_flags) for use after
10350 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
10351 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
10352 (ashrsi3, ashrsi3_v850e2): Likewise.
10353 (bins): Clobber the flags.
10354 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
10355 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
10356 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
10357 (call_value_internal_short, call_value_internal_long): Likewise.
10358 (callt_save_interrupt, callt_return_interrupt): Likewise.
10359 (save_interrupt, return_interrupt): Likewise.
10360 (callt_save_all_interrupt, save_all_interrupt): Likewise.
10361 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
10362 (restore_all_interrupt, _restore_all_interrupt): Likewise.
10363 (All FP comparisons): Only allow after reload has completed.
10365 (divh, divhu): Tweak output template.
10366 (branch_z_normal, branch_z_invert): Remove
10367 (branch_nz_normal, branch_nz_invert): Likewise.
10368 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
10370 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
10371 * config/v850/v850.c (notice_update_cc): Remove.
10372 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
10373 (CC_NO_CARRY): Likewise.
10374 (NOTICE_UPDATE_CC): Define to nothing.
10375 * config/v850/v850.md: Remove block comment on cc0 handling
10376 Remove "cc" attribute from all patterns. Remove cc_status handling
10377 from all patterns. Minor formatting fixes.
10379 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10381 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
10382 (cortex-a76.cortex-a55): Likewise.
10383 * config/aarch64/aarch64-tune.md: Regenerate.
10384 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
10385 cortex-a76.cortex-a55.
10387 2018-06-27 Jeff Law <law@redhat.com>
10389 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
10390 (MULTILIB_DIRNAMES): Similarly.
10392 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
10394 * gimple.h (gimple_return_retbnd): Delete.
10395 (gimple_return_set_retbnd): Likewise.
10396 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
10397 gimple_return_set_retbnd.
10398 * gimple-pretty-print.c (dump_gimple_return): Remove call to
10399 gimple_return_retbnd and adjust.
10400 * tree-inline.h (struct copy_body_data): Remove retbnd field.
10401 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
10402 Explicitly return NULL in a couple more cases. Move assertion
10403 on debug statements and remove unreachable code.
10404 (reset_debug_binding): Do not test id->retbnd.
10405 (expand_call_inline): Do not set it.
10407 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
10409 * configure.ac: Add --disable-gcov option.
10410 * configure: Regenerate.
10411 * Makefile.in: Honour @enable_gcov@.
10412 * doc/install.texi: Document --disable-gcov.
10414 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10416 * config/arm/arm-cpus.in (cortex-a76): New entry.
10417 (cortex-a76.cortex-a55): Likewise.
10418 * config/arm/arm-tables.opt: Regenerate.
10419 * config/arm/arm-tune.md: Likewise.
10420 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
10421 * doc/invoke.texi (ARM Options): Document cortex-a76 and
10422 cortex-a76.cortex-a55.
10424 2018-06-27 Tamar Christina <tamar.christina@arm.com>
10427 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
10429 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
10431 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
10433 (EPILOGUE_USES): Likewise.
10435 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
10437 * tree-inline.c (remap_location): New function extracted from...
10438 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
10439 (copy_phis_for_bb): ...here. Call remap_location.
10440 (copy_cfg_body): Adjust call to copy_edges_for_bb.
10442 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
10444 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
10445 unaligned vsx for 16B memset.
10447 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
10450 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
10451 ieee128_float_type_node to long_double_type_node unless
10452 TARGET_LONG_DOUBLE_128 is set.
10454 2018-06-26 David Malcolm <dmalcolm@redhat.com>
10456 * cfgloop.c (get_loop_location): Convert return type from
10457 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
10458 by implicit construction from rtx_insn *, and using
10459 dump_user_location_t::from_function_decl for the fallback case.
10460 * cfgloop.h (get_loop_location): Convert return type from
10461 location_t to dump_user_location_t.
10462 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
10463 dump_printf_loc to pass in a dump_location_t rather than a
10464 location_t, via the gimple stmt.
10465 * coverage.c (get_coverage_counts): Update calls to
10466 dump_printf_loc to pass in dump_location_t rather than a
10468 * doc/optinfo.texi (Dump types): Convert example of
10469 dump_printf_loc from taking "locus" to taking "insn". Update
10470 description of the "_loc" calls to cover dump_location_t.
10471 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
10473 (dump_user_location_t::dump_user_location_t): New constructors,
10474 from gimple *stmt and rtx_insn *.
10475 (dump_user_location_t::from_function_decl): New function.
10476 (dump_loc): Make static.
10477 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
10478 const dump_location_t &.
10479 (dump_generic_expr_loc): Delete.
10480 (dump_printf_loc): Convert param "loc" from location_t to
10481 const dump_location_t &.
10482 (selftest::test_impl_location): New function.
10483 (selftest::dumpfile_c_tests): New function.
10484 * dumpfile.h: Include "profile-count.h".
10485 (class dump_user_location_t): New class.
10486 (struct dump_impl_location_t): New struct.
10487 (class dump_location_t): New class.
10488 (dump_printf_loc): Convert 2nd param from source_location to
10489 const dump_location_t &.
10490 (dump_generic_expr_loc): Delete.
10491 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
10492 const dump_location_t &.
10493 * gimple-fold.c (fold_gimple_assign): Update call to
10494 dump_printf_loc to pass in a dump_location_t rather than a
10495 location_t, via the gimple stmt.
10496 (gimple_fold_call): Likewise.
10497 * gimple-loop-interchange.cc
10498 (loop_cand::analyze_iloop_reduction_var): Update for change to
10499 check_reduction_path.
10500 (tree_loop_interchange::interchange): Update for change to
10501 find_loop_location.
10502 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
10503 change in return-type of find_loop_location.
10504 (graphite_regenerate_ast_isl): Likewise.
10505 * graphite-optimize-isl.c (optimize_isl): Likewise.
10506 * graphite.c (graphite_transform_loops): Likewise.
10507 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
10508 pass in a dump_location_t rather than a location_t, via the
10510 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
10511 * ipa.c (walk_polymorphic_call_targets): Likewise.
10512 * loop-unroll.c (report_unroll): Convert "locus" param from
10513 location_t to dump_location_t.
10514 (decide_unrolling): Update for change to get_loop_location's
10516 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
10517 location_t to dump_user_location_t.
10518 (grid_find_single_omp_among_assignments_1): Updates calls to
10519 dump_printf_loc to pass in a dump_location_t rather than a
10520 location_t, via the gimple stmt.
10521 (grid_parallel_clauses_gridifiable): Convert "tloc" from
10522 location_t to dump_location_t. Updates calls to dump_printf_loc
10523 to pass in a dump_location_t rather than a location_t, via the
10525 (grid_inner_loop_gridifiable_p): Likewise.
10526 (grid_dist_follows_simple_pattern): Likewise.
10527 (grid_gfor_follows_tiling_pattern): Likewise.
10528 (grid_target_follows_gridifiable_pattern): Likewise.
10529 (grid_attempt_target_gridification): Convert initialization
10530 of local "grid" from memset to zero-initialization; FIXME: does
10531 this require C++11? Update call to dump_printf_loc to pass in a
10532 optinfo_location rather than a location_t, via the gimple stmt.
10533 * profile.c (read_profile_edge_counts): Updates call to
10534 dump_printf_loc to pass in a dump_location_t rather than a
10536 (compute_branch_probabilities): Likewise.
10537 * selftest-run-tests.c (selftest::run_tests): Call
10539 * selftest.h (dumpfile_c_tests): New decl.
10540 * tree-loop-distribution.c (pass_loop_distribution::execute):
10541 Update for change in return type of find_loop_location.
10542 * tree-parloops.c (parallelize_loops): Likewise.
10543 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
10544 "locus" from location_t to dump_user_location_t.
10545 (canonicalize_loop_induction_variables): Likewise.
10546 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
10547 for change in return type of find_loop_location.
10548 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
10549 to dump_printf_loc to pass in a dump_location_t rather than a
10550 location_t, via the stmt.
10551 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
10553 * tree-vect-loop-manip.c (find_loop_location): Convert return
10554 type from source_location to dump_user_location_t.
10555 (vect_do_peeling): Update for above change.
10556 (vect_loop_versioning): Update for change in type of
10558 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
10559 from location_t to dump_user_location_t.
10560 (vect_estimate_min_profitable_iters): Update for change in type
10562 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
10563 location_t to dump_location_t.
10564 (vect_slp_bb): Update for change in type of vect_location.
10565 * tree-vectorizer.c (vect_location): Convert from source_location
10566 to dump_user_location_t.
10567 (try_vectorize_loop_1): Update for change in vect_location's type.
10568 (vectorize_loops): Likewise.
10569 (increase_alignment): Likewise.
10570 * tree-vectorizer.h (vect_location): Convert from source_location
10571 to dump_user_location_t.
10572 (find_loop_location): Convert return type from source_location to
10573 dump_user_location_t.
10574 (check_reduction_path): Convert 1st param from location_t to
10575 dump_user_location_t.
10576 * value-prof.c (check_counter): Update call to dump_printf_loc to
10577 pass in a dump_user_location_t rather than a location_t; update
10578 call to error_at for change in type of "locus".
10579 (check_ic_target): Update call to dump_printf_loc to
10580 pass in a dump_user_location_t rather than a location_t, via the
10583 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
10585 * config/s390/s390.h (enum processor_flags): Do not use
10586 default tune parameter when -march was specified.
10588 2018-06-26 Jakub Jelinek <jakub@redhat.com>
10591 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
10592 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
10595 2018-06-26 Richard Biener <rguenther@suse.de>
10597 PR tree-optimization/86287
10599 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
10600 (vect_analyze_loop): Initialize n_stmts.
10602 2018-06-26 Richard Biener <rguenther@suse.de>
10604 PR middle-end/86271
10605 * fold-const.c (fold_convertible_p): Pointer extension
10608 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
10611 * dwarf2out.c (loc_list_has_views): Adjust comments.
10612 (dw_loc_list): Split single cross-partition range with
10615 2018-06-25 Jeff Law <law@redhat.com>
10617 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
10618 on -mbig-switch by default.
10620 * config/v850/predicates.md (const_float_1_operand): Fix match_code
10622 (const_float_0_operand): Remove unused predicate.
10623 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
10624 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
10625 (recipsf2): New expander. Original pattern now called
10627 (recipdf2, recipdf2_insn): Similarly.
10628 (rsqrtsf2, rsqrtsf2_insn): Similarly
10629 (rsqrtdf2, rsqrtdf2_insn): Similarly
10631 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
10633 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
10634 Simplify logic for FreeBSD (twice).
10636 2018-06-25 Martin Sebor <msebor@redhat.com>
10638 PR tree-optimization/86204
10639 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
10640 a strnlen result if it's less than the length of the string.
10642 2018-06-25 Martin Sebor <msebor@redhat.com>
10644 PR tree-optimization/85700
10645 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
10646 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
10647 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
10649 2018-06-25 Martin Sebor <msebor@redhat.com>
10651 * doc/extend.texi (Zero-length arrays): Update and clarify.
10653 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
10655 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
10656 added IEEE/IBM long double multilib support on PowerPC little
10657 endian Linux systems.
10658 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
10659 (MULTILIB_DEFAULTS): Likewise.
10660 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10662 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
10663 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
10664 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
10666 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
10668 PR middle-end/86311
10669 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
10670 (REORDER_45): Likewise.
10672 2018-06-25 Jeff Law <law@redhat.com>
10674 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
10675 dividend to 32 bits. Adjust length.
10676 (udivmodhi4): Cleanup output template. Fix length.
10678 2018-06-25 Carl Love <cel@us.ibm.com>
10680 * config/rs6000/vsx.md: Change word selector to prefered location.
10682 2018-06-25 Richard Biener <rguenther@suse.de>
10684 PR tree-optimization/86304
10685 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
10686 epilogue-if-converted loops as well.
10688 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
10690 * lto-section-out.c (lto_begin_section): Do not print section
10691 name for noaddr and unnumbered dumps.
10693 2018-06-25 Richard Biener <rguenther@suse.de>
10695 * tree-vectorizer.h (struct vec_info_shared): New structure
10696 with parts split out from struct vec_info and loop_nest from
10697 struct _loop_vec_info.
10698 (struct vec_info): Adjust accordingly.
10699 (struct _loop_vec_info): Likewise.
10700 (LOOP_VINFO_LOOP_NEST): Adjust.
10701 (LOOP_VINFO_DATAREFS): Likewise.
10702 (LOOP_VINFO_DDRS): Likewise.
10703 (struct _bb_vec_info): Likewise.
10704 (BB_VINFO_DATAREFS): Likewise.
10705 (BB_VINFO_DDRS): Likewise.
10706 (struct _stmt_vec_info): Add dr_aux member.
10707 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
10708 (DR_MISALIGNMENT_UNINITIALIZED): New.
10709 (set_dr_misalignment): Adjust.
10710 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
10711 (vect_analyze_loop): Adjust prototype.
10712 (vect_analyze_loop_form): Likewise.
10713 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
10714 Compute dependences lazily.
10715 (vect_record_base_alignments): Use shared datarefs/ddrs.
10716 (vect_verify_datarefs_alignment): Likewise.
10717 (vect_analyze_data_refs_alignment): Likewise.
10718 (vect_analyze_data_ref_accesses): Likewise.
10719 (vect_analyze_data_refs): Likewise.
10720 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
10721 constructor parameter for shared part.
10722 (vect_analyze_loop_form): Pass in shared part and adjust.
10723 (vect_analyze_loop_2): Pass in storage for the number of
10724 stmts. Move loop nest finding to the caller. Compute
10726 (vect_analyze_loop): Pass in shared part.
10727 (vect_transform_loop): Verify shared datarefs are unchanged.
10728 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
10729 constructor parameter for shared part.
10730 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
10731 (vect_slp_bb): Verify shared datarefs are unchanged before
10733 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
10735 (new_stmt_vec_info): Initialize DR_AUX misalignment to
10736 DR_MISALIGNMENT_UNINITIALIZED.
10737 * tree-vectorizer.c (vec_info::vec_info): Add constructor
10738 parameter for shared part.
10739 (vec_info::~vec_info): Adjust.
10740 (vec_info_shared::vec_info_shared): New.
10741 (vec_info_shared::~vec_info_shared): Likewise.
10742 (vec_info_shared::save_datarefs): Likewise.
10743 (vec_info_shared::check_datarefs): Likewise.
10744 (try_vectorize_loop_1): Construct shared part live for analyses
10745 of a single loop for multiple vector sizes.
10746 * tree-parloops.c (gather_scalar_reductions): Adjust.
10748 2018-06-25 Richard Biener <rguenther@suse.de>
10750 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
10751 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
10752 (vect_analyze_data_refs): Remove similar code from here and
10753 simplify accordingly.
10755 2018-06-25 Richard Biener <rguenther@suse.de>
10757 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
10758 for reverse storage order accesses rather than asserting
10759 they cannot happen here.
10761 2018-06-25 Tom de Vries <tdevries@suse.de>
10764 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
10765 Use data16 instead of .byte for insn prefix.
10767 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
10770 * parser.c (make_char_string_pack): Pass this literal chars
10771 through cpp_interpret_string.
10772 (cp_parser_userdef_numeric_literal): Check the result of
10773 make_char_string_pack.
10775 2018-06-24 Maya Rashish <coypu@sdf.org>
10777 * ginclude/stddef.h: Simplify conditions around avoiding
10778 re-definition of __size_t.
10780 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
10782 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
10783 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
10785 2018-06-22 Maya Rashish <coypu@sdf.org>
10787 * doc/invoke.texi (mno-fancy-math-387): Update for changes
10788 made to OpenBSD and NetBSD through the years.
10790 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
10792 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
10793 behavior of vec_pack (vector double, vector double) to match
10794 behavior of vec_float2 (vector double, vector double).
10796 2018-06-22 Olivier Hainque <hainque@adacore.com>
10798 * gimplify.c (gimplify_function_tree): Prevent creation
10799 of a trampoline for the address of the current function
10800 passed to entry/exit instrumentation hooks.
10802 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
10805 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
10808 2018-06-22 Martin Liska <mliska@suse.cz>
10810 PR tree-optimization/86263
10811 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
10812 Bail out if is_enabled is false.
10813 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
10815 (jump_table_cluster::is_enabled): New function.
10817 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
10819 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
10820 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
10821 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
10822 (lto_input_ts_binfo_tree_pointers): Likewise.
10823 * tree-streamer-out.c (streamer_write_tree_bitfields,
10824 write_ts_binfo_tree_pointers): Likewise.
10825 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
10827 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
10829 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
10831 2018-06-22 Martin Liska <mliska@suse.cz>
10833 * symbol-summary.h (get): Make it pure and inline move
10834 functionality from ::get function.
10835 (get): Remove and inline into ::get and ::get_create.
10836 (get_create): Move code from ::get function.
10838 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10841 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
10842 -x assembler-with-cpp.
10844 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10846 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
10847 _FILE_OFFSET_BITS=64 for C++.
10849 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
10851 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
10852 conversion insn that shows up when pr85657-3.c is compiled using
10853 IEEE 128-bit long double.
10854 (neg<mode>2_internal): Use the correct mode to check whether the
10855 mode is IBM extended.
10856 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
10857 multiply and divide external functions from being created more
10860 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
10862 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
10864 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
10865 the edge can be forwarded.
10866 (cfg_layout_merge_blocks): Likewise.
10868 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
10870 * except.c (finish_eh_generation): Commit edge insertions only after
10871 the EH edges have been redirected from post-landing to landing pads.
10873 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
10875 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
10876 create_tmp_var_for to create the FRAME decl.
10877 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
10879 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
10881 * tree-inline.c (copy_edges_for_bb): Minor tweak.
10882 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
10883 debug statement when resetting its value.
10884 (expand_call_inline): Copy the locus of the call onto the assignment
10885 of the return value, if any. Use local variable in more cases.
10887 2018-06-21 Martin Liska <mliska@suse.cz>
10889 * ipa-pure-const.c (propagate_nothrow): Use
10890 funct_state_summaries->get.
10891 (dump_malloc_lattice): Likewise.
10892 (propagate_malloc): Likewise.
10894 2018-06-21 Richard Biener <rguenther@suse.de>
10896 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
10897 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
10898 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
10900 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
10901 BLOCK_ABSTRACT_ORIGIN unconditionally.
10903 2018-06-21 David Malcolm <dmalcolm@redhat.com>
10905 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
10907 * ipa-reference.c (ipa_reference_c_finalize): Delete
10908 ipa_ref_opt_sum_summaries and set it to NULL.
10910 2018-06-21 Tom de Vries <tdevries@suse.de>
10912 PR tree-optimization/85859
10913 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
10914 test with comment from bb_no_side_effects_p.
10916 2018-06-21 Richard Biener <rguenther@suse.de>
10918 PR tree-optimization/86232
10919 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
10920 max for constant niter.
10922 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
10924 * config/aarch64/aarch64-simd.md
10925 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
10927 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
10929 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
10930 Make opernads of the unspec commutative.
10932 2018-06-21 Richard Biener <rguenther@suse.de>
10934 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
10935 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
10936 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
10937 (vect_analyze_data_ref_dependence): Re-order checks to deal with
10939 (vect_record_base_alignments): Do not record base alignment
10940 for gathers or scatters.
10941 (vect_compute_data_ref_alignment): Drop return value that is always
10942 true. Bail out early for gathers or scatters.
10943 (vect_enhance_data_refs_alignment): Bail out early for gathers
10945 (vect_find_same_alignment_drs): Likewise.
10946 (vect_analyze_data_refs_alignment): Remove dead code.
10947 (vect_slp_analyze_and_verify_node_alignment): Likewise.
10948 (vect_analyze_data_refs): For possible gathers or scatters do
10949 not create an alternate DR, just check their possible validity
10950 and mark them. Adjust DECL_NONALIASED handling to not rely
10951 on DR_BASE_ADDRESS.
10952 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
10953 update inits of gathers or scatters.
10954 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
10955 Also copy gather/scatter flag to pattern vinfo.
10957 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
10959 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
10960 behavior of vec_packsu (vector unsigned long long, vector unsigned
10961 long long) to match behavior of vec_packs with same signature.
10963 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
10964 Thomas Schwinge <thomas@codesourcery.com>
10965 Cesar Philippidis <cesar@codesourcery.com>
10967 * gimplify.c (gimplify_scan_omp_clauses): Add support for
10968 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
10969 (gimplify_adjust_omp_clauses): Likewise.
10970 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
10971 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
10972 (gimplify_omp_target_update): Update handling of acc update and
10974 * omp-low.c (install_var_field): Remove unused parameter
10975 base_pointers_restrict.
10976 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
10977 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
10979 (omp_target_base_pointers_restrict_p): Delete.
10980 (scan_omp_target): Update call to scan_sharing_clauses.
10981 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
10983 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
10984 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
10985 (convert_local_omp_clauses): Likewise.
10986 * tree-pretty-print.c (dump_omp_clause): Likewise.
10987 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
10989 (omp_clause_code_name): Likewise.
10991 2018-06-20 Jakub Jelinek <jakub@redhat.com>
10994 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
10997 PR tree-optimization/86231
10998 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
10999 anti-range don't overwrite *vr0min before using it to compute *vr0max.
11001 2018-06-20 Tom de Vries <tdevries@suse.de>
11003 PR tree-optimization/86097
11004 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
11005 iv type if signedness of iv type is not the same as that of *nit.
11007 2018-06-20 Jakub Jelinek <jakub@redhat.com>
11009 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
11010 EDGE_EH edges, verify they are all EDGE_EH.
11012 2018-06-20 Maya Rashish <coypu@sdf.org>
11014 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
11016 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11018 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
11019 * config/aarch64/aarch64.c (xgene1_tunings): Add
11020 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
11021 (aarch64_mode_valid_for_sched_fusion_p):
11022 Allow 16-byte modes.
11023 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
11024 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
11026 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
11028 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
11029 * config/aarch64/iterators.md (VQ2): New mode iterator.
11031 2018-06-20 Martin Liska <mliska@suse.cz>
11033 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
11034 Change default ratio from 10 to 8.
11036 2018-06-20 Martin Liska <mliska@suse.cz>
11038 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
11040 (bit_test_cluster::find_bit_tests): Likewise.
11041 (switch_decision_tree::analyze_switch_statement): Find clusters.
11042 * tree-switch-conversion.h (struct jump_table_cluster): Document
11045 2018-06-20 Martin Liska <mliska@suse.cz>
11047 * tree-switch-conversion.c (switch_conversion::collect):
11048 Record m_uniq property.
11049 (switch_conversion::expand): Bail out for special conditions.
11050 (group_cluster::~group_cluster): New.
11051 (group_cluster::group_cluster): Likewise.
11052 (group_cluster::dump): Likewise.
11053 (jump_table_cluster::emit): New.
11054 (switch_decision_tree::fix_phi_operands_for_edges): New.
11055 (struct case_node): Remove struct.
11056 (jump_table_cluster::can_be_handled): New.
11057 (case_values_threshold): Moved to header.
11058 (reset_out_edges_aux): Likewise.
11059 (jump_table_cluster::is_beneficial): New.
11060 (bit_test_cluster::can_be_handled): Likewise.
11061 (add_case_node): Remove.
11062 (bit_test_cluster::is_beneficial): New.
11063 (case_bit_test::cmp): New.
11064 (bit_test_cluster::emit): New.
11065 (expand_switch_as_decision_tree_p): Remove.
11066 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
11067 (fix_phi_operands_for_edge): Likewise.
11068 (switch_decision_tree::analyze_switch_statement): New.
11069 (compute_cases_per_edge): Move ...
11070 (switch_decision_tree::compute_cases_per_edge): ... here.
11071 (try_switch_expansion): Likewise.
11072 (switch_decision_tree::try_switch_expansion): Likewise.
11073 (record_phi_operand_mapping): Likewise.
11074 (switch_decision_tree::record_phi_operand_mapping): Likewise.
11075 (emit_case_decision_tree): Likewise.
11076 (switch_decision_tree::emit): Likewise.
11077 (balance_case_nodes): Likewise.
11078 (switch_decision_tree::balance_case_nodes): Likewise.
11079 (dump_case_nodes): Likewise.
11080 (switch_decision_tree::dump_case_nodes): Likewise.
11081 (emit_jump): Likewise.
11082 (switch_decision_tree::emit_jump): Likewise.
11083 (emit_cmp_and_jump_insns): Likewise.
11084 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
11085 (emit_case_nodes): Likewise.
11086 (switch_decision_tree::emit_case_nodes): Likewise.
11087 (conditional_probability): Remove.
11088 * tree-switch-conversion.h (enum cluster_type): New.
11090 (struct cluster): Likewise.
11091 (cluster::cluster): Likewise.
11092 (struct simple_cluster): Likewise.
11093 (simple_cluster::simple_cluster): Likewise.
11094 (struct group_cluster): Likewise.
11095 (struct jump_table_cluster): Likewise.
11096 (struct bit_test_cluster): Likewise.
11097 (struct min_cluster_item): Likewise.
11098 (struct case_tree_node): Likewise.
11099 (case_tree_node::case_tree_node): Likewise.
11100 (jump_table_cluster::case_values_threshold): Likewise.
11101 (struct case_bit_test): Likewise.
11102 (struct switch_decision_tree): Likewise.
11103 (struct switch_conversion): Likewise.
11104 (switch_decision_tree::reset_out_edges_aux): Likewise.
11106 2018-06-20 Martin Liska <mliska@suse.cz>
11108 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
11109 (hoist_edge_and_branch_if_true): Likewise.
11110 (expand_switch_using_bit_tests_p): Likewise.
11111 (struct case_bit_test): Likewise.
11112 (case_bit_test_cmp): Likewise.
11113 (emit_case_bit_tests): Likewise.
11114 (switch_conversion::switch_conversion): New class.
11115 (struct switch_conv_info): Remove old struct.
11116 (collect_switch_conv_info): More to ...
11117 (switch_conversion::collect): ... this.
11118 (check_range): Likewise.
11119 (switch_conversion::check_range): Likewise.
11120 (check_all_empty_except_final): Likewise.
11121 (switch_conversion::check_all_empty_except_final): Likewise.
11122 (check_final_bb): Likewise.
11123 (switch_conversion::check_final_bb): Likewise.
11124 (create_temp_arrays): Likewise.
11125 (switch_conversion::create_temp_arrays): Likewise.
11126 (free_temp_arrays): Likewise.
11127 (gather_default_values): Likewise.
11128 (switch_conversion::gather_default_values): Likewise.
11129 (build_constructors): Likewise.
11130 (switch_conversion::build_constructors): Likewise.
11131 (constructor_contains_same_values_p): Likewise.
11132 (switch_conversion::contains_same_values_p): Likewise.
11133 (array_value_type): Likewise.
11134 (switch_conversion::array_value_type): Likewise.
11135 (build_one_array): Likewise.
11136 (switch_conversion::build_one_array): Likewise.
11137 (build_arrays): Likewise.
11138 (switch_conversion::build_arrays): Likewise.
11139 (gen_def_assigns): Likewise.
11140 (switch_conversion::gen_def_assigns): Likewise.
11141 (prune_bbs): Likewise.
11142 (switch_conversion::prune_bbs): Likewise.
11143 (fix_phi_nodes): Likewise.
11144 (switch_conversion::fix_phi_nodes): Likewise.
11145 (gen_inbound_check): Likewise.
11146 (switch_conversion::gen_inbound_check): Likewise.
11147 (process_switch): Use the newly created class.
11148 (switch_conversion::expand): New.
11149 (switch_conversion::~switch_conversion): New.
11150 * tree-switch-conversion.h: New file.
11152 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
11154 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
11155 tree-vect-patterns.c.
11156 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
11157 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
11158 (vect_recog_sad_pattern): Likewise.
11159 (vect_recog_widen_sum_pattern): Likewise.
11160 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
11161 (vect_recog_widen_shift_pattern): Remove the type_in argument.
11162 (vect_recog_rotate_pattern): Likewise.
11163 (vect_recog_mult_pattern): Likewise.
11164 (vect_recog_vector_vector_shift_pattern): Likewise.
11165 (vect_recog_divmod_pattern): Likewise.
11166 (vect_recog_mixed_size_cond_pattern): Likewise.
11167 (vect_recog_bool_pattern): Likewise.
11168 (vect_recog_mask_conversion_pattern): Likewise.
11169 (vect_try_gather_scatter_pattern): Likewise.
11170 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
11171 (vect_recog_over_widening_pattern): Likewise.
11172 (vect_recog_gather_scatter_pattern): Likewise.
11173 (vect_recog_func_ptr): Move from tree-vectorizer.h
11174 (vect_vect_recog_func_ptrs): Move further down the file.
11175 (vect_recog_func): Likewise. Remove the third argument.
11176 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
11177 (vect_pattern_recog_1): Expect the pattern function to do any
11178 necessary target tests. Also expect it to provide a vector type.
11179 Remove the type_in handling.
11181 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
11183 * tree-vect-patterns.c (vect_pattern_detected): New function.
11184 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
11185 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
11186 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
11187 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
11188 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
11189 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
11190 (vect_recog_mask_conversion_pattern)
11191 (vect_try_gather_scatter_pattern): Likewise.
11193 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
11195 * tree-vect-patterns.c (vect_get_internal_def): New function.
11196 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
11197 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
11198 (search_type_for_mask_1): Use it.
11200 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
11202 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
11203 redundant WIDEN_SUM_EXPR handling.
11204 (vect_recog_sad_pattern): Likewise.
11206 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
11208 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
11209 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
11210 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
11211 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
11212 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
11214 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
11216 * tree-vect-stmts.c (vectorizable_call): Make sure that we
11217 use the stmt_vec_info of the original bb statement for the
11218 new zero assignment, even if the call is part of a pattern.
11220 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
11222 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
11223 that the sequence is attached to the original statement rather
11224 than the pattern statement.
11225 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
11226 PATTERN_DEF_SEQ from the original statement rather than
11227 the main pattern statement.
11228 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
11229 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
11230 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
11232 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
11234 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
11235 definition statements before the early exit for statements that aren't
11237 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
11239 (vect_transform_loop): ...here. Process pattern definition
11240 statements without first checking whether the main pattern
11241 statement is live or relevant.
11243 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
11245 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
11246 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
11248 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
11250 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
11251 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
11252 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
11253 (expand_block_compare): Change select_block_compare_mode call.
11254 (expand_strncmp_align_check): Use new functions, fix comment.
11255 (emit_final_str_compare_gpr): New function.
11256 (expand_strn_compare): Refactor and clean up code.
11257 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
11259 2018-06-19 Tony Reix <tony.reix@atos.com>
11260 Damien Bergamini <damien.bergamini@atos.com>
11261 David Edelsohn <dje.gcc@gmail.com>
11263 * collect2.c (static_obj): New variable.
11264 (static_libs): New variable.
11265 (is_in_list): Uncomment declaration.
11266 (main): Track AIX libraries linked statically.
11267 (is_in_list): Uncomment definition.
11268 (scan_prog_file): Don't add AIX shared libraries initializer
11269 to constructor list if linking statically.
11271 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
11273 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
11275 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
11277 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
11279 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
11282 2018-06-19 Martin Liska <mliska@suse.cz>
11284 * config/i386/i386.c (ix86_can_inline_p): Do not use
11285 ipa_fn_summaries::get_create.
11286 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
11288 (devirtualization_time_bonus): Likewise.
11289 (ipcp_propagate_stage): Likewise.
11290 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
11291 (edge_set_predicate): Likewise.
11292 (evaluate_conditions_for_known_args): Likewise.
11293 (evaluate_properties_for_edge): Likewise.
11294 (ipa_call_summary::reset): Tranform to ...
11295 (ipa_call_summary::~ipa_call_summary): ... this.
11296 (ipa_fn_summary::reset): Transform to ...
11297 (ipa_fn_summary::~ipa_fn_summary): ... this.
11298 (ipa_fn_summary_t::remove): Rename to ...
11299 (ipa_fn_summary_t::remove_callees): ... this.
11300 (ipa_fn_summary_t::duplicate): Use placement new
11301 instead of memory copy.
11302 (ipa_call_summary_t::duplicate): Likewise.
11303 (ipa_call_summary_t::remove): Remove.
11304 (dump_ipa_call_summary): Change get_create to get.
11305 (ipa_dump_fn_summary): Dump only when summary exists.
11306 (analyze_function_body): Use symbol_summary::get instead
11308 (compute_fn_summary): Likewise.
11309 (estimate_edge_devirt_benefit): Likewise.
11310 (estimate_edge_size_and_time): Likewise.
11311 (inline_update_callee_summaries): Likewise.
11312 (remap_edge_change_prob): Likewise.
11313 (remap_edge_summaries): Likewise.
11314 (ipa_merge_fn_summary_after_inlining): Likewise.
11315 (write_ipa_call_summary): Likewise.
11316 (ipa_fn_summary_write): Likewise.
11317 (ipa_free_fn_summary): Likewise.
11318 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
11319 (struct ipa_call_summary): Likewise.
11320 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
11322 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
11323 (estimate_size_after_inlining): Likewise.
11324 (estimate_growth): Likewise.
11325 (growth_likely_positive): Likewise.
11326 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
11327 (inline_call): Likewise.
11328 * ipa-inline.c (caller_growth_limits): Likewise.
11329 (can_inline_edge_p): Likewise.
11330 (can_inline_edge_by_limits_p): Likewise.
11331 (compute_uninlined_call_time): Likewise.
11332 (compute_inlined_call_time): Likewise.
11333 (want_inline_small_function_p): Likewise.
11334 (edge_badness): Likewise.
11335 (update_caller_keys): Likewise.
11336 (update_callee_keys): Likewise.
11337 (inline_small_functions): Likewise.
11338 (inline_to_all_callers_1): Likewise.
11339 (dump_overall_stats): Likewise.
11340 (early_inline_small_functions): Likewise.
11341 (early_inliner): Likewise.
11342 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
11343 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
11344 * ipa-pure-const.c (malloc_candidate_p): Likewise.
11345 * ipa-split.c (execute_split_functions): Likewise.
11346 * symbol-summary.h: Likewise.
11347 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
11349 2018-06-19 Richard Biener <rguenther@suse.de>
11351 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
11352 (vectorize_loops): ... here. Fix dbgcnt handling.
11353 (try_vectorize_loop): Wrap try_vectorize_loop_1.
11355 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
11358 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
11359 ieee128 argument takes up only one (vector) register, not two (floating
11362 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
11364 * gimplify.c (gimplify_init_constructor): Really never clear for an
11365 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
11367 2018-06-19 Richard Biener <rguenther@suse.de>
11369 PR tree-optimization/86179
11370 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
11371 after failed recognition.
11373 2018-06-18 Martin Sebor <msebor@redhat.com>
11375 PR middle-end/85602
11376 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
11377 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
11378 Handle integer subtraction.
11379 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
11380 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
11382 2018-06-18 David Malcolm <dmalcolm@redhat.com>
11384 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
11385 param from rtx to rtx_insn *.
11386 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
11388 (frv_ifcvt_modify_insn): Likwise.
11389 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
11390 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
11391 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
11392 as_a <rtx_insn *> cast to local "unprotected_region" once
11393 it's been established that it's not NULL or pc_rtx.
11394 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
11395 param "sethi" from rtx to rtx_insn *.
11396 (nds32_group_float_insns): Likewise for param "insn".
11397 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
11399 (vax_output_int_subtract): Likewise.
11400 * config/vax/vax.c (vax_output_int_add): Likewise for param
11402 (vax_output_int_subtract): Likewise.
11403 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
11404 (emit_pattern_after): Likewise for param "after".
11405 (emit_insn_after): Likewise.
11406 (emit_jump_insn_after): Likewise.
11407 (emit_call_insn_after): Likewise.
11408 (emit_debug_insn_after): Likewise.
11409 (emit_pattern_before): Likewise for param "before".
11410 (emit_insn_before): Likewise.
11411 (emit_jump_insn_before): Likewise.
11412 * final.c (get_insn_template): Likewise for param "insn", removing
11414 * output.h (get_insn_template): Likewise for 2nd param.
11415 * rtl.h (emit_insn_before): Likewise.
11416 (emit_jump_insn_before): Likewise.
11417 (emit_debug_insn_before_noloc): Likewise.
11418 (emit_insn_after): Likewise.
11419 (emit_jump_insn_after): Likewise.
11420 (emit_call_insn_after): Likewise.
11421 (emit_debug_insn_after): Likewise.
11422 (set_insn_deleted): Likewise for param.
11424 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
11427 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
11428 floating point modes, so that IFmode is numerically greater than
11429 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
11430 to declare the ordering. This prevents IFmode from being
11431 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
11432 machine. Include rs6000-modes.h to share the fractional values
11433 between genmodes* and the rest of the compiler.
11434 (IFmode): Likewise.
11435 (KFmode): Likewise.
11436 (TFmode): Likewise.
11437 * config/rs6000/rs6000-modes.h: New file.
11438 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
11439 meaning of rs6000_long_double_size so that 126..128 selects an
11440 appropriate 128-bit floating point type.
11441 (rs6000_option_override_internal): Likewise.
11442 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
11443 (TARGET_LONG_DOUBLE_128): Change the meaning of
11444 rs6000_long_double_size so that 126..128 selects an appropriate
11445 128-bit floating point type.
11446 (LONG_DOUBLE_TYPE_SIZE): Update comment.
11447 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
11448 source and destination to match the standard usage.
11449 (truncifkf2): Likewise.
11450 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
11451 ISA 2.07 to use an explicit clobber, instead of passing in a
11453 (copysign<mode>3_soft): Likewise.
11455 2018-06-18 David Malcolm <dmalcolm@redhat.com>
11457 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
11458 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
11459 (vect_slp_analyze_instance_dependence): Likewise.
11460 (vect_enhance_data_refs_alignment): Likewise.
11461 (vect_analyze_data_refs_alignment): Likewise.
11462 (vect_slp_analyze_and_verify_instance_alignment
11463 (vect_analyze_data_ref_accesses): Likewise.
11464 (vect_prune_runtime_alias_test_list): Likewise.
11465 (vect_analyze_data_refs): Likewise.
11466 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
11467 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
11468 (vect_analyze_scalar_cycles_1): Likewise.
11469 (vect_get_loop_niters): Likewise.
11470 (vect_analyze_loop_form_1): Likewise.
11471 (vect_update_vf_for_slp): Likewise.
11472 (vect_analyze_loop_operations): Likewise.
11473 (vect_analyze_loop): Likewise.
11474 (vectorizable_induction): Likewise.
11475 (vect_transform_loop): Likewise.
11476 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
11477 * tree-vect-slp.c (vect_analyze_slp): Likewise.
11478 (vect_make_slp_decision): Likewise.
11479 (vect_detect_hybrid_slp): Likewise.
11480 (vect_slp_analyze_operations): Likewise.
11481 (vect_slp_bb): Likewise.
11482 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
11483 (vectorizable_bswap): Likewise.
11484 (vectorizable_call): Likewise.
11485 (vectorizable_simd_clone_call): Likewise.
11486 (vectorizable_conversion): Likewise.
11487 (vectorizable_assignment): Likewise.
11488 (vectorizable_shift): Likewise.
11489 (vectorizable_operation): Likewise.
11490 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
11492 2018-06-18 Martin Sebor <msebor@redhat.com>
11494 PR tree-optimization/81384
11495 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
11496 * builtins.c (expand_builtin_strnlen): New function.
11497 (expand_builtin): Call it.
11498 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
11499 * builtins.def (BUILT_IN_STRNLEN): New.
11500 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
11501 Warn for bounds in excess of maximum object size.
11502 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
11503 single-value ranges. Handle strnlen.
11504 (handle_builtin_strlen): Handle strnlen.
11505 (strlen_check_and_optimize_stmt): Same.
11506 * doc/extend.texi (Other Builtins): Document strnlen.
11508 2018-06-18 Maya Rashish <coypu@sdf.org>
11510 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
11511 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
11512 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
11514 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
11516 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
11518 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11520 * tree.c (escaped_string::escape): Replace cast to char * by
11521 const_cast<char *> (unescaped).
11523 2018-06-18 Nick Clifton <nickc@redhat.com>
11526 * tree.c (escaped_string): New class. Converts an unescaped
11527 string into its escaped equivalent.
11528 (warn_deprecated_use): Use the new class to convert the
11529 deprecation message, if present.
11530 (test_escaped_strings): New self test.
11531 (test_c_tests): Add test_escaped_strings.
11532 * doc/extend.texi (deprecated): Add a note that the
11533 deprecation message is affected by the -fmessage-length
11534 option, and that control characters will be escaped.
11535 (#pragma GCC error): Document this pragma.
11536 (#pragma GCC warning): Likewise.
11537 * doc/invoke.texi (-fmessage-length): Document this option's
11538 effect on the #warning and #error preprocessor directives and
11539 the deprecated attribute.
11541 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
11543 * tree.c (decl_value_expr_lookup): Revert latest change.
11544 (decl_value_expr_insert): Likewise.
11546 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
11548 * gimplify.c (nonlocal_vlas): Delete.
11549 (nonlocal_vla_vars): Likewise.
11550 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
11552 (gimplify_body): Do not create and destroy nonlocal_vlas.
11553 * tree-nested.c: Include diagnostic.h.
11554 (use_pointer_in_frame): Tweak.
11555 (lookup_field_for_decl): Add assertion and declare the transformation.
11556 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
11557 internal error when the reference is in a wrong context. Do not
11558 create a debug decl by default.
11559 (note_nonlocal_block_vlas): Delete.
11560 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
11561 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
11562 create a debug decl by default.
11563 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
11564 call is in a wrong context.
11565 (fixup_vla_decls): New function.
11566 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
11567 debug variables were created.
11568 * tree.c (decl_value_expr_lookup): Add checking assertion.
11569 (decl_value_expr_insert): Likewise.
11571 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
11573 PR middle-end/82479
11574 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
11575 * tree-scalar-evolution.c (interpret_expr): Likewise.
11576 (expression_expensive_p): Likewise.
11577 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
11578 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
11579 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
11580 (ssa_defined_by_minus_one_stmt_p): New.
11582 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
11584 PR middle-end/64946
11585 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
11586 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
11587 * dojump.c (do_jump): Likewise.
11588 * expr.c (expand_expr_real_2): Check operand type's sign.
11589 * fold-const.c (const_unop): Handle ABSU_EXPR.
11590 (fold_abs_const): Likewise.
11591 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
11592 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
11593 (strip_sign_op_1): Likesise.
11594 * match.pd: Add new pattern to generate ABSU_EXPR.
11595 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
11596 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
11597 * tree-eh.c (operation_could_trap_helper_p): Likewise.
11598 * tree-inline.c (estimate_operator_cost): Likewise.
11599 * tree-pretty-print.c (dump_generic_node): Likewise.
11600 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
11601 * tree.def (ABSU_EXPR): New.
11603 2018-06-16 Jakub Jelinek <jakub@redhat.com>
11605 PR middle-end/86095
11606 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
11607 documented as preserved for backward compatibility only.
11608 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
11610 PR rtl-optimization/86108
11611 * bb-reorder.c (create_forwarder_block): Renamed to ...
11612 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
11613 jump from new landing pad to the second part.
11614 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
11617 2018-06-15 Jakub Jelinek <jakub@redhat.com>
11619 PR middle-end/85878
11620 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
11621 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
11622 Only call store_expr for halves if the mode is the same.
11624 PR middle-end/86123
11625 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
11626 Fix up comment formatting.
11628 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
11630 * typed-splay-tree.h (typed_splay_tree::remove): New function.
11631 (typed_splay_tree::closure,
11632 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
11633 (typed_splay_tree::typed_splay_tree,
11634 typed_splay_tree::operator =): Declared private.
11635 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
11636 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
11637 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
11638 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
11639 typed_splay_tree::splay_tree_splay,
11640 typed_splay_tree::splay_tree_foreach_helper,
11641 typed_splay_tree::splay_tree_insert,
11642 typed_splay_tree::splay_tree_remove,
11643 typed_splay_tree::splay_tree_lookup,
11644 typed_splay_tree::splay_tree_predecessor,
11645 typed_splay_tree::splay_tree_successor,
11646 typed_splay_tree::splay_tree_min,
11647 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
11648 (typed_splay_tree::root, typed_splay_tree::comp,
11649 typed_splay_tree::delete_key,
11650 typed_splay_tree::delete_value): New data members.
11651 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
11652 typed_splay_tree::remove.
11654 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
11656 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
11657 -mginv and -mno-ginv to the assembler.
11658 * config/mips/mips.opt (-mcrc): New option.
11659 (-mginv): Likewise.
11660 * doc/invoke.text (-mcrc): Document.
11661 (-mginv): Likewise.
11663 2018-06-15 Nick Clifton <nickc@redhat.com>
11666 * tree.c (escaped_string): New class. Converts an unescaped
11667 string into its escaped equivalent.
11668 (warn_deprecated_use): Use the new class to convert the
11669 deprecation message, if present.
11670 (test_escaped_strings): New self test.
11671 (test_c_tests): Add test_escaped_strings.
11672 * doc/extend.texi (deprecated): Add a note that the
11673 deprecation message is affected by the -fmessage-length
11674 option, and that control characters will be escaped.
11675 (#pragma GCC error): Document this pragma.
11676 (#pragma GCC warning): Likewise.
11677 * doc/invoke.texi (-fmessage-length): Document this option's
11678 effect on the #warning and #error preprocessor directives and
11679 the deprecated attribute.
11681 2018-06-15 Richard Biener <rguenther@suse.de>
11683 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
11684 here, also noting vector size used.
11685 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
11686 size used in MSG_OPTIMIZED_LOCATIONS dump.
11687 (pass_slp_vectorize::execute): Adjust.
11689 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
11692 * config/arc/arc.c (arc_return_address_register): Fix
11695 2018-06-15 Richard Biener <rguenther@suse.de>
11697 PR middle-end/86159
11698 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
11699 leave useless conversion stripping to force_gimple_operand_gsi.
11700 (gimplify_build2): Likewise.
11701 (gimplify_build1): Likewise.
11703 2018-06-15 Richard Biener <rguenther@suse.de>
11705 PR middle-end/86076
11706 * tree-cfg.c (move_stmt_op): unshare invariant addresses
11707 before adjusting their block.
11709 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
11711 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
11712 multilibs for *-*-rtems*.
11713 * config/riscv/t-rtems: New file.
11715 2018-06-14 Jakub Jelinek <jakub@redhat.com>
11717 PR middle-end/86122
11718 * match.pd ((A +- CST1) +- CST2): Punt if last resort
11719 unsigned_type_for returns NULL.
11722 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
11723 subregs of multi-word pseudos unless the float mode has word size.
11725 2018-06-14 Richard Biener <rguenther@suse.de>
11727 PR middle-end/86139
11728 * tree-vect-generic.c (build_word_mode_vector_type): Remove
11729 duplicate and harmful type_hash_canon.
11730 * tree.c (type_hash_canon): Assert we didn't find ourselves.
11732 2018-06-14 Richard Biener <rguenther@suse.de>
11735 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
11738 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
11740 * config/rtems.h (STDINT_LONG32): Define.
11742 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
11743 Prachi Godbole <prachi.godbole@imgtec.com>
11745 * config/mips/mips-cpus.def: Define P6600.
11746 * config/mips/mips-tables.opt: Regenerate.
11747 * config/mips/mips.c (mips_ucbranch_type): New enum.
11748 (mips_rtx_cost_data): Add support for P6600.
11749 (mips_issue_rate): Likewise.
11750 (mips_multipass_dfa_lookahead): Likewise.
11751 (mips_avoid_hazard): Likewise.
11752 (mips_reorg_process_insns): Likewise.
11753 (mips_classify_branch_p6600): New function.
11754 * config/mips/mips.h (TUNE_P6600): New define.
11755 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
11756 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
11757 * config/mips/mips.md: Include p6600.md.
11758 (processor): Add p6600.
11759 * config/mips/p6600.md: New file.
11760 * doc/invoke.texi: Add p6600 to supported architectures.
11762 2018-06-13 Martin Sebor <msebor@redhat.com>
11764 PR tree-optimization/86114
11765 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
11767 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
11769 2018-06-13 Richard Biener <rguenther@suse.de>
11771 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
11772 Properly set vector type of the intermediate stmt.
11773 * tree-vect-stmts.c (vectorizable_operation): The destination
11774 var always has vectype_out type.
11776 2018-06-13 Jeff Law <law@redhat.com>
11778 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
11779 integer 0 for argument to print_rtl_with_bb.
11780 (rl78_reorg): Likewise.
11782 2018-06-13 David Malcolm <dmalcolm@redhat.com>
11784 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
11785 from rtx to rtx_insn *.
11786 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
11788 (add_sched_insns_for_speculation): Likewise for local "target",
11789 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
11790 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
11791 from rtx_insn ** to rtx_code_label **.
11792 (reorg_emit_nops): Likewise.
11793 (c6x_reorg): Likewise for local "call_labels".
11794 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
11796 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
11797 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
11798 the loops over LABEL_REFS.
11799 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
11801 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
11802 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
11803 (split_branches): Strengthen local "olabel" from rtx to
11804 rtx_insn *, adding a safe_as_a cast.
11805 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
11807 (add_insn_after): Likewise for first two params.
11808 (add_insn_before): Likewise.
11809 (remove_insn): Likewise for param.
11810 (emit_pattern_before_noloc): Likewise for second and third params.
11811 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
11812 (emit_call_insn_before_noloc): Likewise.
11813 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
11815 (emit_barrier_before): Likewise.
11816 (emit_label_before): Strengthen "label" param from "rtx" to
11817 "rtx_code_label *". Strengthen "before" param from "rtx" to
11819 (emit_insn_after_1): Strengthen "after" param from "rtx" to
11821 (emit_pattern_after_noloc): Likewise.
11822 (emit_insn_after_noloc): Likewise.
11823 (emit_jump_insn_after_noloc): Likewise.
11824 (emit_call_insn_after_noloc): Likewise.
11825 (emit_debug_insn_after_noloc): Likewise.
11826 (emit_barrier_after): Likewise.
11827 (emit_label_after): Likewise for both params.
11828 (emit_pattern_after_setloc): Likewise for "after" param. Convert
11829 "loc" param from "int" to "location_t".
11830 (emit_insn_after_setloc): Likewise.
11831 (emit_jump_insn_after_setloc): Likewise.
11832 (emit_call_insn_after_setloc): Likewise.
11833 (emit_debug_insn_after_setloc): Likewise.
11834 (emit_pattern_before_setloc): Likewise for "before" param. Convert
11835 "loc" param from "int" to "location_t".
11836 (emit_pattern_before): Convert NULL_RTX to NULL.
11837 (emit_insn_before_setloc): Convert "loc" param from "int" to
11839 (emit_jump_insn_before_setloc): Likewise.
11840 (emit_call_insn_before_setloc): Likewise.
11841 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
11842 rtx_insn *. Convert "loc" param from "int" to "location_t".
11843 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
11844 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
11845 Convert 3rd param from "int" to "location_t".
11846 (emit_barrier_before, emit_barrier_after, next_real_insn):
11847 Strengthen param from rtx to rtx_insn *.
11848 (emit_label_before): Strengthen 1st param from "rtx" to
11849 "rtx_code_label *". Strengthen 2nd param from "rtx" to
11851 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
11852 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
11853 Strengthen 2nd param from "rtx" to "rtx_insn *".
11854 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
11855 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
11856 Likewise. Convert 3rd param from "int" to "location_t".
11857 (emit_label_after): Strengthen 1st param from "rtx" to
11858 "rtx_code_label *".
11859 (next_real_insn, remove_insn): Strengthen param from "rtx" to
11861 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
11862 from "rtx" to "rtx_insn *".
11864 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
11866 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
11867 bodies streamed in with -Q.
11868 * dumpfile.c (dump_files): Add lto-stream-out dump file.
11869 * dumpfile.h (tree_dump_index): Add lto_stream_out.
11870 * gimple-streamer-out.c: Include gimple-pretty-print.h
11871 (output_bb): Dump stmts streamed.
11872 * lto-section-out.c: Include print-tree.h
11873 (lto_begin_section): Dump sections created.
11874 (lto_output_decl_index): Dump decl encoded.
11875 * lto-streamer-out.c: Include print-tree.h
11876 (create_output_block): Dump output block created.
11877 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
11878 (output_function): Dump function output.
11879 (output_constructor): Dump constructor streamed.
11880 (write_global_stream): Output indexes encoded.
11881 (produce_asm_for_decls): Dump streams encoded.
11882 * lto-streamer.c (streamer_dump_file): New global var.
11883 * lto-streamer.h (streamer_dump_file): Declare.
11884 * passes.c (ipa_write_summaries): Initialize streamer dump.
11885 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
11888 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
11891 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
11892 offsets for register save directives. Emit a second batch of save
11893 directives, if need be, when the function accesses prior frames.
11895 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
11897 * config/arc/fpu.md (fmasf4): Force operand to register.
11898 (fnmasf4): Likewise.
11900 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
11902 * config/arc/arc-protos.h (arc_pad_return): Remove.
11903 * config/arc/arc.c (machine_function): Remove force_short_suffix
11905 (arc_print_operand): Adjust printing of '&'.
11906 (arc_verify_short): Remove conditional printing of short suffix.
11907 (arc_final_prescan_insn): Remove reference to size_reason.
11908 (pad_return): New function.
11909 (arc_reorg): Call pad_return.
11910 (arc_pad_return): Remove.
11911 (arc_init_machine_status): Remove reference to force_short_suffix.
11912 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
11913 (attr length): When attribute iscompact is true force to 2
11914 regardless; in the case of maybe check if we want to force the
11915 instruction to have 4 bytes length.
11916 (nopv): Change it to generate 4 byte long nop as well.
11917 (blockage): New pattern.
11918 (simple_return): Remove call to arc_pad_return.
11919 (p_return_i): Likewise.
11921 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
11923 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
11925 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
11927 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
11930 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
11932 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
11933 for ARC700 and ARCv2.
11935 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
11938 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
11939 operands[2] instead of operands[1].
11942 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
11944 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
11945 case, check whether the outer register overlaps an unallocatable
11946 register, not just whether it fits the required class.
11948 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
11950 * poly-int.h (can_div_trunc_p): Add new overload in which all values
11952 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
11953 (memrefs_conflict_p): Likewise.
11954 (init_alias_analysis): Likewise.
11955 * cfgexpand.c (expand_debug_expr): Likewise.
11956 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
11957 * cse.c (fold_rtx): Likewise.
11958 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
11959 * expr.c (emit_block_move_hints): Likewise.
11960 (clear_storage_hints, push_block, emit_push_insn): Likewise.
11961 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
11962 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
11963 (emit_group_store): Likewise.
11964 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
11965 to read the PRE/POST_MODIFY increment.
11966 * calls.c (store_one_arg): Use strip_offset.
11967 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
11969 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
11971 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
11972 (simplify_binary_operation_1): Extend CONST_INT handling to
11974 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
11975 than a HOST_WIDE_INT.
11976 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
11978 (adjust_mems, add_stores): Update accodingly.
11979 (vt_canonicalize_addr): Track polynomial offsets.
11980 (emit_note_insn_var_location): Likewise.
11981 (vt_add_function_parameter): Likewise.
11982 (vt_initialize): Likewise.
11984 2018-06-12 Jeff Law <law@redhat.com>
11986 * config.gcc (alpha*-*-freebsd*): Remove.
11987 * config/alpha/freebsd.h: Remove.
11989 2018-06-12 David Malcolm <dmalcolm@redhat.com>
11992 * spellcheck-tree.c (levenshtein_distance): Rename to...
11993 (get_edit_distance): ...this, and update for underlying renaming.
11994 * spellcheck-tree.h (levenshtein_distance): Rename to...
11995 (get_edit_distance): ...this.
11996 * spellcheck.c (levenshtein_distance): Rename to...
11997 (get_edit_distance): ...this. Convert from Levenshtein distance
11998 to Damerau-Levenshtein distance by supporting transpositions of
11999 adjacent characters. Rename "v1" to "v_next" and "v0" to
12001 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
12002 (selftest::test_edit_distance_unit_test_oneway): ...this, and
12003 update for underlying renaming.
12004 (selftest::levenshtein_distance_unit_test): Rename to...
12005 (selftest::test_get_edit_distance_unit): ...this, and update for
12006 underlying renaming.
12007 (selftest::test_find_closest_string): Add example from PR 69968
12008 where transposition helps
12009 (selftest::test_metric_conditions): Update for renaming.
12010 (selftest::test_metric_conditions): Likewise.
12011 (selftest::spellcheck_c_tests): Likewise.
12012 * spellcheck.h (levenshtein_distance): Rename both overloads to...
12013 (get_edit_distance): ...this.
12014 (best_match::consider): Update for renaming.
12016 2018-06-12 Martin Sebor <msebor@redhat.com>
12018 PR tree-optimization/85259
12019 * builtins.c (compute_objsize): Handle constant offsets.
12020 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
12021 true iff a warning has been issued.
12022 * gimple.h (gimple_nonartificial_location): New function.
12023 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
12024 gimple_nonartificial_location and handle -Wno-system-headers.
12025 (handle_builtin_stxncpy): Same.
12027 2018-06-12 Martin Sebor <msebor@redhat.com>
12030 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
12032 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
12034 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12035 BUILTIN_VEC_XST entries for pointer to double and long long.
12037 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
12040 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
12042 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
12045 2018-06-12 Martin Liska <mliska@suse.cz>
12047 * doc/options.texi: Document IntegerRange.
12049 2018-06-12 Martin Liska <mliska@suse.cz>
12051 * config/i386/i386.opt: Make MPX-related options as Deprecated.
12052 * opt-functions.awk: Handle Deprecated flag.
12053 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
12055 (read_cmdline_option): Report warning for a deprecated option.
12056 * opts.h (struct cl_option): Add new field cl_deprecated.
12057 (CL_ERR_DEPRECATED): New.
12059 2018-06-12 Martin Liska <mliska@suse.cz>
12061 * doc/options.texi: Document Deprecated option flag.
12063 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
12065 * config/arc/arc-arch.h (arc_extras): New enum.
12066 (arc_cpu_t):Add field extra.
12067 (arc_cpu_types): Consider the extras.
12068 * config/arc/arc-cpus.def: Add extras info.
12069 * config/arc/arc-opts.h (processor_type): Consider extra field.
12070 * config/arc/arc.c (arc_override_options): Handle extra field.
12072 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
12074 * config/arc/arc-arch.h: Update ARC_OPTX macro.
12075 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
12077 * config/arc/arc.c (arc_init): Update pic warning.
12078 (irq_range): Update irq range parsing warnings.
12079 (arc_override_options): Update various warning messages.
12080 (arc_handle_aux_attribute): Likewise.
12082 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
12084 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
12086 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
12088 * doc/sourcebuild.texi: Document usage of line number 0 in verify
12089 compiler messages directives.
12091 2018-06-12 Matthew Fortune <mfortune@gmail.com>
12093 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
12094 * config/mips/mips-tables.opt: Regenerate.
12095 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
12097 * doc/invoke.texi: Document -march=i6500.
12099 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
12101 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
12102 (i6400_gpmul): Add cpu_unit.
12103 (i6400_gpdiv): Likewise.
12104 (i6400_msa_add_d): Update reservations.
12105 (i6400_msa_int_add) Likewise.
12106 (i6400_msa_short_logic3) Likewise.
12107 (i6400_msa_short_logic2) Likewise.
12108 (i6400_msa_short_logic) Likewise.
12109 (i6400_msa_move) Likewise.
12110 (i6400_msa_cmp) Likewise.
12111 (i6400_msa_short_float2) Likewise.
12112 (i6400_msa_div_d) Likewise.
12113 (i6400_msa_long_logic1) Likewise.
12114 (i6400_msa_long_logic2) Likewise.
12115 (i6400_msa_mult) Likewise.
12116 (i6400_msa_long_float2) Likewise.
12117 (i6400_msa_long_float4) Likewise.
12118 (i6400_msa_long_float5) Likewise.
12119 (i6400_msa_long_float8) Likewise.
12120 (i6400_fpu_fadd): Include frint type.
12121 (i6400_fpu_store): New define_insn_reservation.
12122 (i6400_fpu_load): Likewise.
12123 (i6400_fpu_move): Likewise.
12124 (i6400_fpu_fcmp): Likewise.
12125 (i6400_fpu_fmadd): Likewise.
12126 (i6400_int_mult): Include imul3nc type and update reservation.
12127 (i6400_int_div): Include idiv3 type and update reservation.
12128 (i6400_int_load): Update to check type not move_type.
12129 (i6400_int_store): Likewise.
12130 (i6400_int_prefetch): Set zero latency.
12132 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
12134 * gcc.c: Document new %@{...} sequence.
12135 (LINK_COMMAND_SPEC): Use it for the -L switches.
12136 (cpp_unique_options): Use it for the -I switches.
12137 (at_file_argbuf): New global variable.
12138 (in_at_file): Likewise.
12139 (alloc_args): Create at_file_argbuf.
12140 (clear_args): Truncate at_file_argbuf.
12141 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
12142 (open_at_file): New function.
12143 (close_at_file): Likewise.
12144 (create_at_file): Delete.
12145 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
12148 (validate_switches_from_spec): Deal with %@{...} sequence.
12149 (validate_switches): Likewise.
12150 (driver::finalize): Call clear_args.
12152 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
12154 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
12156 2018-06-11 Martin Sebor <msebor@redhat.com>
12158 * doc/invoke.texi (-Wall): List -Wc++17-compat.
12159 (Wno-class-memaccess): Add @opindex.
12160 (Wno-templates, Wno-multiple-inheritance): Same.
12161 (Wno-virtual-inheritance, Wno-namespaces): Same.
12162 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
12163 (Wno-format-overflow, Wno-format-truncation): Same.
12164 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
12165 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
12166 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
12167 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
12168 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
12169 (Wno-misspelled-isr): Same.
12171 2018-06-11 Martin Sebor <msebor@redhat.com>
12173 * PR tree-optimization/86083
12174 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
12176 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
12178 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
12180 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
12183 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
12184 on the correct operand.
12185 (*movdi_internal64): Ditto.
12187 2018-06-11 Martin Liska <mliska@suse.cz>
12189 PR tree-optimization/86089
12190 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
12192 2018-06-11 Julia Koval <julia.koval@intel.com>
12194 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
12195 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
12196 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
12198 2018-06-11 Olivier Hainque <hainque@adacore.com>
12200 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
12201 for Ada with strict dwarf2.
12203 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
12206 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
12209 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
12211 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
12213 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
12215 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
12218 2018-06-08 Martin Liska <mliska@suse.cz>
12220 * tree-cfg.h (debug_function): Fix argument type to match
12223 2018-06-08 Martin Liska <mliska@suse.cz>
12225 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
12226 Remove usage of MPX-related (and removed) fields.
12227 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
12229 2018-06-08 David Malcolm <dmalcolm@redhat.com>
12231 * cfg.c (debug): Use TDF_NONE rather than 0.
12232 * cfghooks.c (debug): Likewise.
12233 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
12234 (struct dump_option_value_info): Convert to...
12235 (struct kv_pair): ...this template type.
12236 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
12238 (optinfo_verbosity_options): Likewise.
12239 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
12241 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
12242 than int for "optgroup_flags" param.
12243 (dump_generic_expr_loc): Use dump_flags_t rather than int for
12245 (dump_dec): Likewise.
12246 (dump_finish): Use TDF_NONE rather than 0.
12247 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
12248 rather than int for "optgroup_flags" param. Use TDF_NONE rather
12249 than 0. Update for change to option_ptr.
12250 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
12251 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
12252 0. Update for changes to optinfo_verbosity_options and
12254 (opt_info_switch_p): Convert optgroup_flags from int to
12256 (dump_basic_block): Use dump_flags_t rather than int
12257 for "dump_kind" param.
12258 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
12259 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
12260 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
12261 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
12262 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
12263 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
12264 TDF_NONE): Convert from macros to...
12265 (enum dump_flag): ...this new enum.
12266 (dump_flags_t): Update to use enum.
12267 (operator|, operator&, operator~, operator|=, operator&=):
12268 Implement for dump_flags_t.
12269 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
12270 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
12271 Convert from macros to...
12272 (enum optgroup_flag): ...this new enum.
12273 (optgroup_flags_t): New typedef.
12274 (operator|, operator|=): Implement for optgroup_flags_t.
12275 (struct dump_file_info): Convert field "alt_flags" to
12276 dump_flags_t. Convert field "optgroup_flags" to
12278 (dump_basic_block): Use dump_flags_t rather than int for param.
12279 (dump_generic_expr_loc): Likewise.
12280 (dump_dec): Likewise.
12281 (dump_register): Convert param "optgroup_flags" to
12283 (opt_info_enable_passes): Likewise.
12284 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
12286 * gimple-pretty-print.c (debug): Likewise.
12287 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
12288 (merged_store_group::apply_stores): Likewise.
12289 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
12290 * gimple.c (verify_gimple_pp): Likewise.
12291 * graphite-poly.c (print_pbb_body): Likewise.
12292 * passes.c (pass_manager::register_one_dump_file): Convert
12293 local "optgroup_flags" to optgroup_flags_t.
12294 * print-tree.c (print_node): Use TDF_NONE rather than 0.
12296 (debug_body): Likewise.
12297 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
12298 to optgroup_flags_t.
12299 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
12301 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
12302 (convert_mult_to_fma): Likewise.
12303 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
12304 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
12305 * tree-vect-data-refs.c (dump_lower_bound): Convert param
12306 "dump_kind" to dump_flags_t.
12308 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
12310 * config/rs6000/rs6000.c (min, max): Delete.
12312 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
12314 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
12315 -mabi=spe and -mabi=no-spe.
12317 2018-06-08 Martin Liska <mliska@suse.cz>
12319 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
12320 where we expect an existing summary.
12322 2018-06-08 Martin Liska <mliska@suse.cz>
12324 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
12325 * ipa-inline.h (estimate_edge_growth): Likewise.
12327 2018-06-08 Martin Liska <mliska@suse.cz>
12329 * cgraph.c (function_version_hasher::hash): Use
12330 cgraph_node::get_uid ().
12331 (function_version_hasher::equal):
12332 * cgraph.h (cgraph_node::get_uid): New method.
12333 * ipa-inline.c (update_caller_keys): Use
12334 cgraph_node::get_uid ().
12335 (update_callee_keys): Likewise.
12336 * ipa-utils.c (searchc): Likewise.
12337 (ipa_reduced_postorder): Likewise.
12338 * lto-cgraph.c (input_node): Likewise.
12339 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
12340 * symbol-summary.h (symtab_insertion): Likewise.
12341 (symtab_removal): Likewise.
12342 (symtab_duplication): Likewise.
12343 * tree-pretty-print.c (dump_function_header): Likewise.
12344 * tree-sra.c (convert_callers_for_node): Likewise.
12346 2018-06-08 Martin Liska <mliska@suse.cz>
12348 * cgraph.c (symbol_table::create_edge): Always assign a new
12350 (symbol_table::free_edge): Do not recycle numbers.
12351 * cgraph.h (cgraph_edge::get): New method.
12352 * symbol-summary.h (symtab_removal): Use it.
12353 (symtab_duplication): Likewise.
12354 (call_summary::hashable_uid): Remove.
12356 2018-06-08 Martin Liska <mliska@suse.cz>
12358 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
12359 (initialize_growth_caches): Remove.
12360 (free_growth_caches): Likewise.
12361 (do_estimate_edge_time): Use edge_growth_cache.
12362 (do_estimate_edge_size): Likewise.
12363 (do_estimate_edge_hints): Likewise.
12364 * ipa-inline.c (reset_edge_caches): Likewise.
12365 (recursive_inlining): Likewise.
12366 (inline_small_functions): Likewise.
12367 * ipa-inline.h (initialize_growth_caches): Remove.
12368 (estimate_edge_size): Likewise.
12369 (estimate_edge_time): Likewise.
12370 (estimate_edge_hints): Likewise.
12371 (reset_edge_growth_cache): Likewise.
12372 * symbol-summary.h (call_summary::remove): New method.
12374 2018-06-08 Martin Liska <mliska@suse.cz>
12376 * ipa-cp.c (class edge_clone_summary): New summary.
12377 (grow_edge_clone_vectors): Remove.
12378 (ipcp_edge_duplication_hook): Remove.
12379 (class edge_clone_summary_t): New call_summary class.
12380 (ipcp_edge_removal_hook): Remove.
12381 (edge_clone_summary_t::duplicate): New function.
12382 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
12383 (create_specialized_node): Likewise.
12384 (ipcp_driver): Initialize edge_clone_summaries and do not
12387 2018-06-08 Martin Liska <mliska@suse.cz>
12389 * symbol-summary.h (get): New function.
12390 (call_summary::m_initialize_when_cloning): New class member.
12392 2018-06-08 Martin Liska <mliska@suse.cz>
12394 * cgraph.c (cgraph_node::remove): Do not recycle uid.
12395 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
12396 (symbol_table::allocate_cgraph_symbol): Do not set uid.
12397 * passes.c (uid_hash_t): Record removed_nodes by their uids.
12398 (remove_cgraph_node_from_order): Use the removed_nodes set.
12399 (do_per_function_toporder): Likwise.
12400 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
12401 instead of summary_uid.
12402 (symtab_removal): Likewise.
12403 (symtab_duplication): Likewise.
12405 2018-06-08 Martin Liska <mliska@suse.cz>
12407 * ipa-cp.c (ipcp_store_bits_results): Use
12408 ipcp_transformation_sum.
12409 (ipcp_store_vr_results): Likewise.
12410 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
12412 (ipcp_transformation_initialize): ... this.
12413 (ipa_set_node_agg_value_chain):
12414 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
12415 (write_ipcp_transformation_info): Likewise.
12416 (read_ipcp_transformation_info): Likewise.
12417 (ipcp_update_bits): Likewise.
12418 (ipcp_update_vr): Likewise.
12419 (ipcp_transform_function): Likewise.
12420 * ipa-prop.h: Rename ipcp_transformation_summary to
12421 ipcp_transformation.
12422 (class ipcp_transformation_t): New function summary.
12423 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
12424 (ipa_get_agg_replacements_for_node): Likewise.
12426 2018-06-08 Martin Liska <mliska@suse.cz>
12428 * ipa-pure-const.c (struct funct_state_d): Do it class instead
12430 (class funct_state_summary_t): New function_summary class.
12431 (has_function_state): Remove.
12432 (get_function_state): Likewise.
12433 (set_function_state): Likewise.
12434 (add_new_function): Likewise.
12435 (funct_state_summary_t::insert): New function.
12436 (duplicate_node_data): Remove.
12437 (remove_node_data): Remove.
12438 (funct_state_summary_t::duplicate): New function.
12439 (register_hooks): Create new funct_state_summaries.
12440 (pure_const_generate_summary): Use it.
12441 (pure_const_write_summary): Likewise.
12442 (pure_const_read_summary): Likewise.
12443 (propagate_pure_const): Likewise.
12444 (propagate_nothrow): Likewise.
12445 (dump_malloc_lattice): Likewise.
12446 (propagate_malloc): Likewise.
12447 (execute): Do not register hooks, just remove summary
12449 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
12452 2018-06-08 Martin Liska <mliska@suse.cz>
12454 * ipa-reference.c (remove_node_data): Remove.
12455 (duplicate_node_data): Likewise.
12456 (class ipa_ref_var_info_summary_t): New class.
12457 (class ipa_ref_opt_summary_t): Likewise.
12458 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
12459 (get_reference_optimization_summary): Use
12460 ipa_ref_opt_sum_summaries.
12461 (set_reference_vars_info): Remove.
12462 (set_reference_optimization_summary): Likewise.
12463 (ipa_init): Create summaries.
12464 (init_function_info): Use function summary.
12465 (ipa_ref_opt_summary_t::duplicate): New function.
12466 (ipa_ref_opt_summary_t::remove): New function.
12467 (get_read_write_all_from_node): Fix GNU coding style.
12468 (propagate): Use function summary.
12469 (write_node_summary_p): Fix GNU coding style.
12470 (stream_out_bitmap): Likewise.
12471 (ipa_reference_read_optimization_summary): Use function summary.
12472 (ipa_reference_c_finalize): Do not release hooks.
12474 2018-06-08 Martin Liska <mliska@suse.cz>
12476 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
12477 (analyze_function_body): Extract multiple calls of get_create.
12478 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
12479 * ipa-inline.c (recursive_inlining): Use ::get method.
12480 * ipa-inline.h (estimate_edge_growth): Likewise.
12482 2018-06-08 Martin Liska <mliska@suse.cz>
12484 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
12486 (hsa_function_summary::hsa_function_summary): Use the new enum
12488 (hsa_gpu_implementation_p): Use hsa_summaries::get.
12489 * hsa-gen.c (hsa_get_host_function): Likewise.
12490 (get_brig_function_name): Likewise.
12491 * ipa-hsa.c (process_hsa_functions): Likewise.
12492 (ipa_hsa_write_summary): Likewise.
12493 * symbol-summary.h (symtab_duplication): Use ::get function/
12494 (get): New function.
12496 2018-06-08 Martin Liska <mliska@suse.cz>
12498 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
12500 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
12501 (hsa_register_kernel): Likewise.
12502 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
12503 * hsa-gen.c (hsa_get_host_function): Likewise.
12504 (get_brig_function_name): Likewise.
12505 (generate_hsa): Likewise.
12506 (pass_gen_hsail::execute): Likewise.
12507 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
12508 (devirtualization_time_bonus): Likewise.
12509 (ipcp_propagate_stage): Likewise.
12510 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
12511 (edge_set_predicate): Likewise.
12512 (evaluate_conditions_for_known_args): Likewise.
12513 (evaluate_properties_for_edge): Likewise.
12514 (ipa_fn_summary::reset): Likewise.
12515 (ipa_fn_summary_t::duplicate): Likewise.
12516 (dump_ipa_call_summary): Likewise.
12517 (ipa_dump_fn_summary): Likewise.
12518 (analyze_function_body): Likewise.
12519 (compute_fn_summary): Likewise.
12520 (estimate_edge_devirt_benefit): Likewise.
12521 (estimate_edge_size_and_time): Likewise.
12522 (estimate_calls_size_and_time): Likewise.
12523 (estimate_node_size_and_time): Likewise.
12524 (inline_update_callee_summaries): Likewise.
12525 (remap_edge_change_prob): Likewise.
12526 (remap_edge_summaries): Likewise.
12527 (ipa_merge_fn_summary_after_inlining): Likewise.
12528 (ipa_update_overall_fn_summary): Likewise.
12529 (read_ipa_call_summary): Likewise.
12530 (inline_read_section): Likewise.
12531 (write_ipa_call_summary): Likewise.
12532 (ipa_fn_summary_write): Likewise.
12533 (ipa_free_fn_summary): Likewise.
12534 * ipa-hsa.c (process_hsa_functions): Likewise.
12535 (ipa_hsa_write_summary): Likewise.
12536 (ipa_hsa_read_section): Likewise.
12537 * ipa-icf.c (sem_function::merge): Likewise.
12538 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
12539 (do_estimate_edge_time): Likewise.
12540 (estimate_size_after_inlining): Likewise.
12541 (estimate_growth): Likewise.
12542 (growth_likely_positive): Likewise.
12543 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
12544 (inline_call): Likewise.
12545 * ipa-inline.c (caller_growth_limits): Likewise.
12546 (can_inline_edge_p): Likewise.
12547 (can_inline_edge_by_limits_p): Likewise.
12548 (compute_uninlined_call_time): Likewise.
12549 (compute_inlined_call_time): Likewise.
12550 (want_inline_small_function_p): Likewise.
12551 (edge_badness): Likewise.
12552 (update_caller_keys): Likewise.
12553 (update_callee_keys): Likewise.
12554 (recursive_inlining): Likewise.
12555 (inline_small_functions): Likewise.
12556 (inline_to_all_callers_1): Likewise.
12557 (dump_overall_stats): Likewise.
12558 (early_inline_small_functions): Likewise.
12559 (early_inliner): Likewise.
12560 * ipa-inline.h (estimate_edge_growth): Likewise.
12561 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
12562 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
12563 * ipa-prop.h (IPA_NODE_REF): Likewise.
12564 (IPA_EDGE_REF): Likewise.
12565 * ipa-pure-const.c (malloc_candidate_p): Likewise.
12566 (propagate_malloc): Likewise.
12567 * ipa-split.c (execute_split_functions): Likewise.
12568 * symbol-summary.h: Rename get to get_create.
12570 (get_create): Likewise.
12571 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
12573 2018-06-08 Martin Liska <mliska@suse.cz>
12575 * symbol-summary.h (release): Move definition out of class
12577 (symtab_removal): Likewise.
12578 (symtab_duplication): Likewise.
12580 2018-06-08 Martin Liska <mliska@suse.cz>
12582 * symbol-summary.h (function_summary): Move constructor
12583 implementation out of class declaration.
12584 (release): Likewise.
12585 (symtab_insertion): Likewise.
12586 (symtab_removal): Likewise.
12587 (symtab_duplication): Likewise.
12590 2018-06-08 Martin Liska <mliska@suse.cz>
12592 * Makefile.in: Remove support for MPX (macros, related functions,
12593 fields in cgraph_node, ...).
12594 * builtin-types.def (BT_BND): Likewise.
12595 (BT_FN_BND_CONST_PTR): Likewise.
12596 (BT_FN_CONST_PTR_BND): Likewise.
12597 (BT_FN_VOID_PTR_BND): Likewise.
12598 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
12599 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
12600 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
12601 (expand_builtin_mempcpy_with_bounds): Likewise.
12602 (expand_builtin_memset_with_bounds): Likewise.
12603 (expand_builtin_memset_args): Likewise.
12604 (std_expand_builtin_va_start): Likewise.
12605 (expand_builtin): Likewise.
12606 (expand_builtin_with_bounds): Likewise.
12607 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
12608 (DEF_LIB_BUILTIN_CHKP): Likewise.
12609 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
12610 (DEF_CHKP_BUILTIN): Likewise.
12611 (BUILT_IN_MEMCPY): Likewise.
12612 (BUILT_IN_MEMMOVE): Likewise.
12613 (BUILT_IN_MEMPCPY): Likewise.
12614 (BUILT_IN_MEMSET): Likewise.
12615 (BUILT_IN_STPCPY): Likewise.
12616 (BUILT_IN_STRCAT): Likewise.
12617 (BUILT_IN_STRCHR): Likewise.
12618 (BUILT_IN_STRCPY): Likewise.
12619 (BUILT_IN_STRLEN): Likewise.
12620 (BUILT_IN_MEMCPY_CHK): Likewise.
12621 (BUILT_IN_MEMMOVE_CHK): Likewise.
12622 (BUILT_IN_MEMPCPY_CHK): Likewise.
12623 (BUILT_IN_MEMSET_CHK): Likewise.
12624 (BUILT_IN_STPCPY_CHK): Likewise.
12625 (BUILT_IN_STRCAT_CHK): Likewise.
12626 (BUILT_IN_STRCPY_CHK): Likewise.
12627 * calls.c (store_bounds): Likewise.
12628 (emit_call_1): Likewise.
12629 (special_function_p): Likewise.
12630 (maybe_warn_nonstring_arg): Likewise.
12631 (initialize_argument_information): Likewise.
12632 (finalize_must_preallocate): Likewise.
12633 (compute_argument_addresses): Likewise.
12634 (expand_call): Likewise.
12635 * cfgexpand.c (expand_call_stmt): Likewise.
12636 (expand_return): Likewise.
12637 (expand_gimple_stmt_1): Likewise.
12638 (pass_expand::execute): Likewise.
12639 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
12640 (cgraph_node::remove): Likewise.
12641 (cgraph_node::dump): Likewise.
12642 (cgraph_node::verify_node): Likewise.
12643 * cgraph.h (chkp_function_instrumented_p): Likewise.
12644 (symtab_node::get_alias_target): Likewise.
12645 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
12646 (cgraph_local_p): Likewise.
12647 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
12648 (cgraph_edge::rebuild_references): Likewise.
12649 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
12650 (walk_polymorphic_call_targets): Likewise.
12651 (cgraph_node::expand_thunk): Likewise.
12652 (symbol_table::output_weakrefs): Likewise.
12653 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
12654 (ix86_handle_option): Likewise.
12655 * config/i386/constraints.md: Likewise.
12656 * config/i386/i386-builtin-types.def (BND): Likewise.
12660 * config/i386/i386-builtin.def (BDESC_END): Likewise.
12661 (BDESC_FIRST): Likewise.
12663 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
12664 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
12665 * config/i386/i386.c (enum reg_class): Likewise.
12666 (ix86_target_string): Likewise.
12667 (ix86_option_override_internal): Likewise.
12668 (ix86_conditional_register_usage): Likewise.
12669 (ix86_valid_target_attribute_inner_p): Likewise.
12670 (ix86_set_indirect_branch_type): Likewise.
12671 (ix86_set_current_function): Likewise.
12672 (ix86_function_arg_regno_p): Likewise.
12673 (init_cumulative_args): Likewise.
12674 (ix86_function_arg_advance): Likewise.
12675 (ix86_function_arg): Likewise.
12676 (ix86_pass_by_reference): Likewise.
12677 (ix86_function_value_regno_p): Likewise.
12678 (ix86_function_value_1): Likewise.
12679 (ix86_function_value_bounds): Likewise.
12680 (ix86_return_in_memory): Likewise.
12681 (ix86_setup_incoming_vararg_bounds): Likewise.
12682 (ix86_va_start): Likewise.
12683 (indirect_thunk_need_prefix): Likewise.
12684 (print_reg): Likewise.
12685 (ix86_print_operand): Likewise.
12686 (ix86_expand_call): Likewise.
12687 (ix86_output_function_return): Likewise.
12688 (reg_encoded_number): Likewise.
12689 (BDESC_VERIFYS): Likewise.
12690 (ix86_init_mpx_builtins): Likewise.
12691 (ix86_init_builtins): Likewise.
12692 (ix86_emit_cmove): Likewise.
12693 (ix86_emit_move_max): Likewise.
12694 (ix86_expand_builtin): Likewise.
12695 (ix86_builtin_mpx_function): Likewise.
12696 (ix86_get_arg_address_for_bt): Likewise.
12697 (ix86_load_bounds): Likewise.
12698 (ix86_store_bounds): Likewise.
12699 (ix86_load_returned_bounds): Likewise.
12700 (ix86_store_returned_bounds): Likewise.
12701 (ix86_class_likely_spilled_p): Likewise.
12702 (ix86_hard_regno_mode_ok): Likewise.
12703 (x86_order_regs_for_local_alloc): Likewise.
12704 (ix86_mitigate_rop): Likewise.
12705 (ix86_bnd_prefixed_insn_p): Likewise.
12706 (ix86_mpx_bound_mode): Likewise.
12707 (ix86_make_bounds_constant): Likewise.
12708 (ix86_initialize_bounds): Likewise.
12709 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
12710 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
12711 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
12712 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
12713 (TARGET_CHKP_BOUND_MODE): Likewise.
12714 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
12715 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
12716 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
12717 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
12718 * config/i386/i386.h (TARGET_MPX): Likewise.
12719 (TARGET_MPX_P): Likewise.
12720 (VALID_BND_REG_MODE): Likewise.
12721 (FIRST_BND_REG): Likewise.
12722 (LAST_BND_REG): Likewise.
12723 (enum reg_class): Likewise.
12724 (BND_REG_P): Likewise.
12725 (BND_REGNO_P): Likewise.
12726 (BNDmode): Likewise.
12727 (ADJUST_INSN_LENGTH): Likewise.
12728 * config/i386/i386.md: Likewise.
12729 * config/i386/i386.opt: Likewise.
12730 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
12731 (defined): Likewise.
12732 (LINK_MPX): Likewise.
12733 (MPX_SPEC): Likewise.
12734 (LIBMPX_SPEC): Likewise.
12735 (LIBMPXWRAPPERS_SPEC): Likewise.
12736 (CHKP_SPEC): Likewise.
12737 * config/i386/predicates.md: Likewise.
12738 * dbxout.c (dbxout_type): Likewise.
12739 * doc/extend.texi: Likewise.
12740 * doc/invoke.texi: Likewise.
12741 * doc/md.texi: Likewise.
12742 * doc/tm.texi: Likewise.
12743 * doc/tm.texi.in: Likewise.
12744 * dwarf2out.c (is_base_type): Likewise.
12745 (gen_formal_types_die): Likewise.
12746 (gen_subprogram_die): Likewise.
12747 (gen_type_die_with_usage): Likewise.
12748 (gen_decl_die): Likewise.
12749 (dwarf2out_late_global_decl): Likewise.
12750 * expr.c (expand_assignment): Likewise.
12751 (emit_storent_insn): Likewise.
12752 (store_expr_with_bounds): Likewise.
12753 (store_expr): Likewise.
12754 (expand_expr_real_1): Likewise.
12755 * expr.h (store_expr_with_bounds): Likewise.
12756 * function.c (use_register_for_decl): Likewise.
12757 (struct bounds_parm_data): Likewise.
12758 (assign_parms_augmented_arg_list): Likewise.
12759 (assign_parm_find_entry_rtl): Likewise.
12760 (assign_parm_is_stack_parm): Likewise.
12761 (assign_parm_load_bounds): Likewise.
12762 (assign_bounds): Likewise.
12763 (assign_parms): Likewise.
12764 (expand_function_start): Likewise.
12765 * gcc.c (CHKP_SPEC): Likewise.
12766 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
12767 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
12768 (wrestrict_dom_walker::check_call): Likewise.
12769 * gimple.c (gimple_build_call_from_tree): Likewise.
12770 * gimple.h (enum gf_mask): Likewise.
12771 (gimple_call_with_bounds_p): Likewise.
12772 (gimple_call_set_with_bounds): Likewise.
12773 * gimplify.c (gimplify_init_constructor): Likewise.
12774 * ipa-cp.c (initialize_node_lattices): Likewise.
12775 (propagate_constants_across_call): Likewise.
12776 (find_more_scalar_values_for_callers_subset): Likewise.
12777 * ipa-hsa.c (process_hsa_functions): Likewise.
12778 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
12779 * ipa-icf.c (sem_function::merge): Likewise.
12780 * ipa-inline.c (early_inliner): Likewise.
12781 * ipa-pure-const.c (warn_function_noreturn): Likewise.
12782 (warn_function_cold): Likewise.
12783 (propagate_pure_const): Likewise.
12784 * ipa-ref.h (enum GTY): Likewise.
12785 * ipa-split.c (find_retbnd): Likewise.
12786 (consider_split): Likewise.
12787 (split_function): Likewise.
12788 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
12789 * ipa.c (walk_polymorphic_call_targets): Likewise.
12790 (symbol_table::remove_unreachable_nodes): Likewise.
12791 (process_references): Likewise.
12792 (cgraph_build_static_cdtor_1): Likewise.
12793 * lto-cgraph.c (lto_output_node): Likewise.
12794 (output_refs): Likewise.
12795 (compute_ltrans_boundary): Likewise.
12796 (input_overwrite_node): Likewise.
12797 (input_node): Likewise.
12798 (input_cgraph_1): Likewise.
12799 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
12800 * passes.c (pass_manager::execute_early_local_passes): Likewise.
12801 (class pass_chkp_instrumentation_passes): Likewise.
12802 (make_pass_chkp_instrumentation_passes): Likewise.
12803 * passes.def: Likewise.
12804 * rtl.h (struct GTY): Likewise.
12805 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
12806 * stor-layout.c (layout_type): Likewise.
12807 * symtab.c: Likewise.
12808 * target.def: Likewise.
12809 * targhooks.c (default_chkp_bound_type): Likewise.
12810 (default_chkp_bound_mode): Likewise.
12811 (default_builtin_chkp_function): Likewise.
12812 (default_chkp_function_value_bounds): Likewise.
12813 (default_chkp_make_bounds_constant): Likewise.
12814 (default_chkp_initialize_bounds): Likewise.
12815 * targhooks.h (default_chkp_bound_type): Likewise.
12816 (default_chkp_bound_mode): Likewise.
12817 (default_builtin_chkp_function): Likewise.
12818 (default_chkp_function_value_bounds): Likewise.
12819 (default_chkp_make_bounds_constant): Likewise.
12820 (default_chkp_initialize_bounds): Likewise.
12821 * toplev.c (compile_file): Likewise.
12822 (process_options): Likewise.
12823 * tree-core.h (DEF_BUILTIN): Likewise.
12824 (DEF_BUILTIN_CHKP): Likewise.
12825 * tree-inline.c (declare_return_variable): Likewise.
12826 (remap_gimple_stmt): Likewise.
12827 (copy_bb): Likewise.
12828 (initialize_inlined_parameters): Likewise.
12829 (expand_call_inline): Likewise.
12830 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
12831 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
12832 (make_pass_ipa_chkp_produce_thunks): Likewise.
12833 (make_pass_chkp): Likewise.
12834 (make_pass_chkp_opt): Likewise.
12835 (make_pass_chkp_instrumentation_passes): Likewise.
12836 * tree-pretty-print.c (dump_generic_node): Likewise.
12837 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
12838 * tree-ssa-dce.c (propagate_necessity): Likewise.
12839 (eliminate_unnecessary_stmts): Likewise.
12840 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
12841 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
12842 * tree-ssa-sccvn.h: Likewise.
12843 * tree-ssa-strlen.c (get_string_length): Likewise.
12844 (valid_builtin_call): Likewise.
12845 (adjust_last_stmt): Likewise.
12846 (handle_builtin_strchr): Likewise.
12847 (handle_builtin_strcpy): Likewise.
12848 (handle_builtin_stxncpy): Likewise.
12849 (handle_builtin_memcpy): Likewise.
12850 (handle_builtin_strcat): Likewise.
12851 (strlen_check_and_optimize_stmt): Likewise.
12852 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
12853 * tree-streamer-in.c: Likewise.
12854 * tree-streamer.c (record_common_node): Likewise.
12855 * tree.c (tree_code_size): Likewise.
12856 (wide_int_to_tree_1): Likewise.
12857 (type_contains_placeholder_1): Likewise.
12858 (build_common_tree_nodes): Likewise.
12859 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
12860 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
12861 (POINTER_BOUNDS_P): Likewise.
12862 (BOUNDED_TYPE_P): Likewise.
12863 (BOUNDED_P): Likewise.
12864 (CALL_WITH_BOUNDS_P): Likewise.
12865 (pointer_bounds_type_node): Likewise.
12866 * value-prof.c (gimple_ic): Likewise.
12867 * var-tracking.c (vt_add_function_parameters): Likewise.
12868 * varasm.c (make_decl_rtl): Likewise.
12869 (assemble_start_function): Likewise.
12870 (output_constant): Likewise.
12871 (maybe_assemble_visibility): Likewise.
12872 * varpool.c (ctor_for_folding): Likewise.
12873 * chkp-builtins.def: Remove.
12874 * ipa-chkp.c: Remove.
12875 * ipa-chkp.h: Remove.
12876 * rtl-chkp.c: Remove.
12877 * rtl-chkp.h: Remove.
12878 * tree-chkp-opt.c: Remove.
12879 * tree-chkp.c: Remove.
12880 * tree-chkp.h: Remove.
12882 2018-06-07 Carl Love <cel@us.ibm.com>
12884 * config/rs6000/vsx.md (vextract_fp_from_shorth,
12885 vextract_fp_from_shortl): Add BE support.
12887 2018-06-07 Paul Koning <ni1d@arrl.net>
12889 * compare-elim.c (try_merge_compare): Don't merge compare if
12890 address contains a side effect.
12891 (try_eliminate_compare): Likewise.
12893 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
12895 * config.gcc: Support "tremont".
12896 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
12897 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
12899 * config/i386/i386.c (m_TREMONT): Define.
12900 (processor_target_table): Add "tremont".
12901 (PTA_TREMONT): Define.
12902 (ix86_lea_outperforms): Add TARGET_TREMONT.
12903 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
12904 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
12905 and M_INTEL_GOLDMONT_PLUS.
12906 (fold_builtin_cpu): Add "tremont".
12907 (ix86_add_stmt_cost): Add TARGET_TREMONT.
12908 (ix86_option_override_internal): Add "tremont".
12909 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
12910 (processor_type): Add PROCESSOR_TREMONT.
12911 * config/i386/x86-tune.def: Add m_TREMONT.
12912 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
12914 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
12916 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
12917 symbol defined for msp430i* devices to be lower case.
12919 2018-06-07 Richard Biener <rguenther@suse.de>
12921 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
12922 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
12923 Properly wrap signed arithmetic if overflow wraps.
12925 2018-06-07 Jakub Jelinek <jakub@redhat.com>
12927 PR tree-optimization/69615
12928 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
12929 of a cast from a same precision integral SSA_NAME in a bb dominated
12930 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
12931 cast to utype if rhs2 has already a compatible type.
12933 2018-06-07 Richard Biener <rguenther@suse.de>
12935 PR tree-optimization/85935
12936 * graphite-scop-detection.c (find_params_in_bb): Analyze
12937 condition operands with respect to the correct loop. Assert
12938 the analysis doesn't fail.
12940 2018-06-04 Carl Love <cel@us.ibm.com>
12942 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
12943 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
12944 as it is slightly cheaper.
12945 (first_match_or_eos_index_<mode>):
12946 Calculate index using natural element order.
12947 (first_match_index_<mode>):
12948 Calculate index using natural element order.
12949 (first_match_or_eos_index_<mode>):
12950 Calculate index using natural order.
12951 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
12952 for BE and LE modes.
12953 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
12954 P9V_BUILTIN_VCLZLSBB_V16QI.
12955 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
12958 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
12960 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
12961 indentation and line wrap for many prototypes. Add missing
12962 @smallexample directives around block of prototypes for vec_xl and
12965 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
12967 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
12968 track if we pass or return IEEE 128-bit floating point.
12969 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
12970 C++ mangling that is compatible with GCC 8.1.
12971 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
12972 (init_cumulative_args): Note if we pass or return IEEE 128-bit
12973 floating point types.
12974 (rs6000_function_arg_advance_1): Likewise.
12975 (rs6000_mangle_type): Optionally generate mangled names that match
12976 what GCC 8.1 generated for IEEE 128-bit floating point types.
12977 (rs6000_globalize_decl_name): If we have an external function that
12978 passes or returns IEEE 128-bit types, generate a weak reference
12979 from the mangled name used in GCC 8.1 to the current mangled
12981 (rs6000_init_builtins): Make __ibm128 use the long double type if
12982 long double is IBM extended double. Make __float128 use the long
12983 double type if long double is IEEE 128-bit.
12986 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
12987 macro for __ibm128 built-in functions.
12988 (PACK_IF): Add __ibm128 pack/unpack functions.
12989 (UNPACK_IF): Likewise.
12990 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
12991 enable long double built-in functions if long double is IEEE
12992 128-bit floating point.
12993 (rs6000_invalid_builtin): Update long double built-in function
12995 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
12996 functions, adjust the built-in function to use the long double
12997 built-in function if __ibm128 and long double are the same type.
12998 * doc/extend.texi (PowerPC builtins): Update documention for
12999 __builtin_{,un}pack_longdouble. Add documentation for
13000 __builtin_{,un}pack_ibm128.
13002 2018-06-06 Jim Wilson <jimw@sifive.com>
13004 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
13005 (struct machine_function): New field interrupt_mode.
13006 (riscv_handle_type_attribute): New function. Add forward declaration.
13007 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
13008 (riscv_expand_epilogue): Check interrupt_mode field.
13009 (riscv_set_current_function): Check interrupt attribute args and
13010 set interrupt_mode field.
13011 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
13012 (riscv_sret, riscv_uret): New.
13013 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
13014 new arguments to interrupt attribute.
13016 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
13019 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
13020 Don't handle -mcpu=power8 if -mpower9-vector is also used.
13022 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
13024 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
13025 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
13026 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
13027 several redundant entries.
13029 2018-06-06 David Malcolm <dmalcolm@redhat.com>
13031 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
13032 type from "rtx" to "rtx_insn *".
13033 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
13034 for local "call_insn", removing cast.
13035 (ix86_expand_call): Likewise, introducing a "call_insn" local.
13037 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
13039 PR tree-optimization/86066
13040 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
13041 for BIT_INSERT_EXPR stores.
13043 2018-06-06 Richard Biener <rguenther@suse.de>
13045 PR tree-optimization/86062
13046 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
13047 component refs ontop
13048 of to be offsetted base.
13050 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
13052 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
13053 to be static and remove check on interrupt attribute name.
13055 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
13057 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
13058 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
13060 2018-06-05 Steve Ellcey <sellcey@cavium.com>
13063 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
13065 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
13066 Remove second argument, change how error is called.
13067 (aarch64_layout_arg): Remove second argument from
13068 aarch64_err_no_fpadvsimd call.
13069 (aarch64_init_cumulative_args): Ditto.
13070 (aarch64_gimplify_va_arg_expr): Ditto.
13071 * config/aarch64/aarch64.md (mov<mode>): Ditto.
13073 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
13075 * config/i386/i386.md (simple_return_indirect_internal): New expander.
13076 (*simple_return_indirect_internal<mode>): Rename from
13077 simple_return_indirect_internal. Use W mode iterator.
13078 (rstorssp): New expander.
13079 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
13080 (clrssbsy): New expander.
13081 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
13083 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
13085 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
13087 (cmse_check_pointed_object): Likewise.
13089 2018-06-05 Martin Liska <mliska@suse.cz>
13091 PR gcov-profile/47618
13092 * doc/invoke.texi: Document how -fprofile-dir format
13095 2018-06-05 Richard Biener <rguenther@suse.de>
13097 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
13098 removal pretend DOM info isn't available so we do not update
13099 it and only remove edges, not dominated blocks. Actually free
13100 DOM info in case we removed something. Remove unreachable blocks.
13101 (mfb_keep_latches): Work with either DOM info or marked backedges.
13102 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
13103 first. Mark backedges if DOM info isn't available.
13104 (Re-)compute DOM info after cleanup_control_flow_pre.
13106 2018-06-05 Richard Biener <rguenther@suse.de>
13108 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
13109 (locus_discrim_hasher::hash): Adjust.
13110 (locus_discrim_hasher::equal): Likewise.
13111 (next_discriminator_for_locus): Work on line directly.
13112 (same_line_p): Pass in expanded locus1 as well.
13113 (assign_discriminators): Avoid redundant location expansions.
13115 2018-06-05 Richard Biener <rguenther@suse.de>
13117 PR tree-optimization/86046
13118 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
13119 if required after clearing TREE_ADDRESSABLE.
13121 2018-06-05 Richard Biener <rguenther@suse.de>
13123 PR tree-optimization/86047
13124 * tree-ssa-loop.c (for_each_index): Glob handling of all
13125 decls and constants and really handle all of them.
13127 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13130 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
13131 qualifier_void_pointer and qualifier_const_void_pointer.
13132 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
13133 (arm_init_builtins): Handle the above.
13134 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
13135 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
13136 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
13139 2018-06-05 Martin Liska <mliska@suse.cz>
13141 * auto-profile.c (read_autofdo_file): Do not use
13142 gcov_ctr_summary struct.
13143 (afdo_callsite_hot_enough_for_early_inline): Likewise.
13144 * coverage.c (struct counts_entry): Likewise.
13145 (read_counts_file): Read just single summary entry.
13146 (get_coverage_counts): Use gcov_summary struct.
13147 * coverage.h (get_coverage_counts): Likewise.
13148 * gcov-dump.c (dump_working_sets): Likewise.
13149 (tag_summary): Dump just single summary.
13150 * gcov-io.c (gcov_write_summary): Write just histogram
13152 (gcov_read_summary): Read just single summary.
13153 (compute_working_sets): Use gcov_summary struct.
13154 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
13155 of GCOV_COUNTERS_SUMMABLE.
13156 (GCOV_COUNTERS_SUMMABLE): Remove.
13157 (GCOV_FIRST_VALUE_COUNTER): Replace with
13158 GCOV_COUNTER_V_INTERVAL.
13159 (struct gcov_ctr_summary): Remove.
13160 (struct gcov_summary): Directly use fields of former
13162 (compute_working_sets): Use gcov_summary struct.
13163 * gcov.c (read_count_file): Do not use ctrs fields.
13164 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
13166 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
13168 * profile.c: Likewise.
13169 * profile.h: Likewise.
13171 2018-06-05 Martin Liska <mliska@suse.cz>
13173 PR gcov-profile/84846
13174 * gcov.c (output_lines): Print working directory only
13175 in intermediate format.
13177 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
13179 * config/s390/s390-builtin-types.def: Add void function type.
13180 * config/s390/s390-builtins.def: Use the function type for the
13183 2018-06-04 Jim Wilson <jimw@sifive.com>
13185 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
13187 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
13188 and maybe_eh_return. Change regno to unsigned int. Use new args to
13189 handle EH_RETURN_DATA_REGNO registers properly.
13190 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
13191 (riscv_expand_epilogue): Update comment. Change argument name and
13192 type. Update code to use new name and type. Pass new args to
13193 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
13195 * config/riscv/riscv.md (NORMAL_RETURN): New.
13196 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
13197 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
13198 (eh_return): Call gen_eh_return_internal and emit barrier.
13199 (eh_return_internal): Call riscv_expand_epilogue.
13201 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
13203 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
13204 bit_insertion field and declare can_be_merged_into method.
13205 (merged_store_group::can_be_merged_into): New method.
13206 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
13207 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
13208 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
13210 2018-06-04 Richard Biener <rguenther@suse.de>
13212 PR tree-optimization/85955
13213 * builtins.c (fold_builtin_sincos): Convert pointers to
13214 destination to appropriate type before dereferencing.
13216 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
13218 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
13220 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
13222 * expr.c (expand_expr_real_1): Force the operand into memory if
13223 its TYPE_MODE is BLKmode and if there is no integer mode for
13224 the number of bits being extracted.
13226 2018-06-04 Jakub Jelinek <jakub@redhat.com>
13230 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
13231 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
13233 2018-06-04 Richard Biener <rguenther@suse.de>
13235 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
13236 (cleanup_tree_cfg_noloop): ... single caller. Do
13237 start_recording_case_labels later.
13239 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
13241 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
13242 to _IMMINTRIN_H_INCLUDED.
13243 * config/i386/pconfigintrin.h: Ditto.
13244 * config/i386/waitpkgintrin.h: Ditto.
13245 * config/i386/immintrin.h: Add includes for sgxintrin.h,
13246 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
13247 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
13248 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
13249 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
13250 waitpkgintrin.h and cldemoteintrin.h.
13252 2018-06-04 Richard Biener <rguenther@suse.de>
13254 PR tree-optimization/86038
13255 * tracer.c (find_best_successor): Check probability for
13256 being initialized, bail out if not.
13258 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
13261 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
13262 of bits to ignore when comparing architectures.
13264 2018-06-04 Jakub Jelinek <jakub@redhat.com>
13266 PR tree-optimization/69615
13267 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
13268 maximum or minimum of the type, try to merge it also as if
13269 range1 is + [-, x - 1] or + [x + 1, -].
13272 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
13274 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
13276 PR tree-optimization/86034
13277 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
13278 the unsigned bitfield type in a bit insertion sequence if it does not
13279 have a larger precision than the bitfield size.
13280 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
13282 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
13284 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
13286 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
13288 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
13289 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
13290 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
13291 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
13293 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
13295 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
13296 Disable -fdelete-null-pointer-checks for ELF toolchain.
13298 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
13299 Kito Cheng <kito.cheng@gmail.com>
13301 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
13302 (nds32le-*-*, nds32be-*-*): Integrate checking process.
13303 (nds32*-*-*): Add glibc and uclibc conditions.
13304 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
13305 (TARGET_EXCEPT_UNWIND_INFO): Define.
13306 * config/nds32/elf.h: New file.
13307 * config/nds32/linux.h: New file.
13308 * config/nds32/nds32-elf.opt: New file.
13309 * config/nds32/nds32-linux.opt: New file.
13310 * config/nds32/nds32-fp-as-gp.c
13311 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
13312 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
13314 (nds32_asm_file_end): Ditto.
13315 (nds32_print_operand): Ditto.
13316 (nds32_insert_attributes): Ditto.
13317 (nds32_init_libfuncs): New function.
13318 (TARGET_HAVE_TLS): Define.
13319 (TARGET_INIT_LIBFUNCS): Define.
13320 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
13322 (TARGET_ELF): Apply different mcmodel setting.
13323 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
13324 been migrated into elf.h and linux.h files.
13325 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
13326 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
13327 (mcmodel): The content has been migrated into nds32-elf.opt and
13328 nds32-linux.opt files.
13329 * config/nds32/t-elf: New file.
13330 * config/nds32/t-linux: New file.
13332 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
13333 Shiva Chen <shiva0217@gmail.com>
13335 * config/nds32/constants.md (unspec_volatile_element): Add
13336 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
13337 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
13339 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
13340 (make_pass_nds32_fp_as_gp): Declare.
13341 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
13343 (nds32_asm_function_end_prologue): Remove unused asm output.
13344 (nds32_asm_function_begin_epilogue): Remove unused asm output.
13345 (nds32_asm_file_start): Output necessary fp_as_gp information.
13346 (nds32_option_override): Adjust register usage.
13347 (nds32_expand_prologue): Consider fp_as_gp situation.
13348 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
13349 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
13352 (simple_return): Ditto.
13353 (omit_fp_begin): Output special directive for fp_as_gp.
13354 (omit_fp_end): Output special directive for fp_as_gp.
13355 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
13356 mforbid-fp-as-gp): New options.
13358 2018-06-01 Mark Wielaard <mark@klomp.org>
13360 * dwarf2out.c (dwarf2out_finish): Remove generation of
13361 DW_AT_loclists_base.
13363 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
13365 * gimple-ssa-store-merging.c: Include gimple-fold.h.
13366 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
13367 (struct merged_store_group): Add bit_insertion field.
13368 (dump_char_array): Use standard hexadecimal format.
13369 (merged_store_group::merged_store_group): Set bit_insertion to false.
13370 (merged_store_group::apply_stores): Use optimal buffer size. Deal
13371 with BIT_INSERT_EXPR stores. Move up code updating the mask and
13372 also print the mask in the dump file.
13373 (pass_store_merging::gate): Minor tweak.
13374 (imm_store_chain_info::coalesce_immediate): Fix wrong association
13375 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
13376 stores with INTEGER_CST stores.
13377 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
13378 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
13379 and use it throughout. Generate bit insertion sequences if need be.
13380 (pass_store_merging::process_store): Remove redundant condition.
13381 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
13383 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
13385 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
13386 the 128-bit floating point types. Fix function comment.
13388 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13390 * config/aarch64/aarch64-simd.md
13391 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
13393 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
13396 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
13398 PR tree-optimization/85989
13399 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
13401 (backprop::intersect_uses): Check it when deciding whether this
13402 is a backedge reference.
13403 (backprop::process_block): Add each phi to m_visited_phis
13404 after visiting it, then clear it at the end.
13406 2018-06-01 Richard Biener <rguenther@suse.de>
13408 * tree-vectorizer.h (vect_dr_stmt): New function.
13409 (vect_get_load_cost): Adjust.
13410 (vect_get_store_cost): Likewise.
13411 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
13412 Use vect_dr_stmt instead of DR_SMTT.
13413 (vect_record_base_alignments): Likewise.
13414 (vect_calculate_target_alignment): Likewise.
13415 (vect_compute_data_ref_alignment): Likewise and make static.
13416 (vect_update_misalignment_for_peel): Likewise.
13417 (vect_verify_datarefs_alignment): Likewise.
13418 (vector_alignment_reachable_p): Likewise.
13419 (vect_get_data_access_cost): Likewise. Pass down
13420 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
13421 (vect_get_peeling_costs_all_drs): Likewise.
13422 (vect_peeling_hash_get_lowest_cost): Likewise.
13423 (vect_enhance_data_refs_alignment): Likewise.
13424 (vect_find_same_alignment_drs): Likewise.
13425 (vect_analyze_data_refs_alignment): Likewise.
13426 (vect_analyze_group_access_1): Likewise.
13427 (vect_analyze_group_access): Likewise.
13428 (vect_analyze_data_ref_access): Likewise.
13429 (vect_analyze_data_ref_accesses): Likewise.
13430 (vect_vfa_segment_size): Likewise.
13431 (vect_small_gap_p): Likewise.
13432 (vectorizable_with_step_bound_p): Likewise.
13433 (vect_prune_runtime_alias_test_list): Likewise.
13434 (vect_analyze_data_refs): Likewise.
13435 (vect_supportable_dr_alignment): Likewise.
13436 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
13437 (vect_gen_prolog_loop_niters): Likewise.
13438 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
13439 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
13441 (vect_recog_mask_conversion_pattern): Likewise.
13442 (vect_try_gather_scatter_pattern): Likewise.
13443 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
13444 to vect_get_store_cost.
13445 (vect_get_store_cost): Get stmt_info instead of DR.
13446 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
13447 (vect_get_load_cost): Get stmt_info instead of DR.
13449 2018-06-01 Richard Biener <rguenther@suse.de>
13451 PR middle-end/86017
13452 * gimple-fold.c (var_decl_component_p): Also allow offsetted
13453 vars wrapped in MEM_REFs.
13455 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
13457 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
13458 Fix subreg tests so that we only return a choice between
13459 GENERAL_REGS and FP_REGS if the original classes included both.
13461 2018-06-01 Richard Biener <rguenther@suse.de>
13464 * tree-ssa-structalias.c (get_function_part_constraint):
13465 Handle NULL fi->decl.
13466 (find_func_aliases_for_call): Properly handle indirect
13467 fi from direct call.
13468 (find_func_clobbers): Likewise.
13469 (ipa_pta_execute): Likewise.
13470 (create_variable_info_for): For functions that are ifunc_resolver
13471 resolve to a varinfo that contains the result of the resolver call.
13472 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
13475 2018-05-31 Michael Collison <michael.collison@arm.com>
13477 * config/aarch64/aarch64.md:
13478 (*fix_to_zero_extenddfdi2): New pattern.
13479 * gcc.target/aarch64/fix_extend1.c: New testcase.
13481 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
13483 PR middle-end/78809
13484 PR middle-end/83026
13485 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
13486 and BUILT_IN_STRNCMP_EQ.
13487 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
13488 BUILT_IN_STRNCMP_EQ.
13489 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
13490 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
13491 (gimple_fold_builtin): Likewise.
13492 * tree-ssa-strlen.c (compute_string_length): New function.
13493 (determine_min_obsize): New function.
13494 (handle_builtin_string_cmp): New function to handle calls to
13495 string compare functions.
13496 (strlen_optimize_stmt): Add handling to builtin string compare
13498 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
13499 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
13500 * tree.c (build_common_builtin_nodes): Add new defines of
13501 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
13503 2018-05-31 Jakub Jelinek <jakub@redhat.com>
13506 * bb-reorder.c (pass_partition_blocks::gate): Return false for
13507 functions with naked attribute.
13509 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
13511 * config/i386/sse.md (avx_vec_concat<mode>):
13512 Substitute concat_tg_mode mode attribute with xtg_mode.
13513 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
13514 (concat_tg_mode): Remove mode attribute.
13516 2018-05-31 Martin Sebor <msebor@redhat.com>
13519 * calls.c (alloc_max_size): Correct a logic error/typo.
13520 Treat excessive arguments as infinite. Warn for invalid arguments.
13521 * doc/invoke.texi (-Walloc-size-larger-than): Update.
13523 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
13526 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
13527 and movx for Haswell.
13529 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
13530 Cesar Philippidis <cesar@codesourcery.com>
13532 PR middle-end/85879
13533 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
13534 when emitting error on private/firstprivate reductions.
13535 * omp-low.c (lower_omp_target): Avoid reference-type processing
13536 on pointers for firstprivate clause.
13538 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
13540 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
13543 * config/aarch64/aarch64-simd.md
13544 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
13545 (aarch64_ld1_x3_<mode>): Likewise
13546 (aarch64_st1x2<VALLDIF:mode>): Likewise
13547 (aarch64_st1_x2_<mode>): Likewise
13548 (aarch64_st1x3<VALLDIF:mode>): Likewise
13549 (aarch64_st1_x3_<mode>): Likewise
13550 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
13551 (vld1_s8_x3): Likewise.
13552 (vld1_u16_x3): Likewise.
13553 (vld1_s16_x3): Likewise.
13554 (vld1_u32_x3): Likewise.
13555 (vld1_s32_x3): Likewise.
13556 (vld1_u64_x3): Likewise.
13557 (vld1_s64_x3): Likewise.
13558 (vld1_f16_x3): Likewise.
13559 (vld1_f32_x3): Likewise.
13560 (vld1_f64_x3): Likewise.
13561 (vld1_p8_x3): Likewise.
13562 (vld1_p16_x3): Likewise.
13563 (vld1_p64_x3): Likewise.
13564 (vld1q_u8_x3): Likewise.
13565 (vld1q_s8_x3): Likewise.
13566 (vld1q_u16_x3): Likewise.
13567 (vld1q_s16_x3): Likewise.
13568 (vld1q_u32_x3): Likewise.
13569 (vld1q_s32_x3): Likewise.
13570 (vld1q_u64_x3): Likewise.
13571 (vld1q_s64_x3): Likewise.
13572 (vld1q_f16_x3): Likewise.
13573 (vld1q_f32_x3): Likewise.
13574 (vld1q_f64_x3): Likewise.
13575 (vld1q_p8_x3): Likewise.
13576 (vld1q_p16_x3): Likewise.
13577 (vld1q_p64_x3): Likewise.
13578 (vst1_s64_x2): Likewise.
13579 (vst1_u64_x2): Likewise.
13580 (vst1_f64_x2): Likewise.
13581 (vst1_s8_x2): Likewise.
13582 (vst1_p8_x2): Likewise.
13583 (vst1_s16_x2): Likewise.
13584 (vst1_p16_x2): Likewise.
13585 (vst1_s32_x2): Likewise.
13586 (vst1_u8_x2): Likewise.
13587 (vst1_u16_x2): Likewise.
13588 (vst1_u32_x2): Likewise.
13589 (vst1_f16_x2): Likewise.
13590 (vst1_f32_x2): Likewise.
13591 (vst1_p64_x2): Likewise.
13592 (vst1q_s8_x2): Likewise.
13593 (vst1q_p8_x2): Likewise.
13594 (vst1q_s16_x2): Likewise.
13595 (vst1q_p16_x2): Likewise.
13596 (vst1q_s32_x2): Likewise.
13597 (vst1q_s64_x2): Likewise.
13598 (vst1q_u8_x2): Likewise.
13599 (vst1q_u16_x2): Likewise.
13600 (vst1q_u32_x2): Likewise.
13601 (vst1q_u64_x2): Likewise.
13602 (vst1q_f16_x2): Likewise.
13603 (vst1q_f32_x2): Likewise.
13604 (vst1q_f64_x2): Likewise.
13605 (vst1q_p64_x2): Likewise.
13606 (vst1_s64_x3): Likewise.
13607 (vst1_u64_x3): Likewise.
13608 (vst1_f64_x3): Likewise.
13609 (vst1_s8_x3): Likewise.
13610 (vst1_p8_x3): Likewise.
13611 (vst1_s16_x3): Likewise.
13612 (vst1_p16_x3): Likewise.
13613 (vst1_s32_x3): Likewise.
13614 (vst1_u8_x3): Likewise.
13615 (vst1_u16_x3): Likewise.
13616 (vst1_u32_x3): Likewise.
13617 (vst1_f16_x3): Likewise.
13618 (vst1_f32_x3): Likewise.
13619 (vst1_p64_x3): Likewise.
13620 (vst1q_s8_x3): Likewise.
13621 (vst1q_p8_x3): Likewise.
13622 (vst1q_s16_x3): Likewise.
13623 (vst1q_p16_x3): Likewise.
13624 (vst1q_s32_x3): Likewise.
13625 (vst1q_s64_x3): Likewise.
13626 (vst1q_u8_x3): Likewise.
13627 (vst1q_u16_x3): Likewise.
13628 (vst1q_u32_x3): Likewise.
13629 (vst1q_u64_x3): Likewise.
13630 (vst1q_f16_x3): Likewise.
13631 (vst1q_f32_x3): Likewise.
13632 (vst1q_f64_x3): Likewise.
13633 (vst1q_p64_x3): Likewise.
13635 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
13637 * config/msp430/msp430.c (msp430_output_labelref): Prepend
13638 user_label_prefix to name.
13640 * tree-core.h: Update comment about the format of NAME string
13641 passed to handler in attribute_spec.
13643 * config/msp430/msp430.md: Remove erroneous subreg expression from
13644 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
13645 zero_extend{q,h}isi2.
13647 2018-05-30 Borislav Petkov <bp@suse.de>
13649 * doc/extend.texi: Document some architecture specific
13650 constraints and sort entries.
13652 2018-05-30 Martin Sebor <msebor@redhat.com>
13654 PR middle-end/85369
13655 * builtins.c (expand_builtin_stpcpy_1): New function.
13656 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
13657 only if the former succeeds.
13659 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
13661 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
13664 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
13666 * doc/invoke.texi (-flinker-output): Document
13668 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
13670 * passes.c (ipa_write_summaries): Only modify statements if body
13672 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
13673 incrementally linking.
13674 (ipa_passes): Likewise.
13675 * lto-cgraph.c (lto_output_node): When incrementally linking do not
13676 pass down resolution info.
13677 * common.opt (flag_incremental_link): Update info.
13678 * gcc.c (plugin specs): Turn flinker-output=* to
13679 -plugin-opt=-linker-output-known
13680 * toplev.c (compile_file): Also cut compilation when doing incremental
13682 * flag-types. (enum lto_partition_model): Add
13683 LTO_LINKER_OUTPUT_NOLTOREL.
13684 (invoke.texi): Add -flinker-output docs.
13685 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
13686 link same way as WPA; do not stream in dead initializers.
13688 * dwarf2out.c (dwarf2out_die_ref_for_decl,
13689 darf2out_register_external_decl): Support incremental link.
13691 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
13693 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
13695 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
13697 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
13698 it down to simple_object_copy_lto_debug_sections.
13699 (run_gcc): Determine incremental LTO link time and configure
13700 lto1 into non-wpa mode, disable renaming of debug sections.
13702 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
13704 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
13705 descriptions of various incorrectly documented functions.
13707 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
13710 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
13711 address check not strict.
13713 2018-05-30 Richard Biener <rguenther@suse.de>
13715 PR tree-optimization/85964
13716 * tracer.c (better_p): Drop initialized count check, we only
13717 call the function with initialized counts now.
13718 (find_best_successor): Do find a best edge if one
13719 has uninitialized count.
13720 (find_best_predecessor): Likewise. Do BB frequency check only
13721 if count is initialized.
13723 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
13725 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
13726 (aarch64_ldrstr_offset_compare): New.
13727 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
13728 load/store orderings.
13729 (aarch64_gen_adjusted_ldpstp): Likewise.
13731 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
13733 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
13734 Check for subset of GENERAL_REGS and FP_REGS.
13735 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
13738 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
13740 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
13741 and wi::to_poly_offset. Add the current offset and then check
13742 whether the sum fits, rather than using an unchecked addition of
13743 a checked term. Check for a shwi rather than a uhwi.
13744 * expr.c (get_bit_range): Use tree_to_poly_uint64.
13745 (store_constructor): Use poly_int_tree_p.
13746 (expand_expr_real_1): Likewise.
13747 * function.c (assign_temp): Likewise.
13748 * fold-const.c (const_binop): Use poly_int_tree_p and
13749 wi::to_poly_offset.
13750 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
13752 * ipa-icf-gimple.c (func_checker::compare_operand): Use
13753 to_poly_offset for MEM offsets.
13754 * ipa-icf.c (sem_variable::equals): Likewise.
13755 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
13756 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
13757 wi::to_poly_offset for BIT_FIELD_REF offsets.
13758 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
13759 wi::to_poly_offset.
13760 * var-tracking.c (emit_note_insn_var_location): Use
13761 tree_to_poly_uint64.
13763 2018-05-29 Jim Wilson <jimw@sifive.com>
13765 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
13767 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
13770 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
13771 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
13773 (sse4_1_round<mode>2): Use nonimmediate_operand
13774 for operand 1 predicate.
13776 2018-05-29 Martin Sebor <msebor@redhat.com>
13777 Richard Biener <rguenther@suse.de>
13780 * calls.c (get_size_range): Call determine_value_range instead
13781 of get_value_range..
13782 * tree-vrp.h (determine_value_range): Declared new function.
13783 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
13785 2018-05-29 Richard Biener <rguenther@suse.de>
13787 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
13788 sure to use non-pattern stmts for get_earlier_stmt arguments.
13789 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
13790 called on pattern stmts.
13791 (get_later_stmt): Likewise.
13793 2018-05-29 Martin Liska <mliska@suse.cz>
13795 PR gcov-profile/85759
13796 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
13799 2018-05-29 Jakub Jelinek <jakub@redhat.com>
13801 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
13803 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
13807 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
13808 VEC_PACK_FLOAT_EXPR): New tree codes.
13809 * tree-pretty-print.c (op_code_prio): Handle
13810 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
13811 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
13812 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
13813 * tree-inline.c (estimate_operator_cost): Likewise.
13814 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
13815 * fold-const.c (const_binop): Likewise.
13816 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
13817 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
13818 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
13819 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
13820 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
13821 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
13822 * expr.c (expand_expr_real_2): Likewise.
13823 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
13824 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
13825 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
13827 * optabs.c (expand_widen_pattern_expr): For
13828 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
13829 sign from result type rather than operand's type.
13830 (expand_binop_directly): For vec_packu_float_optab and
13831 vec_packs_float_optab allow result type to be different from operand's
13833 * optabs-tree.c (optab_for_tree_code): Handle
13834 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
13835 VEC_PACK_FLOAT_EXPR. Formatting fixes.
13836 * tree-vect-generic.c (expand_vector_operations_1): Handle
13837 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
13838 VEC_PACK_FLOAT_EXPR.
13839 * tree-vect-stmts.c (supportable_widening_operation): Handle
13841 (supportable_narrowing_operation): Handle FLOAT_EXPR.
13842 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
13843 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
13844 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
13845 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
13847 (vec_pack<floatprefix>_float_<mode>): New expander.
13848 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
13850 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
13851 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
13852 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
13853 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
13854 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
13856 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
13857 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
13858 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
13859 VEC_PACK_FLOAT_EXPR): Document.
13861 2018-05-29 Richard Biener <rguenther@suse.de>
13863 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
13865 (stmt_vec_info_vec): Make pointer.
13866 (init_stmt_vec_info_vec): Remove.
13867 (free_stmt_vec_info_vec): Likewise.
13868 (set_stmt_vec_info_vec): New function.
13869 (free_stmt_vec_infos): Likewise.
13870 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
13871 (set_vinfo_for_stmt): Likewise.
13872 (get_earlier_stmt): Likewise.
13873 (get_later_stmt): Likewise.
13874 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
13875 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
13876 (vec_info::~vec_info): Free stmt_vec_infos.
13877 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
13878 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
13879 (pass_slp_vectorize::execute): Likewise.
13880 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
13881 (free_stmt_vec_info_vec): Likewise.
13882 (set_stmt_vec_info_vec): New function.
13883 (free_stmt_vec_infos): Likewise.
13884 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
13885 the global stmt_vec_info_vec.
13886 * tree-parloops.c (gather_scalar_reductions): Use
13887 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
13890 2018-05-29 Richard Biener <rguenther@suse.de>
13892 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
13894 2018-05-29 Martin Liska <mliska@suse.cz>
13895 David Malcolm <dmalcolm@redhat.com>
13897 * vec.c (test_reverse): New.
13898 (vec_c_tests): Add new test.
13899 * vec.h (vl_ptr>::reverse): New function.
13901 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
13903 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
13905 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
13908 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
13910 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
13912 2018-05-28 Richard Biener <rguenther@suse.de>
13914 PR tree-optimization/85933
13915 * tree-vect-data-refs.c (vect_record_base_alignments): Only
13916 look at stmts marked as vectorizable.
13918 2018-05-28 Richard Biener <rguenther@suse.de>
13920 PR tree-optimization/85934
13921 * tree-vect-generic.c (expand_vector_operations_1): Hoist
13922 vector boolean check before scalar optimization.
13924 2018-05-28 Jakub Jelinek <jakub@redhat.com>
13926 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
13929 2018-05-28 Mark Wielaard <mark@klomp.org>
13931 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
13932 if it is an expression containing a minus sign.
13934 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
13936 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
13938 2018-05-27 Paul Koning <ni1d@arrl.net>
13940 * config/pdp11/pdp11.md (truncsihi2): Remove.
13942 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
13943 Chung-Ju Wu <jasonwucj@gmail.com>
13945 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
13947 (unaligned_store_dw): Ditto.
13948 * config/nds32/nds32-memory-manipulation.c
13949 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
13950 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
13951 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
13952 (emit_setmem_word_loop): Rename to ...
13953 (emit_setmem_doubleword_loop): ... this.
13954 (nds32_gen_dup_4_byte_to_word_value): New function.
13955 (nds32_gen_dup_8_byte_to_double_word_value): New function.
13956 (nds32_expand_setmem_loop): Refine implementation.
13957 (nds32_expand_setmem_loop_v3m): Ditto.
13958 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
13961 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
13963 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
13965 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
13967 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
13968 (nds32_init_machine_status): Initialize machine->attr_naked_p and
13969 machine->attr_no_prologue_p.
13970 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
13971 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
13972 (nds32_expand_epilogue): Consider attr_naked_p.
13973 (nds32_expand_epilogue_v3pop): Likewise.
13974 (nds32_can_use_return_insn): Likewise.
13975 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
13976 attr_no_prologue_p fields.
13977 * config/nds32/nds32.opt (mret-in-naked-func): New option.
13979 2018-05-27 Jakub Jelinek <jakub@redhat.com>
13982 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
13984 * config/i386/sse.md
13985 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
13987 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
13989 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
13991 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
13993 (*<floatsuffix>floatv2div2sf2): Rename to ...
13994 (*float<floatunssuffix>v2div2sf2): ... this.
13995 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
13996 (float<floatunssuffix>v2div2sf2_mask): ... this.
13997 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
13998 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
13999 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
14001 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
14003 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
14005 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
14007 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
14009 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
14011 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
14012 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
14013 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
14014 gen_ufix_truncv8dfv8si2.
14015 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
14016 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
14017 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
14018 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
14019 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
14020 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
14021 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
14022 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
14024 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
14028 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
14030 2018-05-25 Jim Wilson <jimw@sifive.com>
14032 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
14033 * config/riscv/riscv.c (struct machine_function): Add
14034 interrupt_handler_p and attribute_checked_p fields.
14035 (riscv_attribute_table): Add interrupt.
14036 (riscv_interrupt_type_p): New.
14037 (riscv_save_reg_p): Save extra regs for interrupt handler.
14038 (riscv_use_save_libcall): Return false for interrupt handler.
14039 (riscv_first_stack_step): Add forward declaration.
14040 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
14041 for interrupt handler with large frame. Use it for saved reg list.
14042 (riscv_expand_prologue): Move flag_stack_usage_info support to
14043 eliminate duplication.
14044 (riscv_expand_epilogue): Generate mret for interrupt handler.
14045 (riscv_epilogue_uses): New.
14046 (riscv_can_use_return_insn): Return false for interrupt handler.
14047 (riscv_function_ok_for_sibcall): Likewise.
14048 (riscv_set_current_function): Add interrupt handler support.
14049 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
14050 * config/riscv/riscv.md (UNSPECV_MRET): New.
14052 (riscv_frflags, riscv_fsflags): Use tab after opcode.
14054 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
14056 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
14058 PR tree-optimization/85712
14059 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
14060 this candidate has already been replaced in-situ by a copy.
14062 2018-05-25 Jason Merrill <jason@redhat.com>
14064 PR c++/80485 - inline function non-zero address.
14065 * symtab.c (nonzero_address): Check DECL_COMDAT.
14067 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
14070 * config/alpha/alpha.md (ashlsi3): New insn pattern.
14071 (*ashlsi_se): Rename from *ashldi_se. Define as sign
14072 extension of SImode operation. Use const123_operand predicate.
14073 (*saddsi_1): Remove.
14074 (*saddl_se_1): Ditto.
14075 (*ssubsi_1): Ditto.
14076 (*ssubl_se_1): Ditto.
14077 * config/alpha/predicates.md (const123_operand): New predicate.
14078 * config/alpha/constraints.md (P): Use IN_RANGE.
14080 2018-05-25 Richard Biener <rguenther@suse.de>
14082 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
14084 (ref_maybe_used_by_stmt_p): Likewise.
14085 (stmt_may_clobber_ref_p): Likewise.
14086 (stmt_may_clobber_ref_p_1): Likewise.
14087 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
14089 (ref_maybe_used_by_stmt_p): Likewise.
14090 (stmt_may_clobber_ref_p): Likewise.
14091 (stmt_may_clobber_ref_p_1): Likewise.
14092 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
14093 the alias oracle to disambiguate DRs with stmts DR analysis
14095 (vect_analyze_data_refs): Do not give up on not analyzable
14096 DRs for BB vectorization. Remove code truncating the dataref
14099 2018-05-25 Jakub Jelinek <jakub@redhat.com>
14102 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
14103 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
14104 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
14106 2018-05-25 Richard Biener <rguenther@suse.de>
14108 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
14109 function, combining stmt data ref gathering and fatal analysis
14111 (vect_analyze_data_refs): Remove now redudnant code and simplify.
14112 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
14113 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
14114 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
14115 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
14117 2018-05-25 Bin Cheng <bin.cheng@arm.com>
14119 PR tree-optimization/85720
14120 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
14121 SCC if all partitions are builtins.
14122 (version_loop_by_alias_check): New parameter. Generate cancelable
14123 runtime alias check if all partitions are builtins.
14124 (distribute_loop): Update call to above function.
14126 2018-05-25 Bin Cheng <bin.cheng@arm.com>
14128 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
14129 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
14130 (parm_default_def_partition_arg): Ditto.
14131 (set_parm_default_def_partition): Ditto.
14132 (get_parm_default_def_partitions): Ditto and make it static.
14133 (get_undefined_value_partitions): Ditto and make it static.
14134 (remove_ssa_form): Refactor call to init_var_map here.
14135 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
14136 computation for loop region.
14137 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
14138 (register_default_def): Delete.
14139 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
14140 (parm_default_def_partition_arg): Ditto.
14141 (set_parm_default_def_partition): Ditto.
14142 (get_parm_default_def_partitions): Ditto and make it static.
14143 (get_undefined_value_partitions): Ditto and make it static.
14144 (coalesce_with_default, coalesce_with_default): Update comment.
14145 (create_coalesce_list_for_region): New func factored out from
14146 create_outofssa_var_map.
14147 (populate_coalesce_list_for_outofssa): New func factored out from
14148 create_outofssa_var_map and coalesce_ssa_name.
14149 (create_outofssa_var_map): Delete.
14150 (coalesce_ssa_name): Refactor to support live range computation.
14151 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
14152 (get_parm_default_def_partitions): Delete.
14153 (get_undefined_value_partitions): Ditto.
14154 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
14155 computation for loop region.
14156 (new_tree_live_info, loe_visit_block): Ditto.
14157 (live_worklist, set_var_live_on_entry): Ditto.
14158 (calculate_live_on_exit, verify_live_on_entry): Ditto.
14159 * tree-ssa-live.h (struct _var_map): New fields.
14160 (init_var_map): Change decl.
14161 (region_contains_p): New.
14163 2018-05-25 Bin Cheng <bin.cheng@arm.com>
14165 * tree-ssa-live.h (live_merge_and_clear): Delete.
14167 2018-05-25 Richard Biener <rguenther@suse.de>
14170 * tree-dump.c (dequeue_and_dump): Remove access to removed
14171 operand 2 of a SWITCH_EXPR.
14173 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
14175 * doc/sourcebuild.texi (vect_double_cond_arith): Include
14176 multiplication and division.
14177 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
14178 (cond_udiv@var{m}, cond_umod@var{m}): Document.
14179 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
14180 (cond_udiv_optab, cond_umod_optab): New optabs.
14181 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
14182 (IFN_COND_RDIV): New internal functions.
14183 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
14184 TRUNC_MOD_EXPR and RDIV_EXPR.
14185 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
14186 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
14188 (SVE_INT_BINARY): Include mult.
14189 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
14190 (optab, sve_int_op): Handle mult.
14191 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
14193 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
14194 for SVE_INT_BINARY_SD.
14196 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
14198 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
14199 (optab, sve_int_op): Handle div and udiv.
14200 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
14201 for SVE_INT_BINARY_SD.
14202 (*<optab><mode>3): New insn for the same.
14204 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
14206 * tree-vect-patterns.c: Include predict.h.
14207 (vect_recog_divmod_pattern): Restrict check for division support
14208 to when optimizing for size.
14210 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
14212 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
14213 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
14214 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
14215 (gimple_match_op::set_op): Likewise.
14216 (gimple_resimplify4): Declare.
14217 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
14218 (expr::gen_transform): Likewise.
14219 (decision_tree::gen): Generate a simplification routine for 4 operands.
14220 * gimple-match-head.c (gimple_simplify): Add an overload for
14221 4 operands. In the top-level function, handle up to 4 call
14222 arguments and call gimple_resimplify4.
14223 (gimple_resimplify4): New function.
14224 (build_call_internal): Pass a fourth operand.
14225 (maybe_push_to_seq): Likewise.
14226 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
14227 Fold VEC_COND_EXPRs of an operation and a default value into
14228 an IFN_COND_* function if possible.
14229 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
14231 (SVE_COND_FP_BINARY): Include them.
14232 (optab, sve_fp_op): Handle them.
14233 (SVE_INT_BINARY_REV): New code iterator.
14234 (SVE_COND_FP_BINARY_REV): New int iterator.
14235 (commutative): New int attribute.
14236 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
14238 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
14240 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
14241 (*cond_<optab><mode>): New patterns for reversed operands.
14243 2018-05-25 Richard Biener <rguenther@suse.de>
14245 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
14246 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
14247 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
14248 (STMT_VINFO_GROUPED_ACCESS): Adjust.
14249 * tree-vect-data-refs.c (everywhere): Adjust users.
14250 * tree-vect-loop.c (everywhere): Likewise.
14251 * tree-vect-slp.c (everywhere): Likewise.
14252 * tree-vect-stmts.c (everywhere): Likewise.
14253 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
14255 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14257 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
14259 (gcc_cv_as_section_exclude): ... this.
14260 Try Solaris as #exclude syntax.
14261 * configure: Regenerate.
14262 * config.in: Regenerate.
14263 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
14265 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
14266 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
14268 * varasm.c (default_elf_asm_named_section): Don't check if
14269 HAVE_GAS_SECTION_EXCLUDE is defined.
14271 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
14273 * doc/md.texi: Update the documentation of the cond_* optabs
14274 to mention the new final operand. Fix GET_MODE_NUNITS call.
14275 Describe the scalar case too.
14276 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
14277 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
14279 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
14280 (get_conditional_internal_fn): Update comment.
14281 * tree-vect-loop.c (vectorizable_reduction): Pass the original
14282 accumulator value as a final argument to conditional functions.
14283 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
14284 a define_expand and add an "else" operand. Assert for now that
14285 the else operand is equal to operand 2. Use SVE_INT_BINARY and
14286 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
14287 (*cond_<optab><mode>): New patterns.
14288 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
14289 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
14290 (UNSPEC_COND_EOR): Delete.
14291 (optab): Remove associated mappings.
14292 (SVE_INT_BINARY): New code iterator.
14293 (sve_int_op): Remove int attribute and add "minus" to the code
14295 (SVE_COND_INT_OP): Delete.
14296 (SVE_COND_FP_OP): Rename to...
14297 (SVE_COND_FP_BINARY): ...this.
14299 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
14301 * optabs.c (can_reuse_operands_p): New function.
14302 (maybe_legitimize_operands): Try to reuse the results for
14305 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
14307 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
14308 Add {q} suffix to insn mnemonic.
14310 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
14312 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
14313 (msp430_warn_func_return): New.
14315 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
14317 * fold-const.c (tree_nonzero_bits): New function.
14318 * fold-const.h (tree_nonzero_bits): Likewise.
14319 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
14320 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
14322 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
14326 * varasm.c (assemble_alias): Check ifunc_resolver only on
14329 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
14332 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
14333 when memory input operand is handled.
14335 2018-05-24 Luis Machado <luis.machado@linaro.org>
14337 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
14339 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
14341 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
14343 * match.pd: Delay FMA folds until after vectorization.
14345 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
14348 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
14349 address check not strict.
14351 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
14353 * gimple-match.h (gimple_match_op): New class.
14354 (mprts_hook): Replace parameters with a gimple_match_op *.
14355 (maybe_build_generic_op): Likewise.
14356 (gimple_simplified_result_is_gimple_val): Replace parameters with
14357 a const gimple_match_op *.
14358 (gimple_simplify): Replace code_helper * and tree * parameters with
14359 a gimple_match_op * parameter.
14360 (gimple_resimplify1): Replace code_helper *, tree and tree *
14361 parameters with a gimple_match_op * parameter.
14362 (gimple_resimplify2): Likewise.
14363 (gimple_resimplify3): Likewise.
14364 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
14365 parameters with a gimple_match_op * parameter.
14366 * gimple-match-head.c (gimple_simplify): Change prototypes of
14367 auto-generated functions to take a gimple_match_op * instead of
14368 separate code_helper * and tree * parameters. Make the same
14369 change in the top-level overload and update calls to the
14370 gimple_resimplify routines. Update calls to the auto-generated
14371 functions and to maybe_push_res_to_seq in the publicly-facing
14372 operation-specific gimple_simplify overloads.
14373 (gimple_match_op::MAX_NUM_OPS): Define.
14374 (gimple_resimplify1): Replace rcode and ops with a single res_op
14375 parameter. Update call to gimple_simplify.
14376 (gimple_resimplify2): Likewise.
14377 (gimple_resimplify3): Likewise.
14378 (mprts_hook): Replace parameters with a gimple_match_op *.
14379 (maybe_build_generic_op): Likewise.
14380 (build_call_internal): Replace type, nargs and ops with
14381 a gimple_match_op *.
14382 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
14383 with a single gimple_match_op *. Update calls to mprts_hook,
14384 build_call_internal and gimple_simplified_result_is_gimple_val.
14385 Factor out code that is common to the tree_code and combined_fn cases.
14386 * genmatch.c (expr::gen_transform): Replace tem_code and
14387 tem_ops with a gimple_match_op called tem_op. Update calls
14388 to the gimple_resimplify functions and maybe_push_res_to_seq.
14389 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
14390 res_ops. Update call to the gimple_resimplify functions.
14391 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
14392 (decision_tree::gen): Make the functions take a gimple_match_op *
14393 called res_op instead of separate res_code and res_ops parameters.
14394 Update call accordingly.
14395 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
14396 and ops with a single res_op parameter. Update calls to
14397 maybe_build_generic_op and maybe_push_res_to_seq.
14398 (fold_stmt_1): Update calls to gimple_simplify and
14399 replace_stmt_with_simplification.
14400 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
14401 and gimple_simplified_result_is_gimple_val.
14402 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
14404 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
14405 with a gimple_match_op *.
14406 (vn_nary_build_or_lookup): Likewise. Update call to
14407 vn_nary_build_or_lookup_1.
14408 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
14409 gimple_match_op *. Update calls to the gimple_resimplify routines
14410 and to gimple_simplified_result_is_gimple_val.
14411 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
14412 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
14413 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
14414 (visit_nary_op): Likewise.
14415 (visit_reference_op_load): Likewise.
14417 2018-05-23 Luis Machado <luis.machado@linaro.org>
14419 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
14420 modifier for printing the step amount.
14422 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14425 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
14428 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
14430 * doc/sourcebuild.texi (Endianness): New subsubsection.
14432 2018-05-23 Luis Machado <luis.machado@linaro.org>
14434 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
14435 <prefetch_dynamic_strides>: New const bool field.
14436 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
14437 prefetch_dynamic_strides.
14438 (exynosm1_prefetch_tune): Likewise.
14439 (thunderxt88_prefetch_tune): Likewise.
14440 (thunderx_prefetch_tune): Likewise.
14441 (thunderx2t99_prefetch_tune): Likewise.
14442 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
14444 (aarch64_override_options_internal): Update to set
14445 PARAM_PREFETCH_DYNAMIC_STRIDES.
14446 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
14447 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
14448 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
14449 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
14450 prefetch-dynamic-strides setting.
14452 2018-05-23 Luis Machado <luis.machado@linaro.org>
14454 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
14455 <minimum_stride>: New const int field.
14456 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
14457 minimum_stride field defaulting to -1.
14458 (exynosm1_prefetch_tune): Likewise.
14459 (thunderxt88_prefetch_tune): Likewise.
14460 (thunderx_prefetch_tune): Likewise.
14461 (thunderx2t99_prefetch_tune): Likewise.
14462 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
14463 <default_opt_level>: Set to 3.
14464 (aarch64_override_options_internal): Update to set
14465 PARAM_PREFETCH_MINIMUM_STRIDE.
14466 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
14467 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
14468 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
14469 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
14470 stride is constant and is below the minimum stride threshold.
14472 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14474 * config/arm/arm-cpus.in (mode26): Delete.
14475 (armv4): Delete mode26 reference.
14476 * config/arm/arm.c (arm_configure_build_target): Delete use of
14479 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
14481 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
14483 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
14484 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
14486 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
14487 (floatunsdidf2): Ditto.
14489 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
14491 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
14492 (fixuns_trunc<mode>si2_avx512f): Ditto.
14493 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
14494 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
14495 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
14497 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
14499 PR rtl-optimization/79985
14500 * df-scan.c (df_insn_refs_collect): Remove special case for
14501 global registers and asm statements.
14503 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
14505 * extend.texi (Global Register Variables): Rewrite the bullet list.
14506 Note that the register is available for allocation. Note that access
14507 via inline asm must use constraints. Add note about async-signal
14508 handlers. Remove paragraph about automagic register selection.
14510 2018-05-23 Richard Biener <rguenther@suse.de>
14512 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
14513 of fixed offset from memset VN.
14515 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
14517 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
14518 first_interp field.
14519 (alloc_cand_and_find_basis): Initialize first_interp field.
14520 (slsr_process_mul): Modify first_interp field.
14521 (slsr_process_add): Likewise.
14522 (slsr_process_cast): Modify first_interp field for each new
14524 (slsr_process_copy): Likewise.
14525 (dump_candidate): Dump first_interp field.
14526 (replace_mult_candidate): Process all interpretations, not just
14528 (replace_rhs_if_not_dup): Likewise.
14529 (replace_one_candidate): Likewise.
14531 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
14533 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
14535 (aarch64_needs_frame_chain): New function.
14536 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
14538 2018-05-23 Sudakshina Das <sudi.das@arm.com>
14541 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
14542 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
14543 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
14544 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
14545 as true for strict-align.
14546 (aarch64_can_inline_p): Perform checks even when callee has no
14547 attributes to check for strict alignment.
14548 * doc/extend.texi (AArch64 Function Attributes): Document
14550 * doc/invoke.texi: (AArch64 Options): Likewise.
14552 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
14554 PR tree-optimization/85853
14555 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
14556 the handling of the root of the node to...
14557 (vect_slp_analyze_node_operations_1): ...this new function,
14558 and run the whole thing with the child nodes' def types
14559 set according to their SLP node's def type.
14561 2018-05-23 Richard Biener <rguenther@suse.de>
14563 PR middle-end/85874
14564 * tree-data-ref.c (create_runtime_alias_checks): Defer
14565 and ignore overflow warnings.
14567 2018-05-23 Yury Gribov <tetra2005@gmail.com>
14569 PR tree-optimization/85822
14570 * tree-vrp.c (is_masked_range_test): Fix handling of negative
14573 2018-05-23 Richard Biener <rguenther@suse.de>
14575 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
14576 memset constants via native_interpret_expr.
14578 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
14581 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
14583 (cgraph_node::create_alias): Likewise.
14584 (cgraph_node::get_availability): Check ifunc_resolver instead
14585 of looking up ifunc attribute.
14586 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
14587 * varasm.c (do_assemble_alias): Likewise.
14588 (assemble_alias): Likewise.
14589 (default_binds_local_p_3): Likewise.
14590 * cgraph.h (cgraph_node): Add ifunc_resolver.
14591 (cgraph_node::only_called_directly_or_aliased_p): Return false
14592 for IFUNC resolver.
14593 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
14595 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
14596 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
14597 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
14598 instead of looking up ifunc attribute.
14600 2018-05-22 Luis Machado <luis.machado@linaro.org>
14602 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
14604 2018-05-22 Martin Sebor <msebor@redhat.com>
14606 PR middle-end/85359
14607 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
14608 only when expasion succeeds.
14609 (expand_builtin_strcmp): Same.
14610 (expand_builtin_strncmp): Same.
14612 2018-05-22 Martin Sebor <msebor@redhat.com>
14614 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
14616 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
14617 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14619 * config/aarch64/aarch64-ldpstp.md: Replace uses of
14620 aarch64_mem_pair_operand with memory_operand and delete operand swapping
14622 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
14623 Add check for legitimate_address.
14624 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
14625 (aarch64_swap_ldrstr_operands): New.
14626 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
14629 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
14630 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14632 * config/aarch64/aarch64.md: New patterns to generate stp
14634 (store_pair_sw, store_pair_dw): New patterns to generate stp for
14635 single words and double words.
14636 (load_pair_sw, load_pair_dw): Likewise.
14637 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
14639 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
14641 * config/aarch64/aarch64-ldpstp.md: Modify peephole
14642 for different mode ldpstp and add peephole for merged zero stores.
14643 Likewise for loads.
14644 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
14646 (aarch64_gen_store_pair): Rename calls to match new patterns.
14647 (aarch64_gen_load_pair): Rename calls to match new patterns.
14648 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
14649 (load_pair<DREG:mode><DREG2:mode>): ... This.
14650 (store_pair<mode>): Rename to...
14651 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
14652 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
14653 New mode iterators.
14654 (V_INT_EQUIV): Handle SImode.
14655 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
14658 2018-05-22 Martin Sebor <msebor@redhat.com>
14661 * calls.c (maybe_warn_nonstring_arg): Use string length to set
14662 or ajust the presumed bound on an operation to avoid unnecessary
14665 2018-05-22 Martin Sebor <msebor@redhat.com>
14667 PR tree-optimization/85826
14668 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
14669 assuming that a DECL necesarily has a constant size.
14671 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
14673 PR middle-end/85862
14674 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
14676 2018-05-22 Richard Biener <rguenther@suse.de>
14678 PR tree-optimization/85834
14679 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
14680 non-constant and non-zero memset arguments.
14682 2018-05-22 Martin Liska <mliska@suse.cz>
14685 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
14687 2018-05-22 Richard Biener <rguenther@suse.de>
14689 PR tree-optimization/85863
14690 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
14691 comparisons when vectype is specified.
14692 (vectorizable_condition): Do not specify vectype for
14693 vect_is_simple_cond when SLP vectorizing.
14695 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
14698 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
14699 define __ibm128 as long double.
14700 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
14701 as a distinct type when IEEE 128-bit support is enabled.
14702 (init_float128_ieee): Fix up conversions between IFmode and IEEE
14703 128-bit types to use the correct functions.
14704 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
14705 convert between 128-bit floating point types that have different
14706 modes but the same representation, instead of using gen_lowpart to
14708 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
14710 (IFKF_reg): New attributes to give the register constraints for
14712 (extend<mode>tf2_internal): New insns to mark an explicit
14713 conversion between 128-bit floating point types that have a
14714 different mode but share the same representation.
14716 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
14718 PR tree-optimization/85814
14719 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
14720 a null return from get_strinfo when unsharing the next
14721 strinfo in the chain.
14723 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
14726 * varasm.c (weak_finish): Clean up weak_decls.
14728 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14730 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
14731 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
14732 UNSPEC_UADALP values.
14733 * config/aarch64/iterators.md (ABAL): New int iterator.
14736 (sur): Add mappings for the above.
14737 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
14739 (aarch64_<sur>abal<mode>_4): Likewise.
14740 (aarch64_<sur>adalp<mode>_3): Likewise.
14741 (<sur>sadv16qi): New define_expand.
14743 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
14745 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
14746 (*movdf_internal): Ditto.
14747 (*rcpsf2_sse): Ditto.
14748 (*rsqrtsf2_sse): Ditto.
14749 (*sqrt<mode>2_sse): Ditto.
14751 2018-05-21 Tamar Christina <tamar.christina@arm.com>
14753 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
14755 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
14756 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
14757 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
14758 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
14760 * config/aarch64/arm_neon.h: Likewise.
14761 * config/aarch64/iterators.md (VQ_I): New.
14763 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
14765 * config.gcc: Add arc/t-multilib-linux to tmake_file for
14767 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
14770 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
14772 * config/nds32/constraints.md (S): New constraint.
14773 * config/nds32/nds32.md (call_internal): Use constraint S.
14774 (call_value_internal): Likewise.
14775 (sibcall_internal): Likewise.
14776 (sibcall_value_internal): Likewise.
14778 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
14779 Chung-Ju Wu <jasonwucj@gmail.com>
14781 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
14782 into consideration.
14784 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
14785 Chung-Ju Wu <jasonwucj@gmail.com>
14787 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
14788 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
14789 (nds32_rtx_costs_impl): Simplify.
14790 (nds32_address_cost_impl): Simplify.
14791 (nds32_init_rtx_costs): New function.
14792 (nds32_rtx_costs_speed_prefer): Likewise.
14793 (nds32_rtx_costs_size_prefer): Likewise.
14794 (nds32_address_cost_speed_prefer): Likewise.
14795 (nds32_address_cost_speed_fwprop): Likewise.
14796 (nds32_address_cost_size_prefer): Likewise.
14797 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
14798 * config/nds32/nds32.c (nds32_option_override): Use
14799 nds32_init_rtx_costs function.
14801 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
14803 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
14804 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
14805 (TARGET_PIPELINE_N8): Likewise.
14806 (TARGET_PIPELINE_N10): Likewise.
14807 (TARGET_PIPELINE_N13): Likewise.
14808 (TARGET_PIPELINE_GRAYWOLF): Likewise.
14810 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
14812 * config/nds32/nds32-fpu.md: Update copyright year.
14814 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
14816 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
14818 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
14820 * config/nds32/nds32.c
14821 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
14822 * config/nds32/nds32.opt (minline-asm-r15): New option.
14824 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
14826 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
14828 * config/nds32/nds32.md (abssi2): New pattern.
14830 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
14832 * config/i386/i386.md (rex64namesuffix): New mode attribute.
14833 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
14834 Merge insn pattern from sse_cvtsi2ss<round_name> and
14835 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
14836 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
14837 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
14838 using SWI48 mode iterator.
14839 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
14840 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
14841 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
14842 pattern from sse_cvttss2si<round_saeonly_name>
14843 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
14844 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
14845 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
14846 using SWI48 mode iterator.
14847 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
14848 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
14849 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
14850 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
14851 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
14852 using SWI48 mode iterator.
14853 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
14854 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
14855 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
14856 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
14857 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
14858 SWI48 mode iterator.
14859 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
14860 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
14861 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
14862 pattern from sse_cvttsd2si<round_saeonly_name>
14863 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
14865 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
14867 * config/nds32/nds32-md-auxiliary.c
14868 (nds32_valid_smw_lwm_base_p): Refine.
14869 (nds32_output_smw_single_word): Refine.
14870 (nds32_output_smw_double_word): New.
14871 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
14873 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
14875 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
14876 (nds32_output_stack_pop): Refine.
14877 (nds32_expand_unaligned_load): Refine.
14878 (nds32_expand_unaligned_store): Refine.
14880 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
14881 Chung-Ju Wu <jasonwucj@gmail.com>
14883 * config/nds32/constants.md: Add TP_REGNUM constant.
14884 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
14885 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
14887 * config/nds32/nds32-doubleword.md: Consider flag_pic.
14888 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
14889 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
14890 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
14891 and PIC code generation.
14892 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
14894 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
14896 * config/nds32/nds32.md: Support TLS and PIC.
14897 * config/nds32/nds32.c: Support TLS and PIC.
14898 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
14899 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
14902 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
14904 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
14905 mode with E_ prefix.
14907 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
14908 Chung-Ju Wu <jasonwucj@gmail.com>
14910 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
14911 * config/nds32/nds32-md-auxiliary.c
14912 (symbolic_reference_mentioned_p): New.
14913 (nds32_legitimize_ict_address): New.
14914 (nds32_expand_ict_move): New.
14915 (nds32_indirect_call_referenced_p): New.
14916 (nds32_symbol_binds_local_p): Delete.
14917 (nds32_long_call_p): Modify.
14918 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
14919 * config/nds32/nds32-protos.h
14920 (symbolic_reference_mentioned_p): Declare.
14921 (nds32_legitimize_ict_address): Declare.
14922 (nds32_expand_ict_move): Declare.
14923 (nds32_indirect_call_referenced_p): Declare.
14924 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
14925 (nds32_relax_group): Use nds32_ict_const_p as condition.
14926 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
14927 (nds32_asm_file_start): Output ict_model directive in asm code.
14928 (nds32_legitimate_address_p): Consider indirect call.
14929 (nds32_print_operand): Consider indirect call.
14930 (nds32_print_operand_address): Consider indirect call.
14931 (nds32_insert_attributes): Handle "indirect_call" attribute.
14932 (TARGET_LEGITIMATE_ADDRESS_P): Define.
14933 (TARGET_LEGITIMATE_CONSTANT_P): Define.
14934 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
14935 (TARGET_DELEGITIMIZE_ADDRESS): Define.
14936 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
14937 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
14938 (TARGET_ICT_MODEL_SMALL): Define.
14939 (TARGET_ICT_MODEL_LARGE): Define.
14940 * config/nds32/nds32.md (movsi): Consider ict model.
14941 (call, call_value): Consider ict model.
14942 (sibcall, sibcall_value): Consider ict model.
14943 * config/nds32/nds32.opt (mict-model): New option.
14944 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
14947 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
14948 Monk Chiang <sh.chiang04@gmail.com>
14949 Jim Wilson <jimw@sifive.com>
14951 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
14952 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
14953 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
14954 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
14955 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
14956 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
14957 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
14958 compute save_libcall_adjustment properly.
14959 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
14960 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
14961 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
14962 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
14963 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
14964 (ABI_SPEC): Handle mabi=ilp32e.
14965 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
14966 (RVE): Add RVE mask.
14967 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
14968 <-march>: Add rv32e as an example.
14970 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
14973 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
14974 (intra_create_variable_infos): Handle C++ constructors.
14976 2018-05-18 Martin Liska <mliska@suse.cz>
14978 * passes.def: Remove a redundant pass.
14980 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
14983 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
14985 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14987 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
14989 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
14991 (armv2, armv2a, armv3, armv3m): Delete architectures.
14992 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
14993 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
14994 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
14996 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
14997 (*mulsidi3adddi): Likewise.
14998 (mulsidi3): Likewise.
14999 (*mulsidi3_nov6): Likewise.
15000 (umulsidi3): Likewise.
15001 (umulsidi3_nov6): Likewise.
15002 (umaddsidi4): Likewise.
15003 (*umulsidi3adddi): Likewise.
15004 (smulsi3_highpart): Likewise.
15005 (*smulsi3_highpart_nov6): Likewise.
15006 (umulsi3_highpart): Likewise.
15007 (*umulsi3_highpart_nov6): Likewise.
15008 * config/arm/arm.h (arm_arch3m): Delete.
15009 * config/arm/arm.c (arm_arch3m): Delete.
15010 (arm_option_override_internal): Update armv3-related comment.
15011 (arm_configure_build_target): Delete use of isa_bit_mode32.
15012 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
15013 (arm_rtx_costs_internal): Delete check of arm_arch3m.
15014 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
15015 (mulsa3): Likewise.
15016 (mulusa3): Likewise.
15017 * config/arm/arm-protos.h (arm_arch3m): Delete.
15018 * config/arm/arm-tables.opt: Regenerate.
15019 * config/arm/arm-tune.md: Likewise.
15020 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
15021 deleted architectures.
15023 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15025 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
15026 (armv5t, armv5te): New features.
15027 (ARMv5, ARMv5e): Delete fgroups.
15028 (ARMv5t, ARMv5te): Adjust for above changes.
15029 (ARMv6m): Likewise.
15030 (armv5, armv5e): Delete arches.
15031 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
15033 (*call_reg_arm): Likewise.
15034 (*call_value_reg_armv5): Likewise.
15035 (*call_value_reg_arm): Likewise.
15036 (*call_symbol): Likewise.
15037 (*call_value_symbol): Likewise.
15038 (*sibcall_insn): Likewise.
15039 (*sibcall_value_insn): Likewise.
15040 (clzsi2): Likewise.
15041 (prefetch): Likewise.
15042 (define_split and define_peephole2 dependent on arm_arch5):
15044 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
15046 (TARGET_ARM_QBIT): Likewise.
15047 (TARGET_DSP_MULTIPLY): Likewise.
15048 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
15049 (arm_arch5, arm_arch5e): Delete.
15050 (arm_arch5t, arm_arch5te): Declare.
15051 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
15052 (arm_arch5t): Declare.
15053 (arm_option_reconfigure_globals): Update for the above.
15054 (arm_options_perform_arch_sanity_checks): Update comment, replace
15055 use of arm_arch5 with arm_arch5t.
15056 (use_return_insn): Likewise.
15057 (arm_emit_call_insn): Likewise.
15058 (output_return_instruction): Likewise.
15059 (arm_final_prescan_insn): Likewise.
15060 (arm_coproc_builtin_available): Likewise.
15061 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
15062 arm_arch5e with arm_arch5t and arm_arch5te.
15063 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
15064 (arm_arch5t, arm_arch5te): Declare.
15065 * config/arm/arm-tables.opt: Regenerate.
15066 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
15067 * config/arm/t-multilib: Likewise.
15068 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
15069 instead of arm_arch5.
15070 (*call_reg_thumb1): Likewise.
15071 (*call_value_reg_thumb1_v5): Likewise.
15072 (*call_value_reg_thumb1): Likewise.
15073 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
15075 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
15077 2018-05-18 Martin Liska <mliska@suse.cz>
15079 PR gcov-profile/84846
15080 * doc/gcov.texi: Document -t option of gcov tool.
15082 2018-05-18 Martin Liska <mliska@suse.cz>
15084 PR gcov-profile/84846
15085 * gcov.c (print_usage): Add new -t option.
15086 (process_args): Handle the option.
15087 (generate_results): Use stdout as output when requested by
15090 2018-05-18 Martin Liska <mliska@suse.cz>
15092 PR gcov-profile/84846
15093 * coverage.c (coverage_init): Write PWD to .gcno file.
15094 * doc/gcov.texi: Document how working directory is printed.
15095 * gcov-dump.c (dump_gcov_file): Print PWD.
15096 * gcov.c (output_intermediate_file): Likewise.
15097 (read_graph_file): Read PWD string.
15098 (output_lines): Print PWD.
15100 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15102 PR middle-end/85817
15103 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
15104 for retval and return false if all args to phi are zero.
15106 2018-05-18 Richard Biener <rguenther@suse.de>
15108 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
15110 (evrp_dom_walker::before_dom_children): Call it.
15112 2018-05-18 Richard Biener <rguenther@suse.de>
15114 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
15115 results when processing array refs with variable index.
15117 2018-05-18 Toon Moene <toon@moene.org>
15119 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
15120 directly after that of -floop-interchange. Indicate that both
15121 options are enabled by default when specifying -O3.
15123 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15125 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
15126 iterator. Delete separate integer-mode vec_set<mode> expander.
15127 (aarch64_simd_vec_setv2di): Delete.
15128 (vec_setv2di): Delete.
15129 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
15130 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
15131 the "w, r" alternative.
15133 2018-05-18 Martin Liska <mliska@suse.cz>
15135 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
15136 * tree-pass.h (make_pass_lower_switch_O0): New function.
15137 * tree-switch-conversion.c (node_has_low_bound): Remove.
15138 (node_has_high_bound): Likewise.
15139 (node_is_bounded): Likewise.
15140 (class pass_lower_switch): Make it a template type and create
15142 (pass_lower_switch::execute): Add template argument.
15143 (make_pass_lower_switch): New function.
15144 (make_pass_lower_switch_O0): New function.
15145 (do_jump_if_equal): Remove.
15146 (emit_case_nodes): Simplify to just handle all 3 cases and leave
15147 all the hard work to tree optimization passes.
15149 2018-05-18 Martin Liska <mliska@suse.cz>
15151 * dbgcnt.c (limit_low): Renamed from limit.
15152 (limit_high): New variable.
15153 (dbg_cnt_is_enabled): Check for upper limit.
15154 (dbg_cnt): Adjust dumping.
15155 (dbg_cnt_set_limit_by_index): Add new argument for high
15157 (dbg_cnt_set_limit_by_name): Likewise.
15158 (dbg_cnt_process_single_pair): Parse new format.
15159 (dbg_cnt_process_opt): Use strtok.
15160 (dbg_cnt_list_all_counters): Remove 'value' and add
15162 * doc/invoke.texi: Document changes.
15164 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
15166 * doc/sourcebuild.texi (scalar_all_fma): Document.
15167 * tree.def (FMA_EXPR): Delete.
15168 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
15169 * internal-fn.c (ternary_direct): New macro.
15170 (expand_ternary_optab_fn): Likewise.
15171 (direct_ternary_optab_supported_p): Likewise.
15172 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
15173 * builtins.c (fold_builtin_fma): Delete.
15174 (fold_builtin_3): Don't call it.
15175 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
15176 * expr.c (expand_expr_real_2): Likewise.
15177 * fold-const.c (operand_equal_p): Likewise.
15178 (fold_ternary_loc): Likewise.
15179 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
15180 * gimple.c (DEFTREECODE): Likewise.
15181 * gimplify.c (gimplify_expr): Likewise.
15182 * optabs-tree.c (optab_for_tree_code): Likewise.
15183 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
15184 * tree-eh.c (operation_could_trap_p): Likewise.
15185 (stmt_could_throw_1_p): Likewise.
15186 * tree-inline.c (estimate_operator_cost): Likewise.
15187 * tree-pretty-print.c (dump_generic_node): Likewise.
15188 (op_code_prio): Likewise.
15189 * tree-ssa-loop-im.c (stmt_cost): Likewise.
15190 * tree-ssa-operands.c (get_expr_operands): Likewise.
15191 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
15192 * fold-const-call.h (fold_fma): Delete.
15193 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
15194 CFN_FNMA and CFN_FNMS.
15195 (fold_fma): Delete.
15196 * genmatch.c (combined_fn): New enum.
15197 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
15198 (commutative_op): New function.
15199 (commutate): Use it. Handle more than 2 operands.
15200 (dt_operand::gen_gimple_expr): Use commutative_op.
15201 (parser::parse_expr): Allow :c to be used with non-binary
15202 operators if the commutative operand is known.
15203 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
15204 CFN_FMS, CFN_FNMA and CFN_FNMS.
15205 (backprop::process_assign_use): Remove FMA_EXPR handling.
15206 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
15207 (gen_hsa_fma): New function.
15208 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
15209 IFN_FNMA and IFN_FNMS.
15210 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
15211 * gimple-fold.h (follow_all_ssa_edges): Declare.
15212 * gimple-fold.c (follow_all_ssa_edges): New function.
15213 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
15214 gimple_build interface and use follow_all_ssa_edges to fold the result.
15215 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
15216 instead of checking for optabs directly.
15217 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
15218 rather than FMA_EXPRs.
15219 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
15220 call to IFN_FMA instead of an FMA_EXPR.
15222 2018-05-17 Jim Wilson <jimw@sifive.com>
15224 * expr.c (do_tablejump): When converting index to Pmode, if we have a
15225 sign extended promoted subreg, and the range does not have the sign bit
15226 set, then do a sign extend.
15228 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
15229 test, check for sign extended subreg and/or constant operands, and
15230 do a sign extend in that case.
15232 2018-05-17 Steve Ellcey <sellcey@cavium.com>
15234 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
15235 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
15237 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
15238 Change logics_shift_reg to logics_shift_imm.
15239 (thunderx2t99_fp_loadpair_basic): Delete.
15240 (thunderx2t99_fp_storepair_basic): Delete.
15241 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
15242 (thunderx2t99_asimd_polynomial): Delete.
15243 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
15244 and neon_fp_mul_d_scalar_q.
15245 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
15246 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
15247 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
15248 (thunderx2t99_asimd_lut): Add missing tbl types.
15249 (thunderx2t99_asimd_ext): Delete.
15250 (thunderx2t99_asimd_load1_1_mult): Delete.
15251 (thunderx2t99_asimd_load1_2_mult): Delete.
15252 (thunderx2t99_asimd_load1_ldp): New.
15253 (thunderx2t99_asimd_load1): New.
15254 (thunderx2t99_asimd_load2): Add missing *load2* types.
15255 (thunderx2t99_asimd_load3): New.
15256 (thunderx2t99_asimd_load4): New.
15257 (thunderx2t99_asimd_store1_1_mult): Delete.
15258 (thunderx2t99_asimd_store1_2_mult): Delete.
15259 (thunderx2t99_asimd_store2_mult): Delete.
15260 (thunderx2t99_asimd_store2_onelane): Delete.
15261 (thunderx2t99_asimd_store_stp): New.
15262 (thunderx2t99_asimd_store1): New.
15263 (thunderx2t99_asimd_store2): New.
15264 (thunderx2t99_asimd_store3): New.
15265 (thunderx2t99_asimd_store4): New.
15267 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
15269 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
15270 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
15272 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
15273 Segher Boessenkool <segher@kernel.crashing.org>
15276 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
15279 2018-05-17 Richard Biener <rguenther@suse.de>
15281 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
15282 for pruning loop and prune defs feeding only already visited PHIs.
15284 2018-05-17 Richard Biener <rguenther@suse.de>
15286 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
15288 2018-05-17 Bin Cheng <bin.cheng@arm.com>
15289 Richard Biener <rguenther@suse.de>
15291 PR tree-optimization/85793
15292 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
15293 for VMAT_ELEMENTWISE.
15295 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
15297 * internal-fn.h (lookup_internal_fn): Declare
15298 * internal-fn.c (lookup_internal_fn): New function.
15299 * gimple.c (gimple_build_call_from_tree): Handle calls to
15300 internal functions.
15301 * gimple-pretty-print.c (dump_gimple_call): Print "." before
15302 internal function names.
15303 * tree-pretty-print.c (dump_generic_node): Likewise.
15304 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
15306 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
15308 * gimple-fold.h (gimple_build): Make the function forms take
15309 combined_fn rather than built_in_function.
15310 (gimple_simplify): Likewise.
15311 * gimple-match-head.c (gimple_simplify): Likewise.
15312 * gimple-fold.c (gimple_build): Likewise.
15313 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
15314 rather than gimple_build_call_internal.
15315 (get_initial_defs_for_reduction): Likewise.
15316 (vect_create_epilog_for_reduction): Likewise.
15317 (vectorizable_live_operation): Likewise.
15319 2018-05-17 Martin Liska <mliska@suse.cz>
15321 * gimple-ssa-sprintf.c (format_directive): Do not use
15322 space in between 'G_' and '('.
15324 2018-05-17 Jakub Jelinek <jakub@redhat.com>
15327 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
15328 even if the mask is not all ones.
15331 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
15333 (ix86_gimple_fold_builtin): Likewise.
15336 * config/i386/i386.c: Include tree-vector-builder.h.
15337 (ix86_vector_shift_count): New function.
15338 (ix86_fold_builtin): Fold shift builtins by scalar count.
15339 (ix86_gimple_fold_builtin): Likewise.
15341 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
15342 _mm512_setzero): New intrinsics.
15344 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
15345 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15347 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
15348 code generation for cases where splatting a value is not useful.
15349 * simplify-rtx.c (simplify_ternary_operation): Simplify
15350 vec_merge across a vec_duplicate and a paradoxical subreg forming
15351 a vector mode to a vec_concat.
15353 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
15355 * config.gcc: Support "goldmont-plus".
15356 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
15358 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
15359 PROCESSOR_GOLDMONT_PLUS.
15360 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
15361 (processor_target_table): Add "goldmont-plus".
15362 (PTA_GOLDMONT_PLUS): Define.
15363 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
15364 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
15365 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
15366 (fold_builtin_cpu): Add "goldmont-plus".
15367 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
15368 (ix86_option_override_internal): Add "goldmont-plus".
15369 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
15370 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
15371 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
15372 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
15374 2018-05-17 Richard Biener <rguenther@suse.de>
15376 PR tree-optimization/85757
15377 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
15378 remove defs that only feed that PHI from further processing.
15380 2018-05-16 Jim Wilson <jimw@sifive.com>
15382 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
15384 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
15386 2018-05-16 Mark Wielaard <mark@klomp.org>
15388 * dwarf2out.c (count_index_strings): New function.
15389 (output_indirect_strings): Call count_index_strings and generate
15390 header for dwarf_version >= 5.
15392 2018-05-16 Mark Wielaard <mark@klomp.org>
15394 * dwarf2out.c (dwarf_FORM): New function.
15395 (set_indirect_string): Use dwarf_FORM.
15396 (reset_indirect_string): Likewise.
15397 (size_of_die): Likewise.
15398 (value_format): Likewise.
15399 (output_die): Likewise.
15400 (add_skeleton_AT_string): Likewise.
15401 (output_macinfo_op): Likewise.
15402 (index_string): Likewise.
15403 (output_index_string_offset): Likewise.
15404 (output_index_string): Likewise.
15405 (count_index_strings): Likewise.
15407 2018-05-16 Carl Love <cel@us.ibm.com>
15409 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
15410 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
15412 2018-05-16 Martin Jambor <mjambor@suse.cz>
15414 * ipa-prop.c (ipa_free_all_edge_args): Remove.
15415 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
15417 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
15419 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
15420 (fnma<mode>4): Likewise.
15421 (fms<mode>4): Likewise.
15422 (fnms<mode>4): Likewise.
15423 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
15424 (aarch64_fnma<mode>4): Likewise.
15425 (aarch64_fms<mode>4): Likewise.
15426 (aarch64_fnms<mode>4): Likewise.
15427 (aarch64_fnmadd<mode>4): Likewise.
15429 2018-05-16 Jason Merrill <jason@redhat.com>
15431 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
15433 2018-05-16 Richard Biener <rguenther@suse.de>
15435 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
15436 (dump_stmt_cost): Declare.
15437 (add_stmt_cost): Dump cost we add.
15438 (add_stmt_costs): New function.
15439 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
15440 No longer exported.
15441 (vect_analyze_stmt): Adjust prototype.
15442 (vectorizable_condition): Likewise.
15443 (vectorizable_live_operation): Likewise.
15444 (vectorizable_reduction): Likewise.
15445 (vectorizable_induction): Likewise.
15446 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
15447 cost vector to pass to vectorizable_ and record afterwards.
15448 (vect_model_reduction_cost): Take cost vector argument and adjust.
15449 (vect_model_induction_cost): Likewise.
15450 (vectorizable_reduction): Likewise.
15451 (vectorizable_induction): Likewise.
15452 (vectorizable_live_operation): Likewise.
15453 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
15454 SLP_TREE_NUMBER_OF_VEC_STMTS.
15455 (vect_analyze_slp_cost_1): Remove.
15456 (vect_analyze_slp_cost): Likewise.
15457 (vect_slp_analyze_node_operations): Take visited args and
15458 a target cost vector. Avoid processing already visited stmt sets.
15459 (vect_slp_analyze_operations): Use a local cost vector to gather
15460 costs and register those of non-discarded instances.
15461 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
15462 (vect_schedule_slp_instance): Remove copying of
15463 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
15465 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
15466 adding cost. Record cost entry location.
15467 (vect_prologue_cost_for_slp_op): Function to compute cost of
15468 a constant or invariant generated for SLP vect in the prologue,
15469 split out from vect_analyze_slp_cost_1.
15470 (vect_model_simple_cost): Make static. Adjust for SLP costing.
15471 (vect_model_promotion_demotion_cost): Likewise.
15472 (vect_model_store_cost): Likewise, make static.
15473 (vect_model_load_cost): Likewise.
15474 (vectorizable_bswap): Add cost vector arg and adjust.
15475 (vectorizable_call): Likewise.
15476 (vectorizable_simd_clone_call): Likewise.
15477 (vectorizable_conversion): Likewise.
15478 (vectorizable_assignment): Likewise.
15479 (vectorizable_shift): Likewise.
15480 (vectorizable_operation): Likewise.
15481 (vectorizable_store): Likewise.
15482 (vectorizable_load): Likewise.
15483 (vectorizable_condition): Likewise.
15484 (vectorizable_comparison): Likewise.
15485 (can_vectorize_live_stmts): Likewise.
15486 (vect_analyze_stmt): Likewise.
15487 (vect_transform_stmt): Adjust calls to vectorizable_*.
15488 * tree-vectorizer.c: Include gimple-pretty-print.h.
15489 (dump_stmt_cost): New function.
15491 2018-05-16 Richard Biener <rguenther@suse.de>
15493 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
15494 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
15495 * tree-ssa-dse.c: Include tree-ssa-loop.h.
15496 (check_name): New callback.
15497 (dse_classify_store): Track cycles via a visited bitmap of PHI
15498 defs and simplify handling of in-loop and across loop dead stores
15499 and properly fail for loop-variant refs. Handle byte-tracking with
15500 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
15503 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
15505 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
15506 (vect_get_mask_type_for_stmt): Likewise.
15507 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
15509 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
15510 to determine the statement's vector type and the vector type that
15511 should be used for calculating nunits. Deal with cases in which
15512 the type has to be deferred.
15513 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
15514 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
15515 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
15516 (vect_determine_vf_for_stmt): New functions, split out from...
15517 (vect_determine_vectorization_factor): ...here.
15518 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
15519 (vect_get_mask_type_for_stmt): New functions, split out from
15520 vect_determine_vectorization_factor.
15522 2018-05-16 Richard Biener <rguenther@suse.de>
15524 * tree-cfg.c (verify_gimple_assign_ternary): Properly
15525 verify the [VEC_]COND_EXPR embedded comparison.
15527 2018-05-15 Martin Sebor <msebor@redhat.com>
15529 PR tree-optimization/85753
15530 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
15531 RECORD_TYPE in addition to ARRAY_TYPE.
15533 2018-05-15 Martin Sebor <msebor@redhat.com>
15535 PR middle-end/85643
15536 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
15538 2018-05-15 Richard Biener <rguenther@suse.de>
15540 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
15541 add by_clobber_p one. Change algorithm to collect all defs
15542 representing uses we need to walk and try reducing them to
15543 a single one before failing.
15544 (dse_dom_walker::dse_optimize_stmt): Adjust.
15546 2018-05-13 Mark Wielaard <mark@klomp.org>
15548 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
15549 (size_of_loc_descr): Likewise.
15550 (output_loc_operands): Likewise.
15551 (output_loc_operands_raw): Likewise.
15552 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
15553 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
15554 (hash_loc_operands): Likewise.
15555 (compare_loc_operands): Likewise.
15557 2018-05-14 Mark Wielaard <mark@klomp.org>
15559 * dwarf2out.c (count_index_addrs): New function.
15560 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
15562 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15564 PR tree-optimization/83648
15565 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
15566 return value as malloc candidate.
15568 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15571 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
15572 param as true in call to suggest_attribute.
15574 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
15576 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
15577 -mreadonly-in-sdata.
15579 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15581 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
15583 (aarch64_crypto_aesd_fused): Likewise.
15585 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
15587 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
15588 (movsi_aarch64): Likewise.
15589 (load_pairsi): Likewise.
15590 (load_pairdi): Likewise.
15591 (store_pairsi): Likewise.
15592 (store_pairdi): Likewise.
15593 (load_pairsf): Likewise.
15594 (load_pairdf): Likewise.
15595 (store_pairsf): Likewise.
15596 (store_pairdf): Likewise.
15597 (zero_extend): Likewise.
15598 (trunc): Swap alternatives.
15599 (fcvt_target): Add '?' to prefer w over r.
15601 2018-05-14 Jakub Jelinek <jakub@redhat.com>
15604 * config/i386/i386.md: Disallow non-commutative arithmetics in
15605 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
15606 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
15607 in the peephole2 before it.
15609 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
15611 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
15612 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
15613 (ix86_handle_option): Handle -mcldemote.
15614 * config.gcc: New header.
15615 * config/i386/cldemoteintrin.h: New file.
15616 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
15617 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
15619 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
15620 OPTION_MASK_ISA_CLDEMOTE.
15621 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
15622 (ix86_valid_target_attribute_inner_p): Ditto.
15623 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
15624 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
15625 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
15626 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
15627 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
15629 * config/i386/i386.opt: Add -mcldemote.
15630 * config/i386/x86intrin.h: New header.
15631 * doc/invoke.texi: Add -mcldemote.
15633 2018-05-14 Richard Biener <rguenther@suse.de>
15635 * doc/match-and-simplify.texi: Adjust :s documentation.
15637 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
15639 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
15640 intended memcpy size.
15641 (REORDER_45): Likewise.
15643 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
15645 * sort.cc: New file.
15646 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
15647 * vec.c (qsort_chk): Use gcc_qsort.
15648 * Makefile.in (OBJS-libcommon): Add sort.o.
15649 (build/sort.o): New target. Use it...
15650 (BUILD_RTL): ... here, and...
15651 (build/gencfn-macros): ... here, and...
15652 (build/genmatch): ... here.
15654 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
15655 Chung-Ju Wu <jasonwucj@gmail.com>
15657 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
15658 * config/nds32/nds32-graywolf.md: New file.
15659 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
15660 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
15662 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
15663 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
15664 * config/nds32/nds32.md (pipeline_model): Add graywolf.
15665 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
15666 * config/nds32/pipelines.md: Include n15 settings.
15668 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
15669 Chung-Ju Wu <jasonwucj@gmail.com>
15671 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
15672 * config/nds32/nds32-n13.md: New file.
15673 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
15674 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
15676 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
15677 * config/nds32/nds32.md (pipeline_model): Add n13.
15678 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
15679 * config/nds32/pipelines.md: Include n13 settings.
15681 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
15682 Chung-Ju Wu <jasonwucj@gmail.com>
15684 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
15685 * config/nds32/nds32-n10.md: New file.
15686 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
15687 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
15689 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
15690 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
15691 * config/nds32/nds32.md (pipeline_model): Add n10.
15692 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
15693 * config/nds32/pipelines.md: Include n10 settings.
15695 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
15696 Kito Cheng <kito.cheng@gmail.com>
15697 Chung-Ju Wu <jasonwucj@gmail.com>
15699 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
15700 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
15701 Add enum values for DSP extension instructions.
15702 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
15704 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
15705 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
15706 New code iterators.
15707 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
15708 * config/nds32/nds32-dspext.md: New file for DSP implementation.
15709 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
15710 * config/nds32/nds32-intrinsic.md: Likewise.
15711 * config/nds32/nds32_intrinsic.h: Likewise.
15712 * config/nds32/nds32-md-auxiliary.c: Likewise.
15713 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
15714 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
15715 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
15716 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
15717 * config/nds32/nds32-protos.h: New declarations for DSP extension.
15718 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
15719 TYPE_DMAC in switch statement.
15720 * config/nds32/nds32.c: New checking and implementation for DSP
15721 extension instructions.
15722 * config/nds32/nds32.h: Likewise.
15723 * config/nds32/nds32.md: Likewise.
15724 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
15725 * config/nds32/predicates.md: Implement new predicates for DSP
15728 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
15730 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
15731 Reformat alternatives and attributes so it is easier to identify
15732 which constraints/attributes go with which instruction.
15733 (mov<mode>_hardfloat32, FMOVE64): Likewise.
15734 (mov<mode>_softfloat32, FMOVE64): Likewise.
15735 (mov<mode>_hardfloat64, FMOVE64): Likewise.
15736 (mov<mode>_softfloat64, FMOVE64): Likewise.
15738 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
15740 * doc/extend.texi (PowerPC Built-in Functions): Rename this
15742 (Basic PowerPC Built-in Functions): The new name of the
15743 subsection previously known as "PowerPC Built-in Functions".
15744 (Basic PowerPC Built-in Functions Available on all Configurations):
15746 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
15747 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
15748 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
15749 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
15751 2018-05-11 Martin Jambor <mjambor@suse.cz>
15754 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
15757 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
15760 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
15762 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
15764 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
15765 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
15766 (ix86_handle_option): Handle -mwaitpkg.
15767 * config.gcc: New header.
15768 * config/i386/cpuid.h (bit_WAITPKG): New bit.
15769 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
15770 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
15772 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
15773 OPTION_MASK_ISA_WAITPKG.
15774 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
15775 (ix86_option_override_internal): Add PTA_WAITPKG.
15776 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
15777 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
15778 IX86_BUILTIN_TPAUSE.
15779 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
15780 __builtin_ia32_umwait and __builtin_ia32_tpause.
15781 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
15782 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
15783 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
15784 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
15785 UNSPECV_TPAUSE): New.
15786 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
15787 * config/i386/i386.opt: Add -mwaitpkg.
15788 * config/i386/waitpkgintrin.h: New file.
15789 * config/i386/x86intrin.h: New header.
15790 * doc/invoke.texi: Add -mwaitpkg.
15792 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
15795 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
15797 (cortex-m0): Use armv6s-m isa.
15798 (cortex-m0plus): Likewise.
15799 (cortex-m1): Likewise.
15800 (cortex-m0.small-multiply): Likewise.
15801 (cortex-m0plus.small-multiply): Likewise.
15802 (cortex-m1.small-multiply): Likewise.
15804 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
15805 Jakub Jelinek <jakub@redhat.com>
15807 PR tree-optimization/85692
15808 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
15809 source permute as well.
15811 2018-05-11 Martin Liska <mliska@suse.cz>
15814 * doc/extend.texi: Document LLVM style format for no_sanitize
15817 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
15819 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
15820 mode_supports_vsx_dform_quad to mode_supports_dq_form.
15821 (mode_supports_vsx_dform_quad): Likewise.
15822 (mode_supports_vmx_dform): Move these functions to be next to the
15823 other mode_supports functions.
15824 (mode_supports_dq_form): Likewise.
15825 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
15826 mode_supports_dq_form.
15827 (reg_offset_addressing_ok_p): Likewise.
15828 (offsettable_ok_by_alignment): Likewise.
15829 (rs6000_legitimate_offset_address_p): Likewise.
15830 (legitimate_lo_sum_address_p): Likewise.
15831 (rs6000_legitimize_address): Likewise.
15832 (rs6000_legitimize_reload_address): Likewise.
15833 (rs6000_secondary_reload_inner): Likewise.
15834 (rs6000_preferred_reload_class): Likewise.
15835 (rs6000_output_move_128bit): Likewise.
15837 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
15839 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
15840 Generate SImode target register for null target.
15841 <case IX86_BUILTIN_XGETBV>: Ditto.
15842 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
15843 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
15845 2018-05-10 Carl Love <cel@us.ibm.com>
15847 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
15848 dcbtt and dcbtstt if operands[2] is 0.
15850 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
15853 * config/i386/sse.md (usadv64qi): New expander.
15855 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
15857 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
15858 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
15859 -maltivec=be support.
15860 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
15861 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
15862 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
15863 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
15864 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
15865 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
15866 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
15867 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
15868 altivec_vsumsws): Adjust.
15869 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
15870 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
15871 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
15873 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
15874 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
15875 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
15876 (altivec_lve<VI_char>x): Delete expand.
15877 (*altivec_lve<VI_char>x_internal): Rename to...
15878 (altivec_lve<VI_char>x): ... this.
15879 (altivec_lvxl_<mode>): Delete expand.
15880 (*altivec_lvxl_<mode>_internal): Rename to ...
15881 (altivec_lvxl_<mode>): ... this.
15882 (altivec_stvxl_<mode>): Delete expand.
15883 (*altivec_stvxl_<mode>_internal): Rename to ...
15884 (altivec_stvxl_<mode>): ... this.
15885 (altivec_stve<VI_char>x): Delete expand.
15886 (*altivec_stve<VI_char>x_internal): Rename to ...
15887 (altivec_stve<VI_char>x): ... this.
15888 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
15889 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
15890 reduc_plus_scal_<mode>): Adjust.
15891 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
15893 (rs6000_cpu_cpp_builtins): Adjust.
15894 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
15895 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
15896 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
15897 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
15898 -maltivec=be support.
15899 (rs6000_split_vec_extract_var): Adjust.
15900 (rs6000_split_v4si_init): Adjust.
15901 (swap_selector_for_mode): Delete.
15902 (altivec_expand_lvx_be, altivec_expand_stvx_be,
15903 altivec_expand_stvex_be): Delete.
15904 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
15905 -maltivec=be support.
15906 (rs6000_gimple_fold_builtin): Ditto.
15907 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
15909 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
15910 (TARGET_DIRECT_MOVE_64BIT): Adjust.
15911 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
15912 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
15913 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
15914 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
15915 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
15916 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
15917 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
15918 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
15919 anonymous split): Adjust.
15920 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
15921 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
15923 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
15925 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
15926 when --with-gxx-include-dir is also specified.
15927 * configure: Regenerate.
15929 2018-05-09 Jim Wilson <jimw@sifive.com>
15932 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
15933 * config/riscv/t-withmultilib: New.
15934 * config/riscv/withmultilib.h: New.
15935 * doc/install.texi: Document RISC-V --with-multilib-list support.
15937 2018-05-09 Richard Biener <rguenther@suse.de>
15939 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
15941 (vect_bb_vectorization_profitable_p): Adjust. Compute
15942 actual scalar cost using the cost vector and the add_stmt_cost
15945 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
15947 PR rtl-optimization/85645
15948 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
15949 in the REG_CFA_REGISTER note for LR, don't leave it empty.
15951 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
15953 PR rtl-optimization/85645
15954 * shrink-wrap.c (spread_components): Return a boolean saying if
15955 anything was changed.
15956 (try_shrink_wrapping_separate): Iterate spread_components until
15957 nothing changes anymore.
15959 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
15961 PR rtl-optimization/85645
15962 * regrename.c (build_def_use): Also kill the chains that include the
15963 destination of a REG_CFA_REGISTER note.
15965 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
15967 PR rtl-optimization/85645
15968 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
15969 insn that has a REG_CFA_REGISTER note.
15971 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
15973 * cfgexpand.c (expand_clobber): New function.
15974 (expand_gimple_stmt_1): Use it.
15975 * tree-vect-stmts.c (vect_clobber_variable): New function,
15977 (vectorizable_simd_clone_call): ...here.
15978 (vectorizable_store): Emit a clobber either side of an
15979 IFN_STORE_LANES sequence.
15980 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
15982 2018-05-09 Tom de Vries <tom@codesourcery.com>
15985 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
15986 (define_insn "trap_if_false"): Add exit after trap.
15988 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
15990 PR rtl-optimization/85638
15991 * bb-reorder.c: Include common/common-target.h.
15992 (create_forwarder_block): New function extracted from...
15993 (fix_up_crossing_landing_pad): ...here. Rename into...
15994 (dw2_fix_up_crossing_landing_pad): ...this.
15995 (sjlj_fix_up_crossing_landing_pad): New function.
15996 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
15997 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
15998 from both partitions and exit the loop after one iteration.
16000 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
16003 * doc/extend.texi (PowerPC Built-in Functions): Rename this
16005 (Basic PowerPC Built-in Functions): The new name of the
16006 subsection previously known as "PowerPC Built-in Functions".
16007 (Basic PowerPC Built-in Functions Available on all Configurations):
16009 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
16011 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
16013 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
16015 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
16018 2018-05-08 Jim Wilson <jimw@sifive.com>
16020 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
16021 (LD_EMUL_SUFFIX): New.
16022 (LINK_SPEC): Use it.
16024 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
16026 * doc/extend.texi (PowerPC Built-in Functions): Rename this
16028 (Basic PowerPC Built-in Functions): The new name of the
16029 subsection previously known as "PowerPC Built-in Functions".
16030 (Basic PowerPC Built-in Functions Available on all Configurations):
16032 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
16034 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
16036 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
16038 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
16041 2018-05-08 Jakub Jelinek <jakub@redhat.com>
16044 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
16045 after cmpelim optimization.
16047 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
16049 * config.gcc: Support "goldmont".
16050 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
16051 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
16052 PROCESSOR_GOLDMONT.
16053 * config/i386/i386.c (m_GOLDMONT): Define.
16054 (processor_target_table): Add "goldmont".
16055 (PTA_GOLDMONT): Define.
16056 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
16057 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
16058 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
16059 (fold_builtin_cpu): Add "goldmont".
16060 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
16061 (ix86_option_override_internal): Add "goldmont".
16062 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
16063 (processor_type): Add PROCESSOR_GOLDMONT.
16064 * config/i386/i386.md: Add CPU "glm".
16065 * config/i386/glm.md: New file.
16066 * config/i386/x86-tune.def: Add m_GOLDMONT.
16067 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
16069 2018-05-08 Jakub Jelinek <jakub@redhat.com>
16072 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
16074 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
16075 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
16076 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
16079 * config/i386/i386.c (ix86_fold_builtin): Handle
16080 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
16083 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
16084 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
16086 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
16089 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
16090 (check_arch): Likewise.
16091 (check_fpu): Return the result rather than printing it.
16092 (end arch): Fix operator precedence.
16093 (end cpu): Likewise.
16094 (END): Print the result from check_fpu.
16096 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
16097 Alan Hayward <alan.hayward@arm.com>
16098 David Sherwood <david.sherwood@arm.com>
16100 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
16101 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
16102 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
16103 (*fcmuo<mode>_and): New patterns.
16105 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
16107 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
16108 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
16109 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
16110 (cmp_op, sve_imm_con): New code attributes.
16111 (SVE_COND_INT_CMP, imm_con): Delete.
16112 (cmp_op): Remove above unspecs from int attribute.
16113 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
16115 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
16116 comparison-specific unspecs.
16117 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
16118 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
16119 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
16120 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
16121 (*vec_fcm<cmp_op><mode>): Rename to...
16122 (*fcm<cmp_op><mode>): ...this and adjust likewise.
16123 (*vec_fcmuo<mode>): Rename to...
16124 (*fcmuo<mode>): ...this and adjust likewise.
16125 (*pred_fcm<cmp_op><mode>): New pattern.
16126 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
16127 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
16129 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
16131 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
16132 (aarch64_emit_sve_predicated_cond): New function.
16133 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
16134 (aarch64_emit_unspec_cond_or): Replace with...
16135 (aarch64_emit_sve_or_conds): ...this new function. Use
16136 aarch64_emit_sve_ptrue_op for the individual comparisons and
16137 aarch64_emit_binop to OR them together.
16138 (aarch64_emit_inverted_unspec_cond): Replace with...
16139 (aarch64_emit_sve_inverted_cond): ...this new function. Use
16140 aarch64_emit_sve_ptrue_op for the comparison and
16141 aarch64_emit_unop to invert the result.
16142 (aarch64_expand_sve_vec_cmp_float): Update after the above
16143 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
16145 2018-05-07 Nathan Sidwell <nathan@acm.org>
16147 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
16148 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
16149 (Backwards Compatibility): Likewise.
16151 2018-05-07 Luis Machado <luis.machado@linaro.org>
16155 2018-05-07 Luis Machado <luis.machado@linaro.org>
16157 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
16158 <prefetch_dynamic_strides>: New const bool field.
16159 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
16160 prefetch_dynamic_strides.
16161 (exynosm1_prefetch_tune): Likewise.
16162 (thunderxt88_prefetch_tune): Likewise.
16163 (thunderx_prefetch_tune): Likewise.
16164 (thunderx2t99_prefetch_tune): Likewise.
16165 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
16167 (aarch64_override_options_internal): Update to set
16168 PARAM_PREFETCH_DYNAMIC_STRIDES.
16169 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
16170 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
16171 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
16172 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
16173 prefetch-dynamic-strides setting.
16175 2018-05-07 Luis Machado <luis.machado@linaro.org>
16177 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
16178 <minimum_stride>: New const int field.
16179 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
16180 minimum_stride field.
16181 (exynosm1_prefetch_tune): Likewise.
16182 (thunderxt88_prefetch_tune): Likewise.
16183 (thunderx_prefetch_tune): Likewise.
16184 (thunderx2t99_prefetch_tune): Likewise.
16185 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
16186 (aarch64_override_options_internal): Update to set
16187 PARAM_PREFETCH_MINIMUM_STRIDE.
16188 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
16189 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
16190 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
16191 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
16192 stride is constant and is below the minimum stride threshold.
16194 2018-05-07 Luis Machado <luis.machado@linaro.org>
16196 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
16199 2018-05-07 Luis Machado <luis.machado@linaro.org>
16201 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
16202 <prefetch_dynamic_strides>: New const bool field.
16203 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
16204 prefetch_dynamic_strides.
16205 (exynosm1_prefetch_tune): Likewise.
16206 (thunderxt88_prefetch_tune): Likewise.
16207 (thunderx_prefetch_tune): Likewise.
16208 (thunderx2t99_prefetch_tune): Likewise.
16209 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
16211 (aarch64_override_options_internal): Update to set
16212 PARAM_PREFETCH_DYNAMIC_STRIDES.
16213 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
16214 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
16215 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
16216 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
16217 prefetch-dynamic-strides setting.
16219 2018-05-07 Luis Machado <luis.machado@linaro.org>
16221 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
16222 <minimum_stride>: New const int field.
16223 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
16224 minimum_stride field.
16225 (exynosm1_prefetch_tune): Likewise.
16226 (thunderxt88_prefetch_tune): Likewise.
16227 (thunderx_prefetch_tune): Likewise.
16228 (thunderx2t99_prefetch_tune): Likewise.
16229 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
16230 (aarch64_override_options_internal): Update to set
16231 PARAM_PREFETCH_MINIMUM_STRIDE.
16232 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
16233 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
16234 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
16235 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
16236 stride is constant and is below the minimum stride threshold.
16238 2018-05-06 Jakub Jelinek <jakub@redhat.com>
16241 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
16242 the type is addressable. Don't force op into register if it has
16245 2018-05-05 Roland McGrath <mcgrathr@google.com>
16248 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
16249 any section for which we don't know a specific type it should have,
16250 regardless of name. Previously this was done only for the exact
16251 names ".init_array", ".fini_array", and ".preinit_array".
16252 (default_elf_asm_named_section): Add comment about
16253 relationship with default_section_type_flags and SECTION_NOTYPE.
16254 (get_section): Don't consider it a type conflict if one side has
16255 SECTION_NOTYPE and the other doesn't, as long as neither has the
16256 SECTION_BSS et al used in the default_section_type_flags logic.
16258 2018-05-05 Tom de Vries <tom@codesourcery.com>
16261 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
16262 (workaround_barsyncs): New function.
16263 (nvptx_reorg): Use workaround_barsyncs.
16264 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
16265 (define_expand "nvptx_membar_cta"): New define_expand.
16266 (define_insn "*nvptx_membar_cta"): New insn.
16268 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
16270 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
16271 To improve optimization opportunities.
16272 * builtin-types.def: The new needed builtin types for the above.
16274 2018-05-04 Richard Biener <rguenther@suse.de>
16276 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
16277 * gimple-ssa-store-merging.c
16278 (imm_store_chain_info::output_merged_store): Remove redundant create,
16279 release split_store vector contents on failure.
16280 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
16281 scalar stmt vector on cache hit.
16283 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
16285 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
16287 * config.gcc (powerpc-xilinx-eabi*): Remove.
16288 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
16290 (fusion_addis_mem_combo_load): Ditto.
16291 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
16293 (rs6000_cpu_cpp_builtins): Ditto.
16294 * config/rs6000/rs6000-linux.c
16295 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
16296 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
16297 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
16299 (rs6000_setup_reg_addr_masks): Ditto.
16300 (rs6000_init_hard_regno_mode_ok): Ditto.
16301 (rs6000_option_override_internal): Ditto.
16302 (legitimate_lo_sum_address_p): Ditto.
16303 (rs6000_legitimize_address): Ditto.
16304 (rs6000_legitimize_reload_address): Ditto.
16305 (rs6000_legitimate_address_p): Ditto.
16306 (abi_v4_pass_in_fpr): Ditto.
16307 (setup_incoming_varargs): Ditto.
16308 (rs6000_gimplify_va_arg): Ditto.
16309 (rs6000_split_multireg_move): Ditto.
16310 (rs6000_savres_strategy): Ditto.
16311 (rs6000_emit_prologue_components): Ditto.
16312 (rs6000_emit_epilogue_components): Ditto.
16313 (rs6000_emit_prologue): Ditto.
16314 (rs6000_emit_epilogue): Ditto.
16315 (rs6000_elf_file_end): Ditto.
16316 (rs6000_function_value): Ditto.
16317 (rs6000_libcall_value): Ditto.
16318 * config/rs6000/rs6000.h: Ditto.
16319 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
16320 (TARGET_MINMAX): ... this. New.
16321 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
16322 * config/rs6000/rs6000.md: Remove Xilinx FP support.
16323 (*movsi_internal1_single): Delete.
16324 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
16325 mfpu=, mxilinx-fpu): Delete.
16326 * config/rs6000/singlefp.h: Delete.
16327 * config/rs6000/sysv4.h: Remove Xilinx FP support.
16328 * config/rs6000/t-rs6000: Ditto.
16329 * config/rs6000/t-xilinx: Delete.
16330 * config/rs6000/titan.md: Adjust for fp_type removal.
16331 * config/rs6000/vsx.md: Remove Xilinx FP support.
16332 (VStype_simple): Delete.
16333 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
16334 * config/rs6000/xfpu.h: Delete.
16335 * config/rs6000/xfpu.md: Delete.
16336 * config/rs6000/xilinx.h: Delete.
16337 * config/rs6000/xilinx.opt: Delete.
16338 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
16339 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
16341 2018-05-04 Tom de Vries <tom@codesourcery.com>
16344 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
16347 2018-05-04 Richard Biener <rguenther@suse.de>
16349 PR middle-end/85627
16350 * tree-complex.c (update_complex_assignment): We are always in SSA form.
16351 (expand_complex_div_wide): Likewise.
16352 (expand_complex_operations_1): Likewise.
16353 (expand_complex_libcall): Preserve EH info of the original stmt.
16354 (tree_lower_complex): Handle removed blocks.
16355 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
16356 on complex multiplication and division libcall builtins.
16358 2018-05-04 Richard Biener <rguenther@suse.de>
16360 PR middle-end/85574
16361 * fold-const.c (negate_expr_p): Restrict negation of operand
16362 zero of a division to when we know that can happen without
16364 (fold_negate_expr_1): Likewise.
16366 2018-05-04 Jakub Jelinek <jakub@redhat.com>
16369 * real.h (real_nextafter): Declare.
16370 * real.c (real_nextafter): New function.
16371 * fold-const-call.c (fold_const_nextafter): New function.
16372 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
16373 CASE_CFN_NEXTTOWARD.
16374 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
16375 even when arg1_mode is different from arg0_mode.
16377 2018-05-03 Nathan Sidwell <nathan@acm.org>
16379 * doc/extend.texi (Deprecated Features): Remove
16380 -ffriend-injection.
16381 (Backwards Compatibility): Likewise.
16382 * doc/invoke.texi (C++ Language Options): Likewise.
16383 (C++ Dialect Options): Likewise.
16385 2018-05-03 Jakub Jelinek <jakub@redhat.com>
16388 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
16389 _mm512_mask_mullox_epi64): New intrinsics.
16391 2018-05-03 Tom de Vries <tom@codesourcery.com>
16394 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
16395 dump files): Add offload-tree.
16397 2018-05-03 Richard Biener <rguenther@suse.de>
16399 PR tree-optimization/85615
16400 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
16401 to loops not nested in BBs loop father to avoid creating multi-entry
16404 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16406 PR tree-optimization/70291
16407 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
16408 arguments. Change return type to tree. Emit libcall as a new
16409 statement rather than replacing existing one when inplace_p is true.
16410 (expand_complex_multiplication_components): New function.
16411 (expand_complex_multiplication): Expand floating-point complex
16412 multiplication using the above.
16413 (expand_complex_division): Rename inner_type parameter to type.
16414 Update expand_complex_libcall call-site.
16415 (expand_complex_operations_1): Update expand_complex_multiplication
16416 and expand_complex_division call-sites.
16418 2018-05-02 Jakub Jelinek <jakub@redhat.com>
16421 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
16422 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
16423 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
16424 the highest significant bit of the shift count mask is clear. In
16425 check whether and[sq]i3 is needed verify that all significant bits
16426 of the shift count other than the highest are set.
16428 2018-05-02 Tom de Vries <tom@codesourcery.com>
16431 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
16432 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
16433 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
16434 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
16435 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
16436 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
16437 __builtin_goacc_parlevel_size.
16439 2018-05-02 Richard Biener <rguenther@suse.de>
16441 PR tree-optimization/85597
16442 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
16443 do not use split vect_get_vec_defs call but call vect_get_slp_defs
16446 2018-05-02 Tom de Vries <tom@codesourcery.com>
16449 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
16450 dump files): Add ltrans-tree.
16452 2018-05-02 Tom de Vries <tom@codesourcery.com>
16455 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
16456 dump files): Add wpa-ipa.
16458 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
16460 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
16461 powerpc*-*-linux*paired* target.
16462 * config/rs6000/750cl.h: Delete.
16463 * config/rs6000/paired.h: Delete.
16464 * config/rs6000/paired.md: Delete.
16465 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
16467 * config/rs6000/rs6000-builtin.def: Remove paired float support.
16468 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
16469 comment. Remove paired float support.
16470 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
16471 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
16473 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
16474 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
16476 * config/rs6000/rs6000.c: Remove paired float support.
16477 (paired_expand_vector_init, paired_expand_vector_move,
16478 paired_emit_vector_compare, paired_emit_vector_cond_expr,
16479 (paired_expand_lv_builtin, paired_expand_stv_builtin,
16480 paired_expand_builtin, paired_expand_predicate_builtin,
16481 paired_init_builtins): Delete.
16482 * config/rs6000/rs6000.h: Remove paired float support.
16483 * config/rs6000/rs6000.md: Remove paired float support.
16484 (move_from_CR_ov_bit): Delete.
16485 * config/rs6000/rs6000.opt (mpaired): Delete.
16486 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
16487 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
16489 2018-05-02 Richard Biener <rguenther@suse.de>
16491 PR middle-end/85567
16492 * gimplify.c (gimplify_save_expr): When in SSA form allow
16493 SAVE_EXPRs to compute to SSA vars.
16495 2018-05-02 Jakub Jelinek <jakub@redhat.com>
16498 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
16499 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
16500 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
16501 clobber operands[2], instead use a new pseudo. Formatting fixes.
16503 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
16505 PR tree-optimization/85586
16506 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
16507 exit early for statements in the same group if the accesses are
16510 2018-05-02 Tom de Vries <tom@codesourcery.com>
16513 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
16516 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
16518 PR tree-optimization/85143
16519 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
16521 2018-05-01 Tom de Vries <tom@codesourcery.com>
16524 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
16525 not found" error message.
16527 2018-05-01 Tom de Vries <tom@codesourcery.com>
16530 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
16531 * vec.c (test_ordered_remove_if): New function.
16532 (vec_c_tests): Call test_ordered_remove_if.
16533 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
16534 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
16535 * tree-vect-patterns.c (vect_pattern_recog_1): Use
16536 VEC_ORDERED_REMOVE_IF.
16538 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16540 PR tree-optimization/82665
16541 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
16542 pointer subtraction where arguments come from a memchr call.
16544 2018-05-01 Jakub Jelinek <jakub@redhat.com>
16546 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
16547 --push-state --as-needed and --pop-state instead of --as-needed and
16548 --no-as-needed if ld supports it.
16549 * configure: Regenerated.
16552 * doc/install.texi2html: Replace _002d with - and _002a with * in
16553 generated html files using sed.
16555 2018-04-30 David Malcolm <dmalcolm@redhat.com>
16558 * gcc-rich-location.c (blank_line_before_p): New function.
16559 (use_new_line): New function.
16560 (gcc_rich_location::add_fixit_insert_formatted): New function.
16561 * gcc-rich-location.h
16562 (gcc_rich_location::add_fixit_insert_formatted): New function.
16564 2018-04-30 David Malcolm <dmalcolm@redhat.com>
16566 * selftest.c (assert_streq): Rename "expected" and "actual" to
16567 "val1" and "val2". Extend NULL-handling to cover both inputs
16568 symmetrically, while still requiring both to be non-NULL for a pass.
16569 * selftest.h (assert_streq): Rename "expected" and "actual" to
16571 (ASSERT_EQ): Likewise.
16572 (ASSERT_EQ_AT): Likewise.
16573 (ASSERT_KNOWN_EQ): Likewise.
16574 (ASSERT_KNOWN_EQ_AT): Likewise.
16575 (ASSERT_NE): Likewise.
16576 (ASSERT_MAYBE_NE): Likewise.
16577 (ASSERT_MAYBE_NE_AT): Likewise.
16578 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
16579 the assertion to pass.
16580 (ASSERT_STREQ_AT): Likewise.
16582 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
16584 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
16585 interaction with -pie.
16587 2018-04-30 David Malcolm <dmalcolm@redhat.com>
16589 * selftest.h: Fix alphabetization of per-source-file selftest
16592 2018-04-30 Jason Merrill <jason@redhat.com>
16594 PR c++/61982 - dead stores to destroyed objects.
16595 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
16598 2018-04-30 Jason Merrill <jason@redhat.com>
16600 * tree.c (build_clobber): New.
16601 * tree.h: Declare it.
16602 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
16604 2018-04-30 David Malcolm <dmalcolm@redhat.com>
16606 * diagnostic-show-locus.c (layout::layout): Update for
16607 location_get_source_line returning a char_span.
16608 (struct char_span): Move to input.h.
16609 (struct correction): Update for fields in char_span becoming
16611 (struct source_line): Update for location_get_source_line
16612 returning a char_span.
16613 (layout::print_line): Likewise.
16614 * edit-context.c (edited_file::print_content): Likewise.
16615 (edited_file::print_diff_hunk): Likewise.
16616 (edited_file::print_run_of_changed_lines): Likewise.
16617 (edited_file::get_num_lines): Likewise.
16618 (edited_line::edited_line): Likewise.
16619 * final.c (asm_show_source): Likewise.
16620 * input.c (location_get_source_line): Convert return type
16621 from const char * to char_span, losing the final "line_len"
16623 (dump_location_info): Update for the above.
16624 (get_substring_ranges_for_loc): Likewise. Use a char_span
16625 when handling the literal within the line.
16626 (test_reading_source_line): Update for location_get_source_line
16627 returning a char_span.
16628 * input.h (class char_span): Move here from
16629 diagnostic-show-locus.c, converting from a struct to a class.
16630 Make data members private.
16631 (char_span::operator bool): New.
16632 (char_span::length): New.
16633 (char_span::get_buffer): New.
16634 (char_span::operator[]): New.
16635 (char_span::subspan): Make const.
16636 (char_span::xstrdup): New.
16637 (location_get_source_line): Convert return type from const char *
16638 to char_span, losing the final "line_size" param.
16640 2018-04-30 Jan Hubicka <jh@suse.cz>
16642 * lto-wrapper.c (ltrans_priorities): New static var.
16643 (cmp_priority): New.
16644 (run_gcc): Read priorities and if doing parallel build order
16645 the Makefile by them.
16647 2018-04-30 David Malcolm <dmalcolm@redhat.com>
16649 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
16651 2018-04-30 Richard Biener <rguenther@suse.de>
16653 * tree-cfg.c (verify_address): Remove base argument, add
16654 flag whether to check TREE_ADDRESSABLE and do that.
16655 (verify_expr): Remove.
16656 (verify_types_in_gimple_reference): Add pieces from verify_expr.
16657 (verify_gimple_assign_single): Likewise.
16658 (verify_gimple_switch): Likewise.
16659 (verify_expr_location_1): Dereference tp once. Add (disabled)
16660 piece from verify_expr.
16661 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
16663 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
16665 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
16667 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
16669 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
16670 (small_data_pattern): Likewise.
16671 (arc_rewrite_small_data): Likewise.
16672 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
16673 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
16674 (get_symbol_alignment): New function.
16675 (legitimate_small_data_address_p): Likewise.
16676 (legitimate_scaled_address): Update, call
16677 legitimate_small_data_address_p.
16678 (output_sdata): New static variable.
16679 (arc_print_operand): Update how we handle small data operands.
16680 (arc_print_operand_address): Likewise.
16681 (arc_legitimate_address_p): Update, use
16682 legitimate_small_data_address_p.
16683 (arc_rewrite_small_data_p): Remove.
16684 (arc_rewrite_small_data_1): Likewise.
16685 (arc_rewrite_small_data): Likewise.
16686 (small_data_pattern): Likewise.
16687 (compact_sda_memory_operand): Update to use
16688 legitimate_small_data_address_p and get_symbol_alignment.
16689 (prepare_move_operands): Don't rewite sdata pattern.
16690 (prepare_extend_operands): Remove.
16691 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
16693 (zero_extendqisi2): Likewise.
16694 (zero_extendhisi2): Likewise.
16695 (extendqihi2): Likewise.
16696 (extendqisi2): Likewise.
16697 (extendhisi2): Likewise.
16698 (addsi3): Likewise.
16699 (subsi3): Likewise.
16700 (andsi3): Likewise.
16701 * config/arc/constraints.md (Usd): Change it to memory constraint.
16703 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
16705 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
16706 as source of std instructions.
16707 * config/arc/arc.md (movsi_insn): Update pattern predicate to
16708 allow 6-bit constants as source for store instructions.
16709 (movdi_insn): Update instruction pattern to allow 6-bit constants
16710 as source for store instructions.
16712 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
16714 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
16716 2018-04-30 Nathan Sidwell <nathan@acm.org>
16717 Sandra Loosemore <sandra@codesourcery.com>
16719 * dumpfile.c (dump_open): Allow '-' for stdout.
16720 * doc/invoke.texi (Developer Options): Document dump filename
16721 determination early. Document stdin/stdout selection.
16723 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
16725 Microblaze Target: PIC data text relative
16727 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
16728 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
16730 * config/microblaze/microblaze.h (microblaze_constant_address_p):
16731 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
16732 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
16733 New addressing mode for data-text relative position indepenedent code.
16734 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
16735 'ADDRESS_SYMBOLIC_TXT_REL'.
16736 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
16737 (microblaze_legitimate_pic_operand): Exclude function calls from
16738 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
16739 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
16741 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
16742 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
16743 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
16744 for 'address + offset'.
16745 (microblaze_expand_prologue): Add new function prologue call for
16747 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
16748 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
16749 table in case of TARGET_PIC_DATA_TEXT_REL.
16750 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
16751 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
16752 Add new macros 'UNSPEC_TEXT',
16753 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
16754 + exclude function calls from 'UNSPEC_PLT' in case of data text
16756 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
16757 new target hook for generating address diff vector tables in case of
16759 * doc/tm.texi : Regenerate.
16760 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
16761 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
16762 of addr diff vector generation.
16763 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
16764 target hook definition.
16765 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
16766 Add default function for generate_pic_addr_diff_vec -> flag_pic.
16767 * doc/invoke.texi (Add new pic option): Add new microblaze pic
16768 option for data text relative.
16770 2018-04-30 Richard Biener <rguenther@suse.de>
16772 * tree-chrec.h (evolution_function_is_constant_p): Remove
16774 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
16776 2018-04-30 Richard Biener <rguenther@suse.de>
16779 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
16781 2018-04-30 Richard Biener <rguenther@suse.de>
16783 PR tree-optimization/28364
16784 PR tree-optimization/85275
16785 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
16786 copying first exit test.
16788 2018-04-28 Mark Wielaard <mark@klomp.org>
16790 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
16791 dwarf_version >= 5.
16792 (dwarf_AT): Handle DW_AT_addr_base.
16793 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
16795 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
16798 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
16799 (*ashl<dwi>3_doubleword_mask_1): Ditto.
16800 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
16801 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
16803 2018-04-28 Richard Biener <rguenther@suse.de>
16805 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
16806 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
16807 to reflect use. Only add interesting stmts.
16809 2018-04-27 Martin Jambor <mjambor@suse.cz>
16812 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
16813 the jump function allows for passing through aggregate values.
16815 2018-04-27 David Malcolm <dmalcolm@redhat.com>
16817 * input.h (in_system_header_at): Convert from macro to inline
16819 (from_macro_expansion_at): Likewise.
16820 (from_macro_definition_at): Likewise.
16822 2018-04-27 Jeff Law <law@redhat.com>
16824 * config.gcc: Mark tile* targets as deprecated/obsolete.
16826 2018-04-27 Richard Biener <rguenther@suse.de>
16828 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
16831 2018-04-27 Richard Biener <rguenther@suse.de>
16833 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
16835 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
16837 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
16838 with Yd constraint. Set "preferred_for_speed" attribute from
16839 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
16840 with Yd constraint.
16841 (*movdi_internal): Ditto.
16842 (movti_interunit splitters): Remove
16843 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
16844 (movdi_interunit splitters): Ditto.
16845 * config/i386/constraints.md (Ye): Remove.
16846 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
16848 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16851 * config/aarch64/constraints.md (Usg): Limit to 31.
16852 (Usj): Limit to 63.
16854 2018-04-27 Jakub Jelinek <jakub@redhat.com>
16856 PR tree-optimization/85529
16857 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
16858 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
16859 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
16860 zero extension or masking of the MSB bit.
16861 (optimize_range_tests): Add FIRST_BB argument, pass it through
16862 to optimize_range_tests_var_bound.
16863 (maybe_optimize_range_tests, reassociate_bb): Adjust
16864 optimize_range_tests callers.
16866 2018-04-26 Richard Biener <rguenther@suse.de>
16867 Jakub Jelinek <jakub@redhat.com>
16869 * cgraph.h (symbol_table): Just declare debug method here.
16870 * symtab.c (symbol_table::debug): Define.
16872 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
16874 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
16876 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
16878 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
16879 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
16880 (*movdi_internal): Substitute Yi and Yj constraint with x
16881 and Ym and Yn constraint with y constraint. Update "isa"
16882 attribute and set "preferred_for_speed" attribute from
16883 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
16884 (*movsi_internal): Ditto.
16885 (*movdf_internal): Ditto.
16886 (*movsf_internal): Ditto.
16887 (*zero_extendsidi2): Ditto.
16888 * config/i386/sse.md (vec_set<mode>_0): Ditto.
16889 (sse2_loadld): Ditto.
16890 (*vec_extract<ssevecmodelower>_0): Ditto.
16891 (*vec_extractv4si_0_zext_sse4): Ditto.
16892 (vec_concatv2di): Ditto.
16893 (*vec_dup<mode>): Ditto.
16894 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
16895 * config/i386/constraints.md (Yi): Remove.
16900 2018-04-26 Nathan Sidwell <nathan@acm.org>
16902 * dumpfile.c (dump_open): New.
16903 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
16904 (dump_finish): Detect stdio/stderr by value not name.
16906 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
16908 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
16910 2018-04-26 Tom de Vries <tom@codesourcery.com>
16913 * config/nvptx/nvptx.c (verify_neutering_jumps)
16914 (verify_neutering_labels): New function
16915 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
16917 2018-04-26 Tom de Vries <tom@codesourcery.com>
16920 * config/nvptx/nvptx.c (needs_neutering_p): New function.
16921 (nvptx_single): Use needs_neutering_p to skip over insns that do not
16924 2018-04-26 Richard Biener <rguenther@suse.de>
16925 Tom de Vries <tom@codesourcery.com>
16928 * lto-streamer-out.c (output_function): Fixup loops if required to match
16929 discovery done in the reader.
16931 2018-04-26 Richard Biener <rguenther@suse.de>
16933 PR tree-optimization/85116
16934 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
16935 have a loop exit from the single latch predecessor. Remove
16936 case of header with just condition.
16937 (ch_base::copy_headers): Exclude infinite loops from any
16939 (pass_ch::execute): Record exits.
16941 2018-04-26 Richard Biener <rguenther@suse.de>
16943 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
16944 prologue cost vector and pass it to vect_get_load_cost.
16945 (vect_get_peeling_costs_all_drs): Likewise.
16946 (vect_peeling_hash_get_lowest_cost): Likewise.
16947 (vect_enhance_data_refs_alignment): Likewise.
16949 2018-04-26 Richard Biener <rguenther@suse.de>
16951 PR middle-end/85450
16952 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
16953 checking of integer<->pointer conversions.
16954 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
16955 sign-/zero-extending pointer types.
16956 (expand_omp_for_static_chunk): Likewise.
16958 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
16959 Jean Lee <xiaoyur347@gmail.com>
16961 * config/mips/mips.c (mips_asan_shadow_offset): New function.
16962 (TARGET_ASAN_SHADOW_OFFSET): Define.
16963 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
16964 true for -fsanitize=address.
16966 2018-04-25 Mark Wielaard <mark@klomp.org>
16968 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
16971 2018-04-25 Jakub Jelinek <jakub@redhat.com>
16973 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
16974 than "alu", remove explicit "memory" and "imm_disp" attributes.
16975 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
16977 PR middle-end/85414
16978 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
16979 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
16980 gen_lowpart_no_emit.
16982 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
16985 * config/i386/i386.c (ix86_expand_builtin): Change memory
16986 operand to XI, extend p0 to Pmode.
16987 * config/i386/i386.md: Change unspec volatile and operand
16988 1 mode to XI, change operand 0 mode to P.
16990 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
16992 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
16993 GET_MODE_MASK before any checking.
16994 (nds32_can_use_bset_p): Likewise.
16995 (nds32_can_use_btgl_p): Likewise.
16997 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
16999 * config/nds32/nds32-doubleword.md: New define_split pattern for
17000 illegal register number.
17002 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
17004 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
17006 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
17008 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
17010 2018-04-25 Richard Biener <rguenther@suse.de>
17012 * lto-streamer.h (LTO_major_version): Bump to 8.
17014 2018-04-25 Jakub Jelinek <jakub@redhat.com>
17016 * BASE-VER: Set to 9.0.0.
17018 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
17020 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
17021 in __abskf2 and __powikf2.
17023 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17026 * config/aarch64/constraints.md (Usg, Usj): New constraints.
17027 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
17028 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
17029 Use the above on operand 2. Reindent.
17030 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
17032 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
17035 * common/config/i386/i386-common.c (ix86_handle_option): Don't
17037 * config/i386/i386.opt (mcet): Removed.
17038 * doc/install.texi: Remove -mcet documentation.
17039 * doc/invoke.texi: Likewise.
17041 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
17044 * doc/install.texi: Remove -mcet from bootstrap-cet.
17046 2018-04-24 Jakub Jelinek <jakub@redhat.com>
17049 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
17050 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
17054 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
17055 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
17056 containing a CONST_VECTOR.
17058 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
17060 * doc/install.texi: Update newlib dependency for nvptx.
17062 2018-04-24 Jakub Jelinek <jakub@redhat.com>
17065 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
17066 instead of INTVAL when shifting x left.
17068 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
17070 PR tree-optimization/85478
17071 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
17072 vect_grouped_store_supported for single element vectors.
17074 2018-04-24 Richard Biener <rguenther@suse.de>
17077 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
17078 load cost increase to the case of non-constant step.
17080 2018-04-24 Jakub Jelinek <jakub@redhat.com>
17083 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
17084 destination if any_malformed_asm.
17086 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
17088 PR middle-end/85496
17089 * expr.c (store_field): In the bitfield case, if the value comes from
17090 a function call and is returned in registers by means of a PARALLEL,
17091 do not change the mode of the temporary unless BLKmode and VOIDmode.
17093 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
17095 PR rtl-optimization/85423
17096 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
17097 dependencies to debug insns when the previous insn is non-debug.
17099 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
17101 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
17102 enums into a single definition.
17103 (fls): Fix predicates and printing.
17106 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
17108 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
17109 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
17110 and short u6 immediate.
17111 (check_if_valid_sleep_operand): Remove.
17112 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
17114 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
17116 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
17117 flag_always_save_lp condition.
17118 * config/nds32/nds32.opt (malways-save-lp): New option.
17120 2018-04-22 Shiva Chen <shiva0217@gmail.com>
17122 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
17123 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
17124 * config/nds32/nds32.h
17125 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
17126 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
17128 2018-04-22 Shiva Chen <shiva0217@gmail.com>
17130 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
17131 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
17133 2018-04-22 Shiva Chen <shiva0217@gmail.com>
17134 Chung-Ju Wu <jasonwucj@gmail.com>
17136 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
17138 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
17139 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
17141 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
17143 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
17145 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
17147 * config/nds32/nds32-protos.h (nds32_data_alignment,
17148 nds32_local_alignment): Declare.
17149 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
17150 nds32_local_alignment): New functions.
17151 (TARGET_CONSTANT_ALIGNMENT): Define.
17152 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
17154 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
17156 * config/nds32/nds32.c
17157 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
17158 (TARGET_MODES_TIEABLE_P): Likewise.
17160 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
17162 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
17163 level Ofast and Og.
17165 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
17166 Chung-Ju Wu <jasonwucj@gmail.com>
17168 * config/nds32/constants.md (unspec_volatile_element): Add enum values
17169 for unaligned access.
17170 * config/nds32/nds32-intrinsic.c: Implementation of expanding
17172 * config/nds32/nds32-intrinsic.md: Likewise.
17173 * config/nds32/nds32_intrinsic.h: Likewise.
17174 * config/nds32/nds32.h (nds32_builtins): Likewise.
17175 * config/nds32/nds32.opt (munaligned-access): New option.
17176 * config/nds32/nds32.c (nds32_asm_file_start): Display
17177 flag_unaligned_access status.
17179 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
17181 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
17182 -mno-relax is present.
17183 * config/riscv/linux.h (LINK_SPEC): Ditto.
17185 2018-04-20 Martin Sebor <msebor@redhat.com>
17188 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
17190 (gimple_fold_builtin_stxcpy_chk): Same.
17191 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
17193 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
17196 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
17197 __powikf2 when long double is IEEE 128-bit.
17199 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
17201 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
17202 step to make sure stack always aligned.
17204 2018-04-20 Carl Love <cel@us.ibm.com>
17207 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
17208 size check for arg0.
17210 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
17211 Tom de Vries <tom@codesourcery.com>
17214 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
17215 Emit insns for calls too.
17216 (nvptx_find_par): Always look for worker-level predecessor insn.
17217 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
17219 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
17220 (nvptx_process_pars): Propagate frames for calls.
17222 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
17225 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
17227 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
17228 (ix86_handle_option): Don't handle OPT_mibt.
17229 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
17231 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
17233 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
17234 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
17235 (ix86_target_macros): Define __CET__ with flag_cf_protection
17236 for -fcf-protection.
17237 * config/i386/i386.c (isa2_opts): Remove -mibt.
17238 * config/i386/i386.h (TARGET_IBT): Removed.
17239 (TARGET_IBT_P): Likewise.
17240 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
17241 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
17242 * config/i386/i386.opt (mcet): Update help message.
17243 (mshstk): Likewise.
17245 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
17246 -mcet as an alias for -mshstk.
17248 2018-04-20 Richard Biener <rguenther@suse.de>
17250 PR middle-end/85475
17251 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
17252 complexity by forcing a single use of the multiply operand.
17254 2018-04-20 Martin Jambor <mjambor@suse.cz>
17257 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
17258 recursion dependency to only apply to non-clones.
17260 2018-04-20 Martin Jambor <mjambor@suse.cz>
17263 * ipa-cp.c (create_specialized_node): Check that clones of
17264 self-recursive edges exist during IPA-CP.
17266 2018-04-19 Toon Moene <toon@moene.org>
17268 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
17271 2018-04-19 Jakub Jelinek <jakub@redhat.com>
17273 PR tree-optimization/85467
17274 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
17275 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
17276 VECTOR_CST element to type.
17278 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
17281 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
17282 * config/i386/i386.md (builtin_setjmp_setup): Removed.
17283 (builtin_longjmp): Likewise.
17284 (save_stack_nonlocal): New pattern.
17285 (restore_stack_nonlocal): Likewise.
17287 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
17290 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
17291 Replace ASM_OUTPUT_LABEL with fprintf.
17293 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
17296 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
17297 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
17298 * config/i386/i386-c.c (ix86_target_macros_internal): Also
17299 define __IBT__ and __SHSTK__ for -fcf-protection.
17300 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
17302 (ix86_trampoline_init): Likewise.
17303 (x86_output_mi_thunk): Likewise.
17304 (ix86_notrack_prefixed_insn_p): Likewise.
17305 (ix86_option_override_internal): Don't disallow -fcf-protection.
17306 * config/i386/i386.md (rdssp<mode>): Also enable for
17308 (incssp<mode>): Likewise.
17309 (nop_endbr): Likewise.
17310 * config/i386/i386.opt (mcet): Change help message to built-in
17313 (mshstk): Likewise.
17314 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
17315 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
17316 enable CET built-in functions.
17318 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
17320 * common/config/i386/i386-common.c
17321 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
17322 OPTION_MASK_ISA_MOVDIRI_UNSET,
17323 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
17324 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
17325 * config.gcc (movdirintrin.h): New header.
17326 * config/i386/cpuid.h (bit_MOVDIRI,
17327 bit_MOVDIR64B): New bits.
17328 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
17330 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
17331 (VOID, PVOID, PCVOID)): New function types.
17332 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
17333 __builtin_ia32_directstoreu_u64,
17334 __builtin_ia32_movdir64b): New builtins.
17335 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
17336 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
17338 (ix86_valid_target_attribute_inner_p): Ditto.
17339 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
17340 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
17341 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
17342 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
17343 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
17344 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
17345 (movdiri<mode>, movdir64b_<mode>): New.
17346 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
17347 * config/i386/immintrin.h: Include movdirintrin.h.
17348 * config/i386/movdirintrin.h: New file.
17349 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
17351 2018-04-19 Richard Biener <rguenther@suse.de>
17353 PR middle-end/85455
17354 * cfg.c (clear_bb_flags): When loop state says we have
17355 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
17357 2018-04-19 Richard Biener <rguenther@suse.de>
17359 PR tree-optimization/84737
17360 * tree-vect-data-refs.c (vect_copy_ref_info): New function
17361 copying restrict info.
17362 (vect_setup_realignment): Use it.
17363 * tree-vectorizer.h (vect_copy_ref_info): Declare.
17364 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
17365 the first DR to all generated stores.
17366 (vectorizable_load): Likewise for loads.
17368 2018-04-19 Jakub Jelinek <jakub@redhat.com>
17370 PR tree-optimization/85446
17371 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
17372 the integral and pointer types to have the same precision.
17374 * doc/install.texi: Document --disable-cet being the default and
17377 2018-04-18 Martin Liska <mliska@suse.cz>
17379 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
17382 2018-04-18 Martin Liska <mliska@suse.cz>
17385 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
17388 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
17389 arguments if they are comparable.
17391 2018-04-18 Martin Liska <mliska@suse.cz>
17394 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
17397 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
17400 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
17403 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
17404 ENDBR after calling __morestack.
17406 2018-04-18 David Malcolm <dmalcolm@redhat.com>
17409 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
17410 by using gcc_base_ver to generate a gcc_driver_version, and use
17411 it when generating GCC_DRIVER_NAME.
17412 * configure: Regenerate.
17414 2018-04-18 Jakub Jelinek <jakub@redhat.com>
17417 * config.gcc: Obsolete powerpc*-*-*spe*.
17419 2018-04-17 Jakub Jelinek <jakub@redhat.com>
17422 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
17423 (stabstr_D): Change type of unum from unsigned int to
17424 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
17427 2018-04-17 Jim Wilson <jimw@sifive.com>
17430 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
17431 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
17432 Set arg_pointer_offset after using pretend_args_size.
17434 2018-04-17 Jakub Jelinek <jakub@redhat.com>
17436 PR rtl-optimization/85431
17437 * dse.c (record_store): Ignore zero width stores.
17440 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
17441 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
17442 __builtin_stack_restore rather than after it.
17443 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
17444 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
17445 argument instead of virtual_dynamic_stack_rtx.
17447 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
17449 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
17451 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17452 Add note to error message to explain internal mapping of overloaded
17453 built-in function name to non-overloaded built-in function name.
17454 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
17457 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
17460 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
17461 where the inputs overlap with the output.
17463 2018-04-17 Jakub Jelinek <jakub@redhat.com>
17465 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
17466 (=v, v) alternative and explicit "memory" attribute.
17467 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
17468 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
17470 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
17471 "sselog1" type instead of "sselog".
17472 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
17473 "sselog". Remove explicit "memory" attribute.
17474 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
17475 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
17477 (vec_extract_hi_v32hi): Merge all alternatives into one, use
17478 "sselog1" type instead of "sselog". Remove explicit "memory"
17480 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
17481 use "sselog1" type instead of "sselog". Remove explicit "memory"
17483 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
17484 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
17486 (vec_extract_hi_v64qi): Merge all alternatives into one, use
17487 "sselog1" type instead of "sselog". Remove explicit "memory"
17489 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
17490 use "sselog1" type instead of "sselog". Remove explicit "memory"
17494 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
17496 PR middle-end/85414
17497 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
17500 2018-04-17 Martin Jambor <mjambor@suse.cz>
17503 * ipa-cp.c (create_specialized_node): Call
17504 expand_all_artificial_thunks if necessary.
17506 2018-04-17 Martin Liska <mliska@suse.cz>
17509 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
17510 in message, remote space in between '_G' and '('.
17512 2018-04-17 Jakub Jelinek <jakub@redhat.com>
17515 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
17516 avx512f_vmcmp<mode>3<round_saeonly_name>,
17517 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
17518 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
17519 avx512f_rndscale<mode><round_saeonly_name>,
17520 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
17521 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
17522 Use %<iptr>2 instead of %2 for -masm=intel.
17523 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
17524 avx512f_vcvttss2usi<round_saeonly_name>,
17525 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
17527 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
17528 avx512f_vcvttsd2usi<round_saeonly_name>,
17529 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
17530 Use %q1 instead of %1 for -masm=intel.
17531 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
17532 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
17533 of %3 for -masm=intel.
17534 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
17536 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
17538 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
17540 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
17541 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
17543 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
17545 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
17546 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
17547 %g1 and one with %0 and %1.
17548 (avx512er_vmrcp28<mode><round_saeonly_name>,
17549 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
17550 %1 for -masm=intel.
17551 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
17552 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
17553 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
17554 of %0 and %{%4%} for -masm=intel.
17555 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
17556 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
17557 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
17558 order of %0 and %{%5%}%{z%} for -masm=intel.
17560 2018-04-17 Jan Hubicka <jh@suse.cz>
17563 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
17565 2018-04-17 Martin Liska <mliska@suse.cz>
17568 * multiple_target.c (create_dispatcher_calls): Set apostrophes
17569 for target_clone error message. Make default implementation
17570 clone to be a local declaration.
17571 (separate_attrs): Add new argument and check for an empty
17573 (expand_target_clones): Handle it.
17574 (ipa_target_clone): Make redirection just for target_clones
17577 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
17578 Tom de Vries <tom@codesourcery.com>
17580 PR middle-end/84955
17581 * omp-expand.c (expand_oacc_for): Add dummy false branch for
17582 tiled basic blocks without omp continue statements.
17584 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
17587 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
17588 vec_extract expression as having side effects to make sure it gets
17591 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
17594 * config/i386/i386.c (get_builtin_code_for_version): Check
17597 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
17600 * config.gcc: Support "skylake".
17601 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
17603 * config/i386/i386.c (m_SKYLAKE): Define.
17604 (processor_target_table): Add "skylake".
17605 (ix86_option_override_internal): Add "skylake".
17606 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
17607 PROCESSOR_CANNONLAKE.
17608 (get_builtin_code_for_version): Fix priority for
17609 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
17610 PROCESSOR_SKYLAKE-AVX512.
17611 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
17612 (processor_type): Add PROCESSOR_SKYLAKE.
17614 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
17615 Jason Merrill <jason@redhat.com>
17618 * convert.c (convert_to_integer_1): Use direct recursion for
17619 enumeral types and types with a precision less than the number
17620 of bits in their mode.
17622 2018-04-16 Julia Koval <julia.koval@intel.com>
17625 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
17626 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
17628 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
17631 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
17632 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
17633 and -mno-direct-move.
17635 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
17638 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
17639 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
17640 Ensure negative shifts result in {0}.
17642 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
17644 PR rtl-optimization/79916
17645 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
17646 regs (if any) to define how to gnerate SD moves when LRA is in
17649 2018-04-13 Jakub Jelinek <jakub@redhat.com>
17651 PR rtl-optimization/85393
17652 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
17653 * except.c (expand_dw2_landing_pad_for_region): Make static.
17654 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
17655 a label and unconditional jump to old_bb, rather than
17656 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
17659 PR rtl-optimization/85376
17660 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
17661 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
17662 instead of a specific value.
17664 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
17665 Bin Cheng <bin.cheng@arm.com>
17667 PR tree-optimization/82965
17668 PR tree-optimization/83991
17669 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
17670 by_profile_only parameter.
17671 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
17672 information if the loop was predicted to iterate too many times.
17673 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
17675 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
17678 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
17681 2018-04-13 Martin Liska <mliska@suse.cz>
17682 Jakub Jelinek <jakub@redhat.com>
17684 PR middle-end/81657
17685 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
17686 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
17687 * builtins.c (expand_builtin_memory_copy_args): Use
17688 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
17689 handle dest_addr == pc_rtx.
17691 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
17694 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
17695 asked to not generate direct moves.
17696 (fix_trunc<mode>si2_stfiwx): Similar.
17697 (fix_trunc<mode>si2_internal): Similar.
17699 2018-04-12 Jakub Jelinek <jakub@redhat.com>
17702 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
17703 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
17704 lookup if dest in some wider mode is known to be const0_rtx and
17705 if so, record permanent equivalence for it to be ZERO_EXTEND of
17706 the narrower mode destination.
17708 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
17710 * lto-streamer-out.c (output_function): Revert 259346.
17711 * omp-expand.c (expand_oacc_for): Likewise.
17713 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
17715 PR rtl-optimization/85354
17716 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
17717 * sel-sched.c (sel_global_init): ... here.
17719 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
17722 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
17723 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
17724 mode for PE-COFF targets.
17725 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
17726 (i386_pe_asm_lto_end): Likewise.
17727 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
17728 (TARGET_ASM_LTO_END): Likewise.
17729 * config/i386/winnt.c (saved_debug_info_level): New static variable.
17730 (i386_pe_asm_lto_start): New function.
17731 (i386_pe_asm_lto_end): Likewise.
17733 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
17734 Richard Biener <rguenther@suse.de>
17736 PR middle-end/84955
17737 * lto-streamer-out.c (output_function): Fix CFG loop state before
17739 * omp-expand.c (expand_oacc_for): Handle calls to internal
17740 functions like regular functions.
17742 2018-04-12 Richard Biener <rguenther@suse.de>
17745 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
17746 for the early LTO debug to properly generate references to it
17747 during DIE emission. Do not re-use that for the skeleton for
17749 (dwarf2out_early_finish): Likewise.
17751 2018-04-12 Jakub Jelinek <jakub@redhat.com>
17754 * config/i386/sse.md
17755 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
17756 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
17757 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
17758 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
17759 and output is a reg, avoid creating invalid lowpart subreg, but
17760 instead split into a 512-bit move. Don't split if not AVX512VL,
17761 input is xmm16+ reg and output is a mem.
17762 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
17763 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
17764 xmm16+ reg and output is a mem.
17766 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17768 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
17769 also for flag_dwarf2_cfi_asm.
17771 2018-04-12 Jakub Jelinek <jakub@redhat.com>
17773 PR rtl-optimization/85342
17774 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
17775 a bool scalar var inside of the loop instead. Don't try to update
17776 recog_data.operand after failed apply_change_group.
17778 2018-04-12 Tom de Vries <tom@codesourcery.com>
17781 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
17782 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
17783 array with flexible array member as array without given dimension.
17784 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
17785 argument for undefined param to true.
17787 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
17790 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
17791 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
17792 from PowerPC section.
17793 * config/rs6000/sysv4.opt (mcall-): Improve help text.
17794 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
17795 help text that is too long.
17796 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
17797 help text that is too long.
17798 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
17799 help text that is too long.
17801 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
17803 * config/alpha/alpha.md (stack_probe_internal): Rename
17804 from "probe_stack". Update all callers.
17806 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
17808 PR rtl-optimization/84566
17809 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
17810 sched_macro_fuse_insns.
17812 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
17815 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
17816 (compute_block_dependences): ... from here.
17818 2018-04-11 Jakub Jelinek <jakub@redhat.com>
17820 PR tree-optimization/85331
17821 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
17822 from int to HOST_WIDE_INT.
17824 2018-04-11 Martin Jambor <mjambor@suse.cz>
17827 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
17828 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
17829 not the same as the source val.
17830 (cgraph_edge_brings_value_p): New parameter.
17831 (gather_edges_for_value): Pass destination value to
17832 cgraph_edge_brings_value_p.
17833 (perhaps_add_new_callers): Likewise.
17834 (get_info_about_necessary_edges): Likewise and exclude values brought
17835 only by self-recursive edges.
17836 (create_specialized_node): Redirect only clones of self-calling edges.
17837 (+self_recursive_pass_through_p): New function.
17838 (find_more_scalar_values_for_callers_subset): Use it.
17839 (find_aggregate_values_for_callers_subset): Likewise.
17840 (known_aggs_to_agg_replacement_list): Removed.
17841 (decide_whether_version_node): Re-calculate known constants for all
17842 remaining context clones.
17844 2018-04-11 Richard Biener <rguenther@suse.de>
17847 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
17848 from early DWARF output.
17849 (dwarf2out_early_finish): Output line info unconditionally into
17850 early DWARF and add reference to it.
17852 2018-04-11 Jakub Jelinek <jakub@redhat.com>
17855 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
17856 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
17857 other than V2DFmode using iptr mode attribute.
17858 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
17860 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
17862 PR rtl-optimization/84659
17863 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
17865 2018-04-11 Jakub Jelinek <jakub@redhat.com>
17868 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
17870 (output_loc_list): Pass address of a dummy size variable even in the
17871 locview handling loop.
17872 (index_location_lists): Add comment on why skip_loc_list_entry can't
17875 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
17878 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
17881 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
17884 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
17885 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
17886 and -mstring-compare-inline-limit.
17888 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
17891 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
17892 for stack clash protection in a register whenever we need it to be in
17895 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
17897 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
17898 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
17900 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
17903 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
17905 (mlong-double-): Ditto.
17906 * config/rs6000/sysv4.opt (msdata=): Ditto.
17907 (mtls-size=): Ditto.
17909 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
17911 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
17912 erroneous entries for
17913 "vector int vec_ldl (int, long int *)", and
17914 "vector unsigned int vec_ldl (int, unsigned long int *)".
17915 Add comments and entries for
17916 "vector bool char vec_ldl (int, bool char *)",
17917 "vector bool short vec_ldl (int, bool short *)",
17918 "vector bool int vec_ldl (int, bool int *)",
17919 "vector bool long long vec_ldl (int, bool long long *)",
17920 "vector pixel vec_ldl (int, pixel *)",
17921 "vector long long vec_ldl (int, long long *)",
17922 "vector unsigned long long vec_ldl (int, unsigned long long *)".
17923 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
17924 type tree bool_long_long_type_node and correct definition of
17925 bool_V2DI_type_node to make reference to this new type tree.
17926 (rs6000_mangle_type): Replace erroneous reference to
17927 bool_long_type_node with bool_long_long_type_node.
17928 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
17929 comments to emphasize sign distinctions for char and int types and
17930 replace RS6000_BTI_bool_long constant with
17931 RS6000_BTI_bool_long_long constant. Also add comment to restrict
17932 use of RS6000_BTI_pixel.
17933 (bool_long_type_node): Remove this macro definition.
17934 (bool_long_long_type_node): New macro definition
17936 2018-04-10 Jakub Jelinek <jakub@redhat.com>
17938 PR rtl-optimization/85300
17939 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
17940 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
17941 simplify_unary_operation fails.
17943 2018-04-10 Martin Liska <mliska@suse.cz>
17945 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
17946 cgraph_edge and ipa_ref.
17948 2018-04-10 Jakub Jelinek <jakub@redhat.com>
17952 * config/i386/sse.md
17953 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
17954 computation of the VEC_MERGE selector from mask.
17955 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
17956 Fix decoding of the VEC_MERGE selector into mask.
17958 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
17960 PR tree-optimization/85286
17961 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
17963 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
17965 * final.c (final_1): Set insn_last_address as well as
17966 insn_current_address.
17968 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17971 * explow.c (emit_stack_probe): Call validize_mem on memory location
17972 before passing it to gen_probe_stack. Create address operand and
17973 legitimize it for the probe_stack_address case.
17975 2018-04-09 Jan Hubicka <jh@suse.cz>
17978 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
17979 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
17980 * tree.c (free_lang_data_in_type): Fix handling of binfos;
17982 (free_lang_data): Rebuild type inheritance graph.
17984 2018-04-09 Martin Sebor <msebor@redhat.com>
17986 * invoke.texi (-finline-small-functions): Mention other optimization
17988 (-findirect-inlining, -fpartial-inlining): Same.
17989 (-finline-functions-called-once): Same.
17990 (-freorder-blocks-and-partition): Same.
17992 2018-04-09 Jan Hubicka <jh@suse.cz>
17995 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
17996 jumps; choose last target that matches the criteria (i.e.
17997 no partition changes for non-crossing jumps).
17998 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
17999 support for redirecting crossing jumps to non-crossing.
18001 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
18003 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
18004 also for naked functions.
18006 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
18008 * config/arc/arc.md (add_shift): New pattern.
18009 (add_shift2): Likewise.
18010 (sub_shift): Likewise.
18011 (sub_shift_cmp0_noout): Likewise.
18012 (compare_si_ashiftsi): Likewise.
18013 (xbfu_cmp0_noout): New combine pattern.
18014 (xbfu_cmp0"): Likewise.
18015 (movsi_set_cc_insn): Place the predicable variant first.
18016 (commutative_binary_cmp0_noout): Remove clobber.
18017 (commutative_binary_cmp0): New pattern.
18018 (noncommutative_binary_cmp0): Likewise.
18019 (noncommutative_binary_cmp0_noout): Likewise.
18020 (noncommutative_binary_comparison_result_used): Removed.
18021 (rsub_cmp0): New pattern.
18022 (rsub_cmp0_noout): Likewise.
18023 (extzvsi): Changed, keep only meaningful variants.
18024 (SQH, SEZ): New iterators.
18025 (SQH_postfix): New mode attribute.
18026 (SEZ_prefix): New code attribute.
18027 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
18028 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
18029 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
18030 of numerical value.
18031 (noncommutative_operator): Check the availability of barrel
18034 2018-04-09 Richard Biener <rguenther@suse.de>
18036 PR tree-optimization/85284
18037 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
18038 Only use the niter constraining form of simple_iv when the exit
18039 is always executed.
18041 2018-04-09 Tom de Vries <tom@codesourcery.com>
18044 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
18045 (define_expand "*memory_barrier"): New define_expand.
18046 (define_insn "memory_barrier"): New insn.
18048 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
18050 PR rtl-optimization/80463
18051 PR rtl-optimization/83972
18052 PR rtl-optimization/83480
18054 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
18055 correct producer for the insn.
18056 (tidy_control_flow): Fixup seqnos in case of debug insns.
18058 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
18060 PR rtl-optimization/83913
18062 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
18063 different sched-times when merging exprs.
18065 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
18067 PR rtl-optimization/83962
18069 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
18070 tidy_fallthru_edge and tidy_control_flow.
18072 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
18074 PR rtl-optimization/83530
18076 * sel-sched.c (force_next_insn): New global variable.
18077 (remove_insn_for_debug): When force_next_insn is true, also leave only
18078 next insn in the ready list.
18079 (sel_sched_region): When the region wasn't scheduled, make another pass
18080 over it with force_next_insn set to 1.
18082 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
18084 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
18086 * config/nds32/constants.md (unspec_volatile_element): Add enum values
18087 for interrupt control.
18088 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
18089 functions for interrupt control.
18090 * config/nds32/nds32-intrinsic.md: Likewise.
18091 * config/nds32/nds32_intrinsic.h: Likewise.
18092 * config/nds32/nds32.h (nds32_builtins): Likewise.
18094 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
18096 * config/nds32/nds32.c (nds32_init_machine_status,
18097 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
18098 strict_aligned_p field.
18099 (nds32_expand_to_rtl_hook): New function.
18100 (TARGET_EXPAND_TO_RTL_HOOK): Define.
18101 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
18103 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
18104 Chung-Ju Wu <jasonwucj@gmail.com>
18106 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
18107 * config/nds32/nds32-n7.md: New file.
18108 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
18109 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
18111 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
18112 * config/nds32/nds32.md (pipeline_model): Add n7.
18113 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
18114 * config/nds32/pipelines.md: Include n7 settings.
18116 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
18117 Chung-Ju Wu <jasonwucj@gmail.com>
18119 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
18120 * config/nds32/nds32-e8.md: New file.
18121 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
18122 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
18124 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
18125 * config/nds32/nds32.md (pipeline_model): Add e8.
18126 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
18127 * config/nds32/pipelines.md: Include e8 settings.
18129 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
18130 Chung-Ju Wu <jasonwucj@gmail.com>
18132 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
18133 * config/nds32/nds32-n8.md: New file.
18134 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
18135 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
18137 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
18138 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
18139 * config/nds32/nds32.md (pipeline_model): Add n8.
18140 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
18141 * config/nds32/pipelines.md: Include n8 settings.
18143 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
18144 Chung-Ju Wu <jasonwucj@gmail.com>
18146 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
18147 * config/nds32/nds32-n9-2r1w.md: New file.
18148 * config/nds32/nds32-n9-3r2w.md: New file.
18149 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
18150 nds32_register_ports): New or modify for cpu n9.
18151 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
18153 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
18154 * config/nds32/nds32-utils.c: New file.
18155 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
18156 TARGET_MUL_SLOW): Define.
18157 * config/nds32/nds32.md (pipeline_model): New attribute.
18158 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
18159 New options that support cpu n9.
18160 * config/nds32/pipelines.md: Include n9 settings.
18161 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
18163 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
18165 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
18166 information if necessary.
18167 (output_cond_branch_compare_zero): Likewise.
18168 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
18169 (nds32_target_alignment): Refine for alignment.
18170 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
18171 (FUNCTION_BOUNDARY): Modify.
18172 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
18174 * config/nds32/nds32.opt (malways-align, malign-functions): New.
18176 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
18178 * config/nds32/constants.md (unspec_volatile_element): Add values for
18179 TLB operation and data prefetch.
18180 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
18181 functions for TLB operation and data prefetch.
18182 * config/nds32/nds32-intrinsic.md: Likewise.
18183 * config/nds32/nds32_intrinsic.h: Likewise.
18184 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
18185 (nds32_print_operand): Likewise.
18186 * config/nds32/nds32.h (nds32_builtins): Likewise.
18188 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
18189 Andrew Pinski <pinsika@gcc.gnu.org>
18191 PR middle-end/82976
18192 * match.pd: Use constant_boolean_node of correct type instead of
18193 boolean_true_node or boolean_false_node for simplifying
18194 pointer comparisons to zero.
18196 2018-04-07 Jakub Jelinek <jakub@redhat.com>
18198 PR tree-optimization/80021
18199 * tree.c (verify_type_variant): Make error call in verify_variant_match
18200 translatable and remove final full stop.
18202 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
18204 * config/nds32/constants.md (unspec_volatile_element): Add
18205 UNSPEC_VOLATILE_EH_RETURN.
18206 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
18207 nds32_output_stack_pop): Support dwarf exception handling process.
18208 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
18209 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
18210 exception handling process.
18211 (nds32_compute_stack_frame): Likewise.
18212 (nds32_return_addr_rtx): Likewise.
18213 (nds32_initial_elimination_offset): Likewise.
18214 (nds32_expand_prologue): Likewise.
18215 (nds32_expand_epilogue): Likewise.
18216 (nds32_dynamic_chain_address): New function.
18217 * config/nds32/nds32.h (machine_function): Add fields for dwarf
18218 exception handling.
18219 (DYNAMIC_CHAIN_ADDRESS): Define.
18220 (EH_RETURN_DATA_REGNO): Define.
18221 (EH_RETURN_STACKADJ_RTX): Define.
18222 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
18223 patterns for dwarf exception handling.
18225 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
18227 * config/nds32/nds32.h: Clean up obsolete macros.
18229 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
18231 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
18232 Add enum values for particular instructions.
18233 * config/nds32/nds32-intrinsic.c: Implementation of expanding
18234 particular intrinsic functions.
18235 * config/nds32/nds32-intrinsic.md: Likewise.
18236 * config/nds32/nds32_intrinsic.h: Likewise.
18237 * config/nds32/nds32.h (nds32_builtins): Likewise.
18238 * config/nds32/nds32.md (type): Add pbsad and pbsada.
18239 (btst, ave): New patterns for particular instructions.
18241 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
18243 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
18244 Add enum values for atomic load/store and memory sync.
18245 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
18247 * config/nds32/nds32-intrinsic.md: Likewise.
18248 * config/nds32/nds32_intrinsic.h: Likewise.
18249 * config/nds32/nds32.h (nds32_builtins): Likewise.
18251 2018-04-07 Jakub Jelinek <jakub@redhat.com>
18253 PR tree-optimization/85257
18254 * fold-const.c (native_encode_vector): If not all elts could fit
18255 and off is -1, return 0 rather than offset.
18256 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
18257 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
18258 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
18259 adjust buffer in native_interpret_expr call.
18261 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
18263 * config/nds32/constants.md (unspec_volatile_element): Add cache
18264 control enum values.
18265 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
18266 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
18267 * config/nds32/nds32.c (nds32_cctl_names): New.
18268 (nds32_print_operand): Handle cache control register names.
18269 * config/nds32/nds32.h (nds32_builtins): New enum values.
18270 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
18272 * config/nds32/nds32.md (type): Add mmu.
18273 * config/nds32/pipelines.md (simple_insn): Add mmu.
18275 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
18277 * config/nds32/nds32.md (type): Remove call.
18278 * config/nds32/pipelines.md (simple_insn): Likewise.
18280 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
18282 * config/nds32/constants.md (unspec_volatile_element): Add
18283 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
18284 UNSPEC_VOLATILE_FMFCFG.
18285 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
18286 description for fmfcfg and fmfcsr.
18287 (bdesc_1arg): Add fmtcsr.
18288 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
18289 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
18290 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
18291 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
18292 unspec_fmfcfg): New patterns.
18293 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
18294 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
18295 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
18296 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
18297 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
18298 __nds32__fmfcfg): Define.
18300 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
18302 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
18303 intrinsic register names.
18304 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
18305 intrinsic register enum values and macros.
18307 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
18309 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
18310 for load/store addressing form.
18311 (nds32_print_operand_address): Likewise.
18313 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
18316 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
18317 based on LABEL_REF. Remove useless assertion.
18318 (pic_address_needs_scratch): Fix formatting.
18319 (sparc_legitimize_pic_address): Minor tweaks.
18320 (sparc_delegitimize_address): Adjust assertion accordingly.
18321 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
18322 into symbolic_operand.
18323 (movsi_high_pic_label_ref): Likewise.
18324 (movsi_lo_sum_pic_label_ref): Likewise.
18325 (movdi_pic_label_ref): Likewise.
18326 (movdi_high_pic_label_ref): Likewise.
18327 (movdi_lo_sum_pic_label_ref): Likewise.
18329 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
18331 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
18332 custom LIB_SPEC setup.
18334 2018-04-06 Ruslan Bukin <br@bsdpad.com>
18335 Kito Cheng <kito.cheng@gmail.com>
18337 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
18338 * config/riscv/freebsd.h: New.
18340 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
18342 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
18343 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
18346 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
18347 Kito Cheng <kito.cheng@gmail.com>
18349 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
18350 nds32_output_call, nds32_symbol_binds_local_p): New functions.
18351 * config/nds32/nds32-protos.h (nds32_output_call,
18352 nds32_output_return): Declare.
18353 * config/nds32/nds32.md: Refine all the call and return patterns.
18355 2018-04-06 Jakub Jelinek <jakub@redhat.com>
18358 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
18359 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
18361 PR rtl-optimization/84872
18362 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
18363 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
18364 EDGE_CROSSING edge.
18366 2018-04-06 Tamar Christina <tamar.christina@arm.com>
18368 * expr.c (copy_blkmode_to_reg): Revert 254862.
18369 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
18371 2018-04-06 Richard Biener <rguenther@suse.de>
18373 PR middle-end/85244
18374 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
18375 after seeing a component reference with an adjacent field. Treat
18376 refs to arrays at struct end of external decls similar to
18377 refs to unconstrained commons.
18379 2018-04-06 Jakub Jelinek <jakub@redhat.com>
18382 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
18383 look through SAVE_EXPRs with non-side-effects argument. Adjust
18385 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
18388 2018-04-06 Richard Biener <rguenther@suse.de>
18390 PR middle-end/85180
18391 * alias.c (find_base_term): New wrapper around find_base_term
18392 unwinding CSELIB_VAL_PTR changes.
18393 (find_base_term): Do not restore CSELIB_VAL_PTR during the
18396 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18398 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
18400 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
18401 constant definitions.
18402 ("nop"): lr 0,0 -> nopr r0
18403 ("nop_lr0", "nop_lr1"): New insn definitions.
18405 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
18407 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
18408 NDS32_V3PUSH_AVAILABLE_P macro.
18410 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
18411 Chung-Ju Wu <jasonwucj@gmail.com>
18413 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
18414 (nds32*-*-*): Add float and fpu_config into supported_defaults.
18415 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
18416 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
18417 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
18418 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
18419 * config/nds32/constraints.md: New constraints and checking for hard
18420 float configuration.
18421 * config/nds32/iterators.md: New mode iterator and attribute for hard
18422 float configuration.
18423 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
18425 * config/nds32/nds32-fpu.md: New file.
18426 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
18427 deal with hard float code generation.
18428 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
18430 (abi_type, float_reg_number): New enum type.
18431 * config/nds32/nds32-predicates.c: New predicates for hard float.
18432 * config/nds32/nds32-protos.h: Declare functions for hard float.
18433 * config/nds32/nds32.c: Implementation for hard float configuration.
18434 * config/nds32/nds32.h: Definitions for hard float configuration.
18435 * config/nds32/nds32.md: Include hard float machine description and
18436 modify patterns for hard float configuration.
18437 * config/nds32/nds32.opt: New options for hard float configuration.
18438 * config/nds32/predicates.md: New predicates for hard float
18441 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
18443 * common/config/nds32/nds32-common.c
18444 (nds32_option_optimization_table): Enable -mreleax-hint by default.
18446 2018-04-05 Jakub Jelinek <jakub@redhat.com>
18448 PR middle-end/85195
18449 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
18450 CONSTRUCTOR_ELT (ctor, ...)->value.
18452 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
18455 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
18457 2018-04-05 Tom de Vries <tom@codesourcery.com>
18460 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
18463 2018-04-05 Shiva Chen <shiva0217@gmail.com>
18464 Kito Cheng <kito.cheng@gmail.com>
18466 * config/nds32/constraints.md (U33): Fine-tune checking condition.
18467 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
18468 * config/nds32/nds32.h (nds32_16bit_address_type): Add
18469 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
18471 2018-04-05 Shiva Chen <shiva0217@gmail.com>
18472 Kito Cheng <kito.cheng@gmail.com>
18474 * config/nds32/constraints.md (Ufe): New memory constraint.
18475 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
18476 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
18477 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
18479 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
18480 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
18482 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
18484 * config/nds32/nds32.md: Use optimize_size in the condition for
18485 alu-shift instructions.
18487 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
18489 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
18491 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
18493 * config/nds32/nds32.md (negsi2): Refine pattern.
18495 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
18496 Chung-Ju Wu <jasonwucj@gmail.com>
18498 * config/nds32/iterators.md (shift_rotate): New code iterator.
18499 (shift): New code attribute.
18500 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
18501 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
18502 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
18503 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
18504 bit-wise operations.
18505 (andsi3, *andsi3): Ditto.
18506 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
18507 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
18508 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
18509 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
18510 nds32_ior_operand, nds32_xor_operand): New predicates.
18512 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
18514 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
18515 (addsi3, subsi3): ... this.
18517 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
18519 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
18521 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
18523 * config/nds32/nds32.md: Adjust indention.
18525 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
18527 * config/nds32/nds32.md (feature): New attribute.
18529 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
18531 * config/nds32/nds32.md (subtype): New attribute.
18533 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18536 * config/arm/arm-builtins.c (arm_expand_builtin): Change
18537 expansion to perform a bitwise AND of the argument followed by a
18538 boolean negation of the result.
18540 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
18542 PR rtl-optimization/84878
18543 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
18544 the basic block. Assert the use reference is not artificial and that
18545 it has an associated insn.
18547 2018-04-04 Michael Matz <matz@suse.de>
18549 * builtins.c (compute_objsize): Pass correct operand
18550 to array_at_struct_end_p.
18552 2018-04-04 Richard Biener <rguenther@suse.de>
18555 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
18556 from contexts for DINFO_LEVEL_TERSE and below.
18558 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
18560 * config/nds32/nds32-doubleword.md (move_<mode>): Require
18561 resiter_operand condition.
18562 * config/nds32/nds32.md (*move<mode>): Ditto.
18564 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
18565 Monk Chiang <sh.chiang04@gmail.com>
18567 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
18569 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
18571 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
18573 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
18574 Kito Cheng <kito.cheng@gmail.com>
18576 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
18577 nds32_cond_code_str, output_cond_branch,
18578 output_cond_branch_compare_zero, nds32_expand_cbranch,
18579 nds32_expand_cstore, nds32_expand_movcc,
18580 nds32_output_cbranchsi4_equality_zero,
18581 nds32_output_cbranchsi4_equality_reg,
18582 nds32_output_cbranchsi4_equality_reg_or_const_int,
18583 nds32_output_cbranchsi4_greater_less_zero: New functions.
18584 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
18585 nds32_expand_cstore, nds32_expand_movcc,
18586 nds32_output_cbranchsi4_equality_zero,
18587 nds32_output_cbranchsi4_equality_reg,
18588 nds32_output_cbranchsi4_equality_reg_or_const_int,
18589 nds32_output_cbranchsi4_greater_less_zero): Declare.
18590 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
18591 nds32_rimm11s_operand): New predicates.
18592 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
18593 * config/nds32/nds32.md: Rewrite all the branch and conditional move
18596 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
18598 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
18599 * config/nds32/nds32.md: Ditto.
18600 * config/nds32/pipelines.md: Ditto.
18602 2018-04-04 Richard Biener <rguenther@suse.de>
18604 PR tree-optimization/85168
18605 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
18606 propagating abnormals.
18608 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
18610 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
18612 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
18613 Kito Cheng <kito.cheng@gmail.com>
18615 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
18616 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
18617 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
18618 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
18619 * config/nds32/nds32.md (sibcall_internal): New.
18620 (sibcall_register): Remove.
18621 (sibcall_immediate): Remove.
18622 (sibcall_value_internal): New.
18623 (sibcall_value_register): Remove.
18624 (sibcall_value_immediate): Remove.
18625 * config/nds32/predicates.md (nds32_general_register_operand): New.
18626 (nds32_call_address_operand): New.
18628 2018-04-03 Jakub Jelinek <jakub@redhat.com>
18630 PR rtl-optimization/85167
18631 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
18632 bb_defs if *split_p, instead preinitialize it to NULL.
18634 PR tree-optimization/85156
18635 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
18636 evaluating the argument multiple times.
18638 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
18640 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
18642 (_mm_cvtpd_ps): Likewise.
18643 (_mm_cvttpd_epi32): Likewise.
18644 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
18645 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
18646 vector, pixel, and bool following altivec.h include.
18648 2018-04-03 Martin Sebor <msebor@redhat.com>
18650 * doc/extend.texi (Common Function Attributes): Clarify.
18651 (const attribute): Likewise.
18652 (pure attribute): Likewise.
18654 2018-04-03 Jakub Jelinek <jakub@redhat.com>
18657 * config/i386/i386.c (ix86_expand_vector_set): Use
18658 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
18660 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
18662 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
18663 instructions when changing rounding bits to preserve precision bits
18664 in the x87 control word.
18666 2018-04-03 Martin Liska <mliska@suse.cz>
18668 PR tree-optimization/82491
18669 * rtl.h (strip_offset_and_add): Replace += suboffset with
18670 poly_uint64 () + suboffset.
18672 2018-03-29 Martin Liska <mliska@suse.cz>
18673 Martin Jambor <mjambor@suse.cz>
18676 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
18677 param_type is not an integral or pointer type.
18679 2018-04-03 Richard Biener <rguenther@suse.de>
18681 * sese.h (recompute_all_dominators): Remove.
18683 2018-04-02 Martin Sebor <msebor@redhat.com>
18685 * doc/invoke.texi (-Wrestrict): Fix typos.
18687 2018-04-02 Jim Wilson <jimw@sifive.com>
18689 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
18690 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
18691 (<optab>di3, <optab>si3_extend): Likewise.
18692 (<optab>si3_mask, <optab>si3_mask_1): New.
18693 (<optab>di3_mask, <optab>di3_mask_1): New.
18694 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
18695 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
18696 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
18698 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
18700 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
18703 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
18705 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
18706 (nds32_canonicalize_comparison): New function.
18708 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
18709 Kito Cheng <kito.cheng@gmail.com>
18710 Kuan-Lin Chen <kuanlinchentw@gmail.com>
18712 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
18713 * config/nds32/constants.md (unspec_volatile_element): Add
18714 UNSPEC_VOLATILE_RELAX_GROUP.
18715 * config/nds32/nds32-relax-opt.c: New file.
18716 * config/nds32/nds32-predicates.c
18717 (nds32_symbol_load_store_p): New function.
18718 * config/nds32/nds32-protos.h
18719 (nds32_symbol_load_store_p): Declare function.
18720 (make_pass_nds32_relax_opt): Declare new rtl pass function.
18721 * config/nds32/nds32.c
18722 (nds32_register_pass): New function to register pass.
18723 (nds32_register_passes): New function to register passes.
18724 * config/nds32/nds32.md (relax_group): New pattern.
18725 * config/nds32/nds32.opt (mrelax-hint): New option.
18726 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
18728 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
18730 * config/nds32/t-nds32: Modify files dependency.
18732 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
18734 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
18735 (PROFILE_HOOK): Define its implementation.
18737 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
18739 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
18740 type and 32-bit size.
18742 2018-04-01 Jakub Jelinek <jakub@redhat.com>
18744 PR middle-end/85090
18745 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
18746 (V_128_256): New mode iterator.
18747 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
18748 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
18749 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
18751 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
18752 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
18754 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
18757 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
18758 NaN inputs correctly.
18760 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
18763 * config/rs6000/vsx.md (??r): New mode attribute.
18764 (*vsx_mov<mode>_64bit): Use it.
18765 (*vsx_mov<mode>_32bit): Likewise.
18767 2018-03-30 Martin Sebor <msebor@redhat.com>
18769 PR tree-optimization/84818
18770 * builtins.c (check_access): Use warning_n.
18772 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18775 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
18777 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
18780 2018-03-30 Julia Koval <julia.koval@intel.com>
18783 * x86-tune.def (movx, partial_reg_dependency): Enable for
18786 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
18788 PR inline-asm/84985
18789 * lra-constraints.c (process_alt_operands): Move setting
18790 this_alternative_matches below.
18792 2018-03-29 Martin Liska <mliska@suse.cz>
18795 * doc/invoke.texi: Document how LTO works with debug info.
18796 Describe auto-load support of binutils. Mention 'x86-64'
18797 as valid option value of -march option.
18799 2018-03-29 Jakub Jelinek <jakub@redhat.com>
18801 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
18804 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
18805 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
18806 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
18809 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
18812 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
18813 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
18814 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
18815 for RS6000_BTM_POWERPC64.
18816 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
18817 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
18818 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
18821 (DIVDEU): Likewise.
18823 2018-03-28 Carl Love <cel@us.ibm.com>
18826 2017-09-27 Carl Love <cel@us.ibm.com>
18828 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
18829 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
18830 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
18833 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18835 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
18836 instead of __vector bool.
18837 (_mm_max_pu8): Likewise.
18838 (_mm_min_pi16): Likewise.
18840 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
18843 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
18844 (DIVWEUO): Likewise.
18845 (DIVDEO): Likewise.
18846 (DIVDEUO): Likewise.
18847 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
18848 DIVWEUO and DIVDEUO.
18849 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
18850 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
18851 (div_extend): Likewise.
18852 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
18854 (__builtin_divweuo): Likewise.
18855 (__builtin_divdeo): Likewise.
18856 (__builtin_divdeuo): Likewise.
18858 2018-03-28 Jakub Jelinek <jakub@redhat.com>
18861 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
18862 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
18864 PR tree-optimization/82004
18865 * gimple-match-head.c (optimize_pow_to_exp): New function.
18866 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
18867 Don't fold to exp if optimize_pow_to_exp is false.
18869 2018-03-28 Martin Liska <mliska@suse.cz>
18872 * calls.c (initialize_argument_information): Fix trailing space.
18873 * common.opt: Fix typo and provide better explanation for
18874 -fsanitize-coverage option.
18875 * config/i386/i386.opt: Fix typo.
18877 2018-03-28 Jakub Jelinek <jakub@redhat.com>
18878 Martin Liska <mliska@suse.cz>
18881 * gimplify.c (asan_poison_variable): Don't do the check for
18882 gimplify_omp_ctxp here.
18883 (gimplify_decl_expr): Do it here.
18884 (gimplify_target_expr): Likewise.
18886 2018-03-28 Martin Liska <mliska@suse.cz>
18889 * config/i386/i386.c (ix86_function_arg_advance): Do not call
18890 chkp_type_bounds_count if MPX is not enabled.
18892 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
18894 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
18896 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
18899 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
18900 function to create the function decl for complex long double
18901 multiply and divide for -mabi=ieeelongdouble.
18902 (init_float128_ieee): Call it.
18904 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
18907 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
18908 -fcf-protection=branch -mibt.
18909 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
18911 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18914 * config/arm/arm.c (arm_valid_symbolic_address): Handle
18915 arm_word_relocations.
18917 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
18920 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
18921 extern array declarations.
18923 2018-03-27 Richard Biener <rguenther@suse.de>
18925 PR middle-end/84067
18926 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
18927 explicit single_use checks.
18929 2018-03-27 Richard Biener <rguenther@suse.de>
18931 PR tree-optimization/85082
18932 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
18935 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18937 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
18938 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
18939 Turn on fasynchronous-unwind-tables and funwind-tables.
18941 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
18944 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
18945 (*bmi_blsr_<mode>_ccz): Ditto.
18947 2018-03-26 Tom de Vries <tom@codesourcery.com>
18949 PR tree-optimization/85063
18950 * omp-general.c (offloading_function_p): New function. Factor out
18952 * omp-offload.c (pass_omp_target_link::gate): ... here.
18953 * omp-general.h (offloading_function_p): Declare.
18954 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
18955 with attribute omp declare target for offloading functions.
18957 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
18959 PR tree-optimization/84005
18960 * tree-data-ref.h (get_base_for_alignment): Declare.
18961 * tree-data-ref.c (get_base_for_alignment_1): New function.
18962 (get_base_for_alignment): Likewise.
18963 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
18964 get_base_for_alignment to find a suitable base object, instead
18965 of always using drb->base_address.
18967 2018-03-23 Jakub Jelinek <jakub@redhat.com>
18969 PR inline-asm/85022
18970 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
18971 known size by default.
18973 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
18975 PR inline-asm/85030
18976 * lra-constraints.c (process_alt_operands): Don't match BLKmode
18977 and non BLKmode operands.
18979 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18982 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
18983 Clean up attributes.
18985 2018-03-23 Richard Biener <rguenther@suse.de>
18988 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
18989 we are going to emit early debug for LTO.
18991 2018-03-23 Jakub Jelinek <jakub@redhat.com>
18993 PR inline-asm/85034
18994 * function.c (match_asm_constraints_1): Don't optimize if input
18995 doesn't satisfy general_operand predicate for output's mode.
18997 PR inline-asm/85022
18998 * alias.c (write_dependence_p): Don't require for x_canonicalized
18999 non-VOIDmode if x has VOIDmode.
19002 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
19003 just don't try to optimize it rather than assert it never happens.
19005 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
19007 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
19008 macro expansions for definition of ST_INTERNAL_<mode> and
19009 LD_INTERNAL_<mode> builtins.
19010 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
19012 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
19014 (altivec_expand_st_builtin): Likewise.
19015 (altivec_expand_builtin): Remove calls to deleted functions.
19016 (rs6000_address_for_altivec): Delete this function.
19017 * config/rs6000/vector.md: Remove expands for
19018 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
19020 2018-03-22 Sudakshina Das <sudi.das@arm.com>
19023 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
19024 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
19025 re-computing once computed.
19026 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
19027 (arm_init_machine_status): Initialize
19028 machine->static_chain_stack_bytes.
19030 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
19033 * doc/extend.texi: Add four new prototypes for vec_ld.
19034 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
19035 definitions for more logical presentation.
19036 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
19037 entries for V1TI variants of __builtin_altivec_ld builtin.
19038 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
19039 handling of V1TI variant of LVX icode pattern.
19040 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
19041 (rs6000_gimple_fold_builtin): Likewise.
19042 (altivec_init_builtins): Add code to define
19043 __builtin_altivec_lvx_v1ti function.
19045 2018-03-22 Jakub Jelinek <jakub@redhat.com>
19047 PR inline-asm/84941
19048 * function.c (match_asm_constraints_1): Don't do the optimization
19049 if input isn't a REG, SUBREG, MEM or constant.
19051 2018-03-22 Tom de Vries <tom@codesourcery.com>
19053 PR tree-optimization/84956
19054 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
19055 bb_has_abnormal_pred.
19057 2018-03-22 Jakub Jelinek <jakub@redhat.com>
19060 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
19061 DECL_INITIAL (decl) to decl at the end.
19062 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
19063 adjust the comment.
19065 2018-03-21 Joseph Myers <joseph@codesourcery.com>
19067 * doc/extend.texi (__builtin_tgmath): Document when complex
19068 integer types are treated as _Complex _Float64.
19070 2018-03-21 Tom de Vries <tom@codesourcery.com>
19072 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
19074 2018-03-21 Jakub Jelinek <jakub@redhat.com>
19076 PR tree-optimization/84960
19077 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
19078 if it is ENTRY block, move them into single succ of ENTRY in that case.
19080 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
19082 PR tree-optimization/84811
19083 * poly-int.h (poly_span_traits): Remove the T3 parameter and
19084 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
19085 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
19086 (known_subrange_p): Update accordingly. Cast each value involved
19087 in the size comparison, rather than casting the result of the
19090 2018-03-21 Jakub Jelinek <jakub@redhat.com>
19092 PR tree-optimization/84982
19093 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
19094 by flipping the least significant bit rather than all bits from
19095 bitpos to bitpos + bitsize - 1.
19097 2018-03-21 Nathan Sidwell <nathan@acm.org>
19099 * doc/extend.texi (Deprecated Features): Remove mention of
19100 long-deleted deprecations.
19102 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19105 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
19106 * configure: Regenerate.
19108 2018-03-21 Tom de Vries <tom@codesourcery.com>
19110 PR tree-optimization/83126
19111 * tree-parloops.c (num_phis): New function.
19112 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
19114 2018-03-21 Nathan Sidwell <nathan@acm.org>
19116 * doc/extend.texi (Deprecated Features): Update deprecated flags,
19117 mention anon-struct/union members and trailing attributes.
19119 2018-03-21 Bin Cheng <bin.cheng@arm.com>
19121 PR tree-optimization/84969
19122 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
19123 builtin memset partitions if they set different rhs values.
19125 2018-03-21 Jakub Jelinek <jakub@redhat.com>
19127 PR rtl-optimization/84989
19128 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
19129 VEC_DUPLICATE with scalar result mode.
19131 2018-03-21 Martin Liska <mliska@suse.cz>
19134 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
19135 not intended return statement.
19137 2018-03-21 Martin Liska <mliska@suse.cz>
19140 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
19141 (chkp_find_bound_slots_1): Limit number of iterations.
19143 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
19146 * Minor grammar fixes for x86 options.
19148 2018-03-20 Jakub Jelinek <jakub@redhat.com>
19151 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
19152 holding REG_CFA_RESTORE notes, instead turn them into a USE.
19154 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
19157 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
19158 (altivec_lvx_<mode>_1op): Likewise.
19159 (altivec_stvx_<mode>_2op): Likewise.
19160 (altivec_stvx_<mode>_1op): Likewise.
19161 (altivec_lvx_<VM2:mode>): New define_expand.
19162 (altivec_stvx_<VM2:mode>): Likewise.
19163 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
19164 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
19165 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
19166 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
19167 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
19168 (rs6000_gen_lvx): Likewise.
19169 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
19170 (altivec_expand_stv_builtin): Likewise.
19171 (altivec_expand_builtin): Likewise.
19172 * config/rs6000/vector.md: Likewise.
19174 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19177 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
19180 2018-03-20 Richard Biener <rguenther@suse.de>
19183 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
19184 sign-conversions as zero, fall back to standard scalar_stmt
19187 2018-03-20 Martin Liska <mliska@suse.cz>
19190 * predict.c (rebuild_frequencies): Handle case when we have
19191 PROFILE_ABSENT, but flag_guess_branch_prob is false.
19193 2018-03-20 Jakub Jelinek <jakub@redhat.com>
19196 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
19197 flag_section_anchors.
19198 * varasm.c (use_blocks_for_decl_p): Remove hack for
19199 dw2_force_const_mem.
19202 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
19204 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
19205 be created, use lowpart_subreg of operands[0] rather than operands[0]
19207 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
19208 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
19209 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
19210 and n constraint instead of aarch64_shift_imm_di and Usd.
19211 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
19212 (*aarch64_<optab>_reg_minus<mode>3): ... this.
19214 2018-03-20 Sudakshina Das <sudi.das@arm.com>
19217 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
19218 to favor GPR over NEON registers.
19219 (<shift>di3_neon): Likewise.
19221 2018-03-20 Tom de Vries <tom@codesourcery.com>
19224 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
19225 (nvptx_process_pars): Emit bar.sync asap and alap.
19227 2018-03-20 Tom de Vries <tom@codesourcery.com>
19230 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
19231 seen_label if seen_label is already set.
19233 2018-03-20 Jakub Jelinek <jakub@redhat.com>
19236 * config/i386/i386.c (fold_builtin_cpu): For features above 31
19237 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
19238 Use 1U instead of 1. Formatting fixes.
19241 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
19242 instead of TREE_TYPE (s1) for the return value.
19244 2018-03-19 Jakub Jelinek <jakub@redhat.com>
19246 PR tree-optimization/84946
19247 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
19248 bitsize + bitsize in poly_uint64 rather than poly_int64.
19251 * dwarf2asm.c: Include fold-const.c.
19252 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
19253 of decl rather than decl itself.
19255 PR rtl-optimization/84643
19256 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
19258 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
19261 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
19262 calling assemble_variable.
19264 2018-03-19 Sudakshina Das <sudi.das@arm.com>
19267 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
19268 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
19270 2018-03-19 Jim Wilson <jimw@sifive.com>
19273 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
19274 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
19275 (riscv_first_stack_step): Likewise.
19276 (riscv_option_override): Use STACK_BOUNDARY instead of
19277 MIN_STACK_BOUNDARY.
19278 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
19279 MIN_STACK_BOUNDARY.
19280 (BIGGEST_ALIGNMENT): Set to 128.
19281 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
19282 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
19285 2018-03-19 Richard Biener <rguenther@suse.de>
19287 PR tree-optimization/84933
19288 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
19289 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
19291 2018-03-19 Richard Biener <rguenther@suse.de>
19293 PR tree-optimization/84859
19294 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
19295 (cond_if_else_store_replacement): Perform sinking operation on
19296 single-store BBs regardless of MAX_STORES_TO_SINK setting.
19297 Generalize what a BB with a single eligible store is.
19299 2018-03-19 Richard Biener <rguenther@suse.de>
19301 PR tree-optimization/84929
19302 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
19303 chrec_is_positive against non-chrec arg.
19305 2018-03-19 Tamar Christina <tamar.christina@arm.com>
19308 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
19310 2018-03-18 Martin Liska <mliska@suse.cz>
19312 PR rtl-optimization/84635
19313 * regrename.c (build_def_use): Use matches_mode only when
19316 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
19318 PR tree-optimization/84913
19319 * tree-vect-loop.c (vectorizable_reduction): Don't try to
19320 vectorize chains of COND_EXPRs.
19322 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
19324 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
19326 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
19328 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
19330 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
19332 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
19334 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
19335 Kito Cheng <kito.cheng@gmail.com>
19337 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
19338 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
19339 (nds32_adjust_reg_alloc_order): New function.
19340 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
19342 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
19344 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
19345 nds32_print_operand, nds32_print_operand_address): Use
19346 HOST_WIDE_INT_PRINT_DEC instead.
19348 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
19350 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
19352 2018-03-17 Jakub Jelinek <jakub@redhat.com>
19355 * config/i386/i386.c (initial_ix86_tune_features,
19356 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
19357 unsigned long long.
19358 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
19359 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
19360 rather than 1u << ix86_tune. Formatting fix.
19361 (ix86_option_override_internal): Change ix86_arch_mask from
19362 unsigned int to unsigned HOST_WIDE_INT, initialize to
19363 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
19364 (ix86_function_specific_restore): Likewise.
19366 2018-03-16 Jakub Jelinek <jakub@redhat.com>
19369 * postreload.c (reload_combine_recognize_pattern): Perform
19370 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
19371 truncate_int_for_mode the result for the destination's mode.
19374 * hsa-gen.c (mem_type_for_type): Fix comment typo.
19375 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
19377 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
19380 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
19383 * lra-assigns.c (lra_split_hard_reg_for): Don't use
19384 regno_allocno_class_array and sorted_pseudos.
19385 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
19386 insns where regno is used.
19388 2018-03-16 Martin Liska <mliska@suse.cz>
19391 * multiple_target.c (create_dispatcher_calls): Redirect
19392 reference in the symbol table.
19394 2018-03-16 Martin Liska <mliska@suse.cz>
19397 * multiple_target.c (create_dispatcher_calls): Redirect also
19400 2018-03-16 Jakub Jelinek <jakub@redhat.com>
19404 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
19405 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
19406 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
19408 2018-03-16 Julia Koval <julia.koval@intel.com>
19410 * doc/invoke.texi (Skylake Server): Add CLWB.
19411 Cannonlake): Remove CLWB.
19413 2018-03-16 Jakub Jelinek <jakub@redhat.com>
19415 PR tree-optimization/84841
19416 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
19418 (FLOAT_ONE_CONST_TYPE): Define.
19419 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
19420 (sort_by_operand_rank): Put entries with higher constant_type last
19421 rather than first to match comments.
19423 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
19425 * config/nios2/nios2.md (movsi_internal): Fix thinko in
19428 2018-03-15 Jakub Jelinek <jakub@redhat.com>
19431 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
19432 check and use address of target always.
19434 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
19437 * config/i386/i386.c (indirect_thunk_needed): Update comments.
19438 (indirect_thunk_bnd_needed): Likewise.
19439 (indirect_thunks_used): Likewise.
19440 (indirect_thunks_bnd_used): Likewise.
19441 (indirect_return_needed): New.
19442 (indirect_return_bnd_needed): Likewise.
19443 (output_indirect_thunk_function): Add a bool argument for
19445 (output_indirect_thunk_function): Don't generate alias for
19446 function return thunk.
19447 (ix86_code_end): Call output_indirect_thunk_function to generate
19448 function return thunks.
19449 (ix86_output_function_return): Set indirect_return_bnd_needed
19450 and indirect_return_needed instead of indirect_thunk_bnd_needed
19451 and indirect_thunk_needed.
19453 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
19455 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
19456 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
19457 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
19459 2018-03-15 David Malcolm <dmalcolm@redhat.com>
19460 Paul Hua <paul.hua.gm@gmail.com>
19463 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
19465 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
19467 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
19468 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
19469 resp. SFmode cases.
19471 2018-03-15 Tamar Christina <tamar.christina@arm.com>
19474 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
19475 instead of GET_MODE_SIZE when comparing Units.
19477 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
19480 * varasm.c (hash_section): Return an unchangeble hash value
19481 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
19482 Return !aarch64_can_use_per_function_literal_pools_p ().
19484 2018-03-15 Jakub Jelinek <jakub@redhat.com>
19487 * optabs.c (emit_conditional_move): Pass address of cmode's copy
19488 rather than address of cmode as last argument to prepare_cmp_insn.
19490 2018-03-15 Julia Koval <julia.koval@intel.com>
19492 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
19493 F_AVX512VNNI, F_AVX512BITALG): New.
19495 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
19498 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
19499 insn for floating-point loads and stores.
19501 2018-03-14 Carl Love <cel@us.ibm.com>
19503 * config/rs6000/rs6000-c.c: Add macro definitions for
19504 ALTIVEC_BUILTIN_VEC_PERMXOR.
19505 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
19506 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
19507 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
19508 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
19510 * config/doc/extend.texi: Add prototypes for vec_permxor.
19512 2018-03-14 David Malcolm <dmalcolm@redhat.com>
19515 * diagnostic-show-locus.c (class layout_point): Convert m_line
19516 from int to linenum_type.
19517 (line_span::comparator): Use linenum "compare" function when
19518 comparing line numbers.
19519 (test_line_span): New function.
19520 (layout_range::contains_point): Convert param "row" from int to
19522 (layout_range::intersects_line_p): Likewise.
19523 (layout::will_show_line_p): Likewise.
19524 (layout::print_source_line): Likewise.
19525 (layout::should_print_annotation_line_p): Likewise.
19526 (layout::print_annotation_line): Likewise.
19527 (layout::print_leading_fixits): Likewise.
19528 (layout::annotation_line_showed_range_p): Likewise.
19529 (struct line_corrections): Likewise for field m_row.
19530 (line_corrections::line_corrections): Likewise for param "row".
19531 (layout::print_trailing_fixits): Likewise.
19532 (layout::get_state_at_point): Likewise.
19533 (layout::get_x_bound_for_row): Likewise.
19534 (layout::print_line): Likewise.
19535 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
19536 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
19537 * input.c (selftest::test_linenum_comparisons): New function.
19538 (selftest::input_c_tests): Call it.
19539 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
19540 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
19541 * selftest.h (ASSERT_GT): New macro.
19542 (ASSERT_GT_AT): New macro.
19543 (ASSERT_LT): New macro.
19544 (ASSERT_LT_AT): New macro.
19546 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
19548 PR rtl-optimization/84780
19549 * combine.c (distribute_links): Don't make a link based on pc_rtx.
19551 2018-03-14 Martin Liska <mliska@suse.cz>
19553 * tree.c (record_node_allocation_statistics): Use
19554 get_stats_node_kind.
19555 (get_stats_node_kind): New function extracted from
19556 record_node_allocation_statistics.
19557 (free_node): Use get_stats_node_kind.
19559 2018-03-14 Richard Biener <rguenther@suse.de>
19561 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
19562 that the value-set of ANTIC_IN doesn't grow.
19565 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
19567 (BB_VISITED_WITH_VISITED_SUCCS): New define.
19568 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
19570 2018-03-14 Julia Koval <julia.koval@intel.com>
19572 * config.gcc (icelake-client, icelake-server): New.
19574 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
19575 (initial_ix86_arch_features): Ditto.
19576 (PTA_SKYLAKE): Add SGX.
19577 (PTA_ICELAKE): Remove.
19578 (PTA_ICELAKE_CLIENT): New.
19579 (PTA_ICELAKE_SERVER): New.
19580 (ix86_option_override_internal): Split up icelake on icelake client and
19582 (get_builtin_code_for_version): Ditto.
19583 (fold_builtin_cpu): Ditto.
19584 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
19585 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
19586 * config/i386/i386.h (processor_type): Ditto.
19587 * doc/invoke.texi: Ditto.
19589 2018-03-14 Jakub Jelinek <jakub@redhat.com>
19592 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
19593 INTEGER_CST offset, add it together with bitpos / 8 and
19594 sign extend based on POINTER_SIZE.
19598 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
19601 * config/i386/constraints.md (Yc): New register constraint.
19602 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
19603 Use Yc constraint for alternative 2 of operand 0. Remove
19604 preferred_for_speed attribute.
19606 2018-03-14 Richard Biener <rguenther@suse.de>
19608 PR tree-optimization/84830
19609 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
19610 with the old one to avoid oscillations.
19612 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
19615 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
19617 (assign_by_spills): Return a flag of reload assignment failure.
19618 Do not process the reload assignment failures. Do not spill other
19619 reload pseudos if they has the same reg class. Update n if
19621 (lra_assign): Add a return arg. Set up from the result of
19622 assign_by_spills call.
19623 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
19624 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
19625 usage_insns if it is not NULL.
19626 (spill_hard_reg_in_range): New function.
19627 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
19628 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
19629 function prototypes.
19630 (lra_assign): Change prototype.
19631 * lra.c (lra): Add code to deal with fails by splitting hard reg
19634 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
19636 * config/riscv/riscv.opt (mrelax): New option.
19637 * config/riscv/riscv.c (riscv_file_start): Emit ".option
19638 "norelax" when riscv_mrelax is disabled.
19639 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
19641 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19644 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
19645 reassociation for int modes.
19647 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
19649 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
19650 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
19652 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
19653 * config/aarch64/aarch64-sve.md
19654 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
19655 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
19656 (*extend<mode><Vwide>2): Rename to...
19657 (aarch64_sve_extend<mode><Vwide>2): ...this.
19658 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
19659 renaming the old pattern to...
19660 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
19662 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
19663 define_expand, renaming the old pattern to...
19664 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
19665 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
19666 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
19667 account when deciding which SVE instruction the optab should use.
19668 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
19670 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
19672 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
19673 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
19674 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
19675 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
19676 (tlsdesc_small_<mode>): Turn a define_expand and use
19677 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
19678 (tlsdesc_small_advsimd_<mode>): ...this.
19679 (tlsdesc_small_sve_<mode>): New pattern.
19681 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
19683 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
19684 (UNSPEC_UMUL_HIGHPART): New constants.
19685 (MUL_HIGHPART): New int iteraor.
19686 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
19687 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
19689 (*<su>mul<mode>3_highpart): New define_insn.
19691 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
19694 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
19697 2018-03-13 Martin Liska <mliska@suse.cz>
19700 * (sem_item_optimizer::sem_item_optimizer): Initialize new
19702 (sem_item_optimizer::~sem_item_optimizer): Release it.
19703 (sem_item_optimizer::merge_classes): Register variable aliases.
19704 (sem_item_optimizer::fixup_pt_set): New function.
19705 (sem_item_optimizer::fixup_points_to_sets): Likewise.
19706 * ipa-icf.h: Declare new variables and functions.
19708 2018-03-13 Jakub Jelinek <jakub@redhat.com>
19710 PR middle-end/84834
19711 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
19712 integer_pow2p@2 and test integer_pow2p in condition.
19713 (A < 0 ? C : 0): Similarly for @1.
19715 PR middle-end/84831
19716 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
19717 characters starting at p contain '\0' character, don't look beyond
19721 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
19722 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
19725 * reg-stack.c (change_stack): Change update_end var from int to
19726 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
19727 also call set_block_for_insn on the newly added insns and rescan.
19730 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
19731 on the last operand.
19734 * tree.c (stabilize_reference_1): Return save_expr (e) for
19735 STATEMENT_LIST even if it doesn't have side-effects.
19737 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
19739 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
19741 2018-03-12 Renlin Li <renlin.li@arm.com>
19743 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
19744 aarch64_output_scalar_simd_mov_immediate.
19746 2018-03-12 Martin Sebor <msebor@redhat.com>
19748 PR tree-optimization/83456
19749 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
19750 for perfectly overlapping calls to memcpy.
19751 (gimple_fold_builtin_memory_chk): Same.
19752 (gimple_fold_builtin_strcpy): Handle no-warning.
19753 (gimple_fold_builtin_stxcpy_chk): Same.
19754 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
19756 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
19758 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
19759 parameter. Use it for SFmode.
19760 (rs6000_function_arg_advance_1): Adjust.
19761 (rs6000_function_arg): Adjust.
19762 (rs6000_gimplify_va_arg): Pass false for that new parameter.
19764 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
19766 PR rtl-optimization/84169
19767 PR rtl-optimization/84780
19768 * combine.c (can_combine_p): Check for a 2-insn combination whether
19769 the destination register is used between the two insns, too.
19771 2018-03-12 Richard Biener <rguenther@suse.de>
19773 PR tree-optimization/84803
19774 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
19775 for refs DR analysis didn't process.
19777 2018-03-12 Richard Biener <rguenther@suse.de>
19779 PR tree-optimization/84777
19780 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
19781 force-vectorize loops ignore whether we are optimizing for size.
19783 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
19785 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
19786 (TARGET_MD_ASM_ADJUST): Define.
19788 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
19789 Kito Cheng <kito.cheng@gmail.com>
19790 Chung-Ju Wu <jasonwucj@gmail.com>
19792 * config/nds32/nds32.c (nds32_compute_stack_frame,
19793 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
19794 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
19795 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
19796 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
19797 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
19798 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
19799 * config/nds32/nds32.md (prologue, epilogue): Use macro
19800 NDS32_V3PUSH_AVAILABLE_P to do checking.
19802 2018-03-11 Jakub Jelinek <jakub@redhat.com>
19805 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
19806 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
19807 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
19808 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
19809 addition of most attributes on !orig_type_die or the attribute not
19810 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
19812 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
19813 Chung-Ju Wu <jasonwucj@gmail.com>
19815 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
19816 __NDS32_VH__ macro.
19817 * config/nds32/nds32.opt (mvh): New option.
19819 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
19820 Chung-Ju Wu <jasonwucj@gmail.com>
19822 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
19824 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
19825 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
19828 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
19829 Chung-Ju Wu <jasonwucj@gmail.com>
19831 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
19833 * config/nds32/nds32-multiple.md (strlensi): New pattern.
19834 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
19836 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
19837 Kito Cheng <kito.cheng@gmail.com>
19838 Chung-Ju Wu <jasonwucj@gmail.com>
19840 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
19841 UNSPEC_FFMISM and UNSPEC_FLMISM.
19842 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
19843 for ffb, ffmism and flmism.
19844 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
19845 (unspec_ffmism): Ditto.
19846 (unspec_flmism): Ditto.
19847 (nds32_expand_builtin_impl): Check if string extension is available.
19848 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
19849 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
19851 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
19854 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
19857 * lra-assigns.c (assign_by_spills): Return a flag of reload
19858 assignment failure. Do not process the reload assignment
19859 failures. Do not spill other reload pseudos if they has the same
19861 (lra_assign): Add a return arg. Set up from the result of
19862 assign_by_spills call.
19863 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
19864 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
19865 usage_insns if it is not NULL.
19866 (spill_hard_reg_in_range): New function.
19867 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
19868 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
19869 function prototypes.
19870 (lra_assign): Change prototype.
19871 * lra.c (lra): Add code to deal with fails by splitting hard reg
19874 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
19877 * config/i386/i386.opt: Replace Enforcment with Enforcement.
19879 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
19882 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
19883 (dw_val_node): Add val_symbolic_view.
19884 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
19885 (symview_upper_bound): New.
19886 (new_line_info_table): Initialize symviews_since_reset.
19887 (dwarf2out_source_line): Count symviews_since_reset and set
19888 symview_upper_bound.
19889 (dw_val_equal_p): Handle symview.
19890 (add_AT_symview): New.
19891 (print_dw_val): Handle symview.
19892 (attr_checksum, attr_checksum_ordered): Likewise.
19893 (same_dw_val_p, size_of_die): Likewise.
19894 (value_format, output_die): Likewise.
19895 (add_high_low_attributes): Use add_AT_symview for entry_view.
19896 (dwarf2out_finish): Reset symview_upper_bound, clear
19899 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
19902 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
19903 Add strict argument and use it.
19904 (rs6000_split_multireg_move): Update for new strict argument.
19905 (mem_operand_gpr): Disallow all non-offsettable addresses.
19906 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
19908 2018-03-09 Jakub Jelinek <jakub@redhat.com>
19911 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
19912 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
19913 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
19916 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
19917 decl, use remap_type if we want to use the type.
19919 2018-03-09 Martin Sebor <msebor@redhat.com>
19921 PR tree-optimization/84526
19922 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
19924 (builtin_access::generic_overlap): Be prepared to handle non-array
19927 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
19929 PR rtl-optimization/84682
19930 * lra-constraints.c (process_address_1): Check is_address flag
19931 for address constraints.
19932 (process_alt_operands): Likewise.
19933 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
19934 preprocess_constraints.
19935 * recog.h (preprocess_constraints): Add oploc parameter.
19937 * recog.c (preprocess_constraints): Test address_operand for
19938 CT_ADDRESS constraints.
19940 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
19943 * lra-assigns.c (assign_by_spills): Return a flag of reload
19944 assignment failure. Do not process the reload assignment
19945 failures. Do not spill other reload pseudos if they has the same
19947 (lra_assign): Add a return arg. Set up from the result of
19948 assign_by_spills call.
19949 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
19950 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
19951 usage_insns if it is not NULL.
19952 (spill_hard_reg_in_range): New function.
19953 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
19954 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
19955 function prototypes.
19956 (lra_assign): Change prototype.
19957 * lra.c (lra): Add code to deal with fails by splitting hard reg
19960 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19963 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
19964 Accept complain bool parameter. Only emit errors if it is true.
19965 (arm_parse_cpu_option_name): Likewise.
19966 (arm_target_thumb_only): Adjust callers of the above.
19967 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
19968 prototype to take a default true bool parameter.
19969 (arm_parse_arch_option_name): Likewise.
19971 2018-03-09 David Malcolm <dmalcolm@redhat.com>
19972 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
19976 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
19977 * configure: Regenerate.
19978 * configure.ac ("linker --version-script option"): New.
19979 ("linker soname option"): New.
19981 2018-03-09 Richard Biener <rguenther@suse.de>
19983 PR tree-optimization/84775
19984 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
19985 immediate uses of predicate stmts and mark them modified.
19988 PR tree-optimization/84178
19989 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
19991 (version_loop_for_if_conversion): Delay update_ssa call.
19992 (tree_if_conversion): Delay update_ssa until after predicate
19995 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
19998 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
19999 when the function accesses prior frames.
20001 2018-03-08 Jakub Jelinek <jakub@redhat.com>
20004 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
20005 gen_llsym, otherwise call maybe_gen_llsym.
20007 PR inline-asm/84742
20008 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
20009 has ',' character inside of it.
20011 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20014 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
20015 as clobbering CC_REGNUM.
20017 2018-03-08 Richard Biener <rguenther@suse.de>
20019 PR middle-end/84552
20020 * tree-scalar-evolution.c: Include tree-into-ssa.h.
20021 (follow_copies_to_constant): Do not follow SSA names registered
20024 2018-03-08 Richard Biener <rguenther@suse.de>
20026 PR tree-optimization/84178
20027 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
20029 (version_loop_for_if_conversion): Delay update_ssa call.
20030 (tree_if_conversion): Delay update_ssa until after predicate
20033 2018-03-08 David Malcolm <dmalcolm@redhat.com>
20035 PR tree-optimization/84178
20036 * tree-if-conv.c (release_bb_predicate): Remove the
20037 the assertion that the stmts have NULL use_ops.
20038 Discard the statements, asserting that they haven't
20039 yet been added to a BB.
20041 2018-03-08 Richard Biener <rguenther@suse.de>
20043 PR tree-optimization/84746
20044 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
20045 (phi_translate): Pass in destination ANTIC_OUT set.
20046 (phi_translate_1): Likewise. For a simplified result lookup
20047 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
20048 (phi_translate_set): Adjust.
20049 (do_pre_regular_insertion): Likewise.
20050 (do_pre_partial_partial_insertion): Likewise.
20052 2018-03-08 Martin Liska <mliska@suse.cz>
20054 PR gcov-profile/84735
20055 * doc/gcov.texi: Document usage of profile files.
20056 * gcov-io.h: Document changes in the format.
20058 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
20062 * dwarf2out.c (struct dw_line_info_table): Update comments for
20064 (FORCE_RESET_NEXT_VIEW): New.
20065 (FORCE_RESETTING_VIEW_P): New.
20066 (RESETTING_VIEW_P): Check for -1 too.
20067 (ZERO_VIEW_P): Likewise.
20068 (new_line_info_table): Force-reset next view.
20069 (dwarf2out_begin_function): Likewise.
20070 (dwarf2out_source_line): Simplify zero_view_p initialization.
20071 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
20072 view directly. Omit view when omitting .loc at line 0.
20074 2018-03-08 Jakub Jelinek <jakub@redhat.com>
20076 PR tree-optimization/84740
20077 * tree-switch-conversion.c (process_switch): Call build_constructors
20078 only if info.phi_count is non-zero.
20080 PR tree-optimization/84739
20081 * tree-tailcall.c (find_tail_calls): Check call arguments against
20082 DECL_ARGUMENTS (current_function_decl) rather than
20083 DECL_ARGUMENTS (func) when checking for tail recursion.
20085 2018-03-07 Jakub Jelinek <jakub@redhat.com>
20087 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
20088 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
20089 Volker Reichelt's entry and add entries for people that perform
20090 GCC fuzzy testing and report numerous bugs.
20092 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
20095 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
20096 readonly data in sdata, if that is disabled.
20097 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
20098 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
20099 -mreadonly-in-sdata option.
20101 2018-03-07 Martin Sebor <msebor@redhat.com>
20103 PR tree-optimization/84468
20104 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
20105 basic block when looking for nul assignment.
20107 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
20110 * except.h (output_function_exception_table): Adjust prototype.
20111 * except.c (output_function_exception_table): Remove FNNAME parameter
20112 and add SECTION parameter. Ouput one part of the table at a time.
20113 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
20114 the first part of the exception table and emit unwind directives.
20115 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
20116 (i386_pe_seh_cold_init): Likewise.
20117 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
20118 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
20119 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
20120 (ix86_output_call_insn): Emit a nop in one more case for SEH.
20121 * config/i386/winnt.c: Include except.h.
20122 (struct seh_frame_state): Add reg_offset, after_prologue and
20123 in_cold_section fields.
20124 (i386_pe_seh_end_prologue): Set seh->after_prologue.
20125 (i386_pe_seh_cold_init): New function.
20126 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
20127 to seh->in_cold_section.
20128 (seh_emit_push): Record the offset of the push.
20129 (seh_emit_save): Record the offet of the save.
20130 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
20131 Test seh->after_prologue to disregard the epilogue.
20132 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
20133 (i386_pe_end_cold_function): New function.
20135 2018-03-07 Jakub Jelinek <jakub@redhat.com>
20138 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
20139 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
20142 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
20144 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
20145 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
20147 PR middle-end/84723
20148 * multiple_target.c: Include tree-inline.h and intl.h.
20149 (expand_target_clones): Diagnose and fail if node->definition and
20150 !tree_versionable_function_p (node->decl).
20152 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
20154 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
20156 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
20157 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20158 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
20160 2018-03-06 Jakub Jelinek <jakub@redhat.com>
20163 * combine.c (try_combine): Use reg_or_subregno instead of handling
20164 just paradoxical SUBREGs and REGs.
20166 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
20168 * config/arc/arc.c (arc_finalize_pic): Remove function.
20169 (arc_must_save_register): We use single base PIC register, remove
20170 checks to save/restore the PIC register.
20171 (arc_expand_prologue): Likewise.
20172 * config/arc/arc-protos.h (arc_set_default_type_attributes):
20174 (arc_verify_short): Likewise.
20175 (arc_attr_type): Likewise.
20176 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
20177 (walk_stores): Likewise.
20178 (arc_address_cost): Make it static.
20179 (arc_verify_short): Likewise.
20180 (branch_dest): Likewise.
20181 (arc_attr_type): Likewise.
20182 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
20183 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
20184 (arc_final_prescan_insn): Remove inserting the nops due to
20185 hardware hazards. It is done in reorg step.
20186 (insn_length_variant_t): Remove.
20187 (insn_length_parameters_t): Likewise.
20188 (arc_insn_length_parameters): Likewise.
20189 (arc_get_insn_variants): Likewise.
20190 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
20192 2018-03-06 Jakub Jelinek <jakub@redhat.com>
20194 PR inline-asm/84683
20195 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
20198 PR tree-optimization/84687
20199 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
20201 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
20203 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20205 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
20206 Rename to ppc_speculation_barrier.
20207 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
20208 __builtin_ppc_speculation_barrier.
20210 2018-03-05 Jakub Jelinek <jakub@redhat.com>
20213 * combine.c (combine_simplify_rtx): Don't try to simplify if
20214 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
20217 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
20219 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
20220 to 32 bytes when compiling for POWER9.
20222 2018-03-05 Jakub Jelinek <jakub@redhat.com>
20225 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
20226 regparm >= 3 with no arg reg available also for calls with
20227 flag_force_indirect_call. Pass decl to ix86_function_regparm.
20230 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
20232 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
20234 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
20237 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
20239 2018-03-05 Richard Biener <rguenther@suse.de>
20241 PR tree-optimization/84486
20242 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
20243 When inserting a __builtin_assume_aligned call set the LHS
20244 SSA name alignment info accordingly.
20246 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
20248 PR tree-optimization/84114
20249 * config/aarch64/aarch64.c (aarch64_reassociation_width)
20250 Avoid reassociation of FLOAT_MODE addition.
20252 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
20254 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
20255 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
20256 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
20257 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
20258 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
20259 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
20260 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
20262 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
20263 __builtin_ia32_wbinvd): New builtins.
20264 (SPECIAL_ARGS2): New.
20265 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
20266 (SPECIAL_ARGS2): New.
20267 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
20268 (ix86_valid_target_attribute_inner_p): Ditto.
20269 (ix86_init_mmx_sse_builtins): Add special_args2.
20270 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
20271 TARGET_WBNOINVD_P): New.
20272 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
20273 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
20274 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
20275 * config/i386/immintrin.h (_wbinvd): New intrinsic.
20276 * config/i386/pconfigintrin.h: New file.
20277 * config/i386/wbnoinvdintrin.h: Ditto.
20278 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
20280 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
20282 2018-03-05 Richard Biener <rguenther@suse.de>
20284 PR tree-optimization/84670
20285 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
20287 (BB_VISITED_WITH_VISITED_SUCCS): New define.
20288 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
20289 (compute_antic_aux): Only assert the number of values in ANTIC_IN
20290 doesn't grow if all successors (recursively) were visited at least
20293 2018-03-05 Richard Biener <rguenther@suse.de>
20295 PR tree-optimization/84650
20296 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
20297 if executed in the loop pipeline.
20299 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
20301 * doc/configfiles.texi (Configuration Files): Move info about
20302 conditionalizing $target-protos.h to...
20303 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
20304 differs from $target-protos.h.
20306 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
20307 Chung-Ju Wu <jasonwucj@gmail.com>
20309 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
20310 * config/nds32/nds32-multiple.md (setmemsi): Define.
20311 * config/nds32/nds32-memory-manipulation.c
20312 (nds32_gen_dup_4_byte_to_word_value): New.
20313 (emit_setmem_word_loop): New.
20314 (emit_setmem_byte_loop): New.
20315 (nds32_expand_setmem_loop): New.
20316 (nds32_expand_setmem_loop_v3m): New.
20317 (nds32_expand_setmem_unroll): New.
20318 (nds32_expand_setmem): New.
20320 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
20321 Chung-Ju Wu <jasonwucj@gmail.com>
20323 * config/nds32/nds32-memory-manipulation.c
20324 (nds32_emit_load_store): New.
20325 (nds32_emit_post_inc_load_store): New.
20326 (nds32_emit_mem_move): New.
20327 (nds32_emit_mem_move_block): New.
20328 (nds32_expand_movmemsi_loop_unknown_size): New.
20329 (nds32_expand_movmemsi_loop_known_size): New.
20330 (nds32_expand_movmemsi_loop): New.
20331 (nds32_expand_movmemsi_unroll): New.
20332 (nds32_expand_movmemqi): Rename ...
20333 (nds32_expand_movmemsi): ... to this.
20334 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
20335 (movmemsi): ... to this.
20336 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
20337 (nds32_expand_movmemsi): ... to this.
20339 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
20340 Monk Chiang <sh.chiang04@gmail.com>
20341 Chung-Ju Wu <jasonwucj@gmail.com>
20343 * config/nds32/nds32-protos.h
20344 (nds32_expand_load_multiple): New arguments.
20345 (nds32_expand_store_multiple): Ditto.
20346 (nds32_valid_multiple_load_store): Rename ...
20347 (nds32_valid_multiple_load_store_p): ... to this.
20348 * config/nds32/nds32-memory-manipulation.c
20349 (nds32_expand_load_multiple): Refine implementation.
20350 (nds32_expand_store_multiple): Ditto.
20351 * config/nds32/nds32-multiple.md
20352 (load_multiple): Update nds32_expand_load_multiple interface.
20353 (store_multiple): Update nds32_expand_store_multiple interface.
20354 * config/nds32/nds32-predicates.c
20355 (nds32_valid_multiple_load_store): Rename ...
20356 (nds32_valid_multiple_load_store_p): ... to this and refine
20358 * config/nds32/predicates.md
20359 (nds32_load_multiple_and_update_address_operation): New predicate.
20360 (nds32_store_multiple_and_update_address_operation): New predicate.
20362 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
20363 Chung-Ju Wu <jasonwucj@gmail.com>
20365 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
20366 (combo): New attribute.
20367 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
20369 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
20371 * config/nds32/nds32.opt: Change -mcmodel= default value.
20373 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
20374 Monk Chiang <sh.chiang04@gmail.com>
20375 Chung-Ju Wu <jasonwucj@gmail.com>
20377 * config/nds32/constants.md (unspec_element): New enum.
20378 * config/nds32/constraints.md (Umw): New constraint.
20379 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
20380 * config/nds32/nds32-intrinsic.md: Likewise.
20381 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
20382 (nds32_valid_smw_lwm_base_p): New.
20383 (nds32_output_smw_single_word): New.
20384 (nds32_output_lmw_single_word): New.
20385 (nds32_expand_unaligned_load): New.
20386 (nds32_expand_unaligned_store): New.
20387 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
20388 (nds32_output_smw_single_word): Declare.
20389 (nds32_output_lmw_single_word): Declare.
20390 (nds32_expand_unaligned_load): Declare.
20391 (nds32_expand_unaligned_store): Declare.
20392 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
20393 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
20394 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
20395 NDS32_BUILTIN_UASTORE_DW.
20396 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
20399 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
20400 Kito Cheng <kito.cheng@gmail.com>
20401 Chung-Ju Wu <jasonwucj@gmail.com>
20403 * config/nds32/nds32-intrinsic.c
20404 (nds32_expand_builtin_null_ftype_reg): Delete.
20405 (nds32_expand_builtin_reg_ftype_imm): Ditto.
20406 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
20407 (nds32_read_argument): New.
20408 (nds32_legitimize_target): Ditto.
20409 (nds32_legitimize_argument): Ditto.
20410 (nds32_check_constant_argument): Ditto.
20411 (nds32_expand_unop_builtin): Ditto.
20412 (nds32_expand_unopimm_builtin): Ditto.
20413 (nds32_expand_binop_builtin): Ditto.
20414 (nds32_builtin_decl_impl): Ditto.
20415 (builtin_description): Ditto.
20416 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
20417 (nds32_init_builtins_impl): Ditto.
20418 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
20419 (nds32_builtin_decl): New.
20420 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
20421 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
20423 2018-03-02 Jeff Law <law@redhat.com>
20425 * reorg.c (stop_search_p): Handle DEBUG_INSN.
20426 (redundant_insn, fill_simple_delay_slots): Likewise.
20427 (fill_slots_from_thread): Likewise.
20428 * resource.c (mark_referenced_resources): Likewise.
20429 (mark_set_resources, find_dead_or_set_registers): Likewise.
20431 2018-03-02 Jakub Jelinek <jakub@redhat.com>
20433 * substring-locations.h (format_warning_va): Formatting fix for
20434 ATTRIBUTE_GCC_DIAG.
20435 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
20437 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
20438 * substring-locations.c: Include intl.h.
20439 (format_warning_va): Turned into small wrapper around
20440 format_warning_n_va, renamed to ...
20441 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
20442 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
20444 (format_warning_at_substring_n): New function.
20445 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
20446 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
20447 format_warning_at_substring with just a shorter name instead of
20448 const function pointer.
20449 (fmtwarn_n): New function.
20450 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
20451 appropriate, get rid of all the fmtstr temporaries, move conditionals
20452 with G_() wrapped string literals directly into fmtwarn arguments,
20453 cast dir.len to (int), formatting fixes.
20455 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
20457 * doc/invoke.texi: Remove "Cilk Plus" references.
20459 2018-03-02 Jakub Jelinek <jakub@redhat.com>
20460 Richard Biener <rguenther@suse.de>
20463 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
20464 for error or warning attributes if CALL_FROM_THUNK_P is set.
20467 2018-03-02 Jakub Jelinek <jakub@redhat.com>
20470 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
20471 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
20474 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
20475 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
20477 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
20478 instead of -1U in last predictors element's probability member.
20480 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
20483 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
20484 arguments if they are comparable.
20486 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
20488 PR tree-optimization/84634
20489 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
20490 masks and masked_loop_p with a single loop_masks, making sure it's
20491 null for bb vectorization.
20493 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
20495 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
20496 (vect_analyze_data_ref_access): Use loop->safe_len rather than
20497 loop->force_vectorize to check whether there is no alias.
20499 2018-03-02 Jakub Jelinek <jakub@redhat.com>
20502 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
20504 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
20506 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
20507 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
20508 instead of a loop around prev_real_insn.
20509 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
20512 PR inline-asm/84625
20513 * config/i386/i386.c (ix86_print_operand): Use conditional
20514 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
20517 2018-03-02 Richard Biener <rguenther@suse.de>
20519 PR tree-optimization/84427
20520 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
20521 (bitmap_set_subtract_values): Rewrite to handle multiple
20524 (prune_clobbered_mems): Likewise.
20525 (phi_translate): Take edge instead of pred/phiblock.
20526 (phi_translate_1): Likewise.
20527 (phi_translate_set): Likewise. Insert all translated
20528 exprs for a value into the set, keeping possibly multiple
20529 expressions per value.
20530 (compute_antic_aux): Adjust for phi_translate changes.
20531 When intersecting union the expressions and prune those
20532 not in the final value set, keeping possibly multiple
20533 expressions per value. Do not use value-insertion
20534 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
20535 all expressions. Add verification that the value-sets
20536 only shrink during iteration.
20537 (compute_partial_antic_aux): Adjust for the phi_translate changes.
20538 (do_pre_regular_insertion): Likewise.
20539 (do_pre_partial_partial_insertion): Likewise.
20541 2018-03-02 Richard Biener <rguenther@suse.de>
20544 * config/darwin.c (saved_debug_info_level): New static global.
20545 (darwin_asm_lto_start): Disable debug info generation for LTO out.
20546 (darwin_asm_lto_end): Restore debug info generation settings.
20548 2018-03-01 Martin Liska <mliska@suse.cz>
20551 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
20552 volatile arguments.
20554 2018-03-01 Richard Biener <rguenther@suse.de>
20557 * dwarf2out.c (gen_variable_die): Properly handle late VLA
20558 type annotation with LTO when debug was disabled at compile-time.
20560 2018-03-01 Matthew Fortune <mfortune@gmail.com>
20562 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
20564 (mips_final_postscan_insn): Likewise.
20566 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
20568 PR rtl-optimization/84528
20569 * alias.c (init_alias_target): Add commentary.
20570 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
20571 a unique base value if the frame pointer is not eliminated
20572 to the stack pointer.
20574 2018-03-01 Tom de Vries <tom@codesourcery.com>
20576 PR rtl-optimization/83327
20577 * lra-int.h (hard_regs_spilled_into): Declare.
20578 * lra.c (hard_regs_spilled_into): Define.
20579 (init_reg_info): Init hard_regs_spilled_into.
20580 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
20581 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
20582 (process_bb_lives): Handle hard_regs_spilled_into.
20583 (lra_create_live_ranges_1): Before doing liveness propagation, clear
20584 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
20586 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
20588 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
20589 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
20590 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
20591 * config/rs6000/aix72.h: New file.
20593 2018-02-28 Jakub Jelinek <jakub@redhat.com>
20595 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
20596 instead of warning_at with conditional singular and plural messages
20600 * stor-layout.c (update_alignment_for_field): For
20601 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
20602 && !DECL_PACKED (field), do the alignment update, just use
20603 only desired_align instead of MAX (type_align, desired_align)
20605 (place_field): Don't do known_align < desired_align handling
20606 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
20607 is non-NULL, instead do it after rli->prev_field handling and
20608 only if not within a bitfield word. For DECL_PACKED (field)
20609 use type_align of BITS_PER_UNIT.
20611 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
20613 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
20614 superfluous parentheses and trailing spaces.
20616 2018-02-28 Richard Biener <rguenther@suse.de>
20618 PR tree-optimization/84584
20619 * graphite-scop-detection.c (scop_detection::add_scop): Discard
20620 SCoPs with fake exit edge.
20622 2018-02-28 Martin Liska <mliska@suse.cz>
20625 * timevar.c (timer::print): Fix format to properly print 100%
20628 2018-02-28 Richard Biener <rguenther@suse.de>
20630 PR middle-end/84607
20631 * genmatch.c (capture_info::walk_match): Do not mark
20632 captured expressions without operands as expr_p given
20633 they act more like predicates and should be subject to
20634 "lost tail" side-effect preserving.
20636 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
20638 PR rtl-optimization/81611
20639 * auto-inc-dec.c (attempt_change): Move dead note from
20640 mem_insn if it's the next use of regno
20641 (find_address): Take address use of reg holding
20642 non-incremented value. Add parm to limit search to the named
20644 (merge_in_block): Attempt to use a mem insn that is the next
20645 use of the original regno.
20647 2018-02-27 Martin Sebor <msebor@redhat.com>
20650 * doc/invoke.texi (-Wmissing-attributes): New option.
20651 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
20653 2018-02-27 Martin Sebor <msebor@redhat.com>
20655 PR translation/84207
20656 * diagnostic-core.h (warning_n, error_n, inform_n): Change
20657 n argument to unsigned HOST_WIDE_INT.
20658 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
20659 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
20660 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
20661 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
20663 2018-02-27 Richard Biener <rguenther@suse.de>
20665 PR tree-optimization/84512
20666 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
20667 Do not use the estimate returned from record_stmt_cost for
20668 the scalar iteration cost but sum properly using add_stmt_cost.
20670 2018-02-27 Richard Biener <rguenther@suse.de>
20672 PR tree-optimization/84466
20673 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
20674 Adjust last change to less strictly validate use operands.
20676 2018-02-27 Martin Liska <mliska@suse.cz>
20678 PR gcov-profile/84548
20679 * gcov.c (process_file): Allow partial overlap and consider it
20680 also as group functions.
20681 (output_lines): Properly calculate range of lines for a group.
20683 2018-02-27 Martin Liska <mliska@suse.cz>
20685 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
20686 'ggc' suffixes. Change first column width.
20687 (timer::print): Fix formatting of the column.
20689 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
20691 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
20692 preserve inline entry blocks for the sake of debug inline
20693 entry point markers alone.
20694 (remove_unused_locals): Suggest in comments a better place to
20695 force the preservation of inline entry blocks that are
20696 otherwise unused, but do not preserve them.
20698 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
20700 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
20702 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
20705 * config/i386/constraints.md (Bs): Replace
20706 ix86_indirect_branch_register with
20707 TARGET_INDIRECT_BRANCH_REGISTER.
20709 * config/i386/i386.md (indirect_jump): Likewise.
20710 (tablejump): Likewise.
20711 (*sibcall_memory): Likewise.
20712 (*sibcall_value_memory): Likewise.
20713 Peepholes of indirect call and jump via memory: Likewise.
20714 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
20715 (*sibcall_value_GOT_32): Likewise.
20716 * config/i386/predicates.md (indirect_branch_operand): Likewise.
20717 (GOT_memory_operand): Likewise.
20718 (call_insn_operand): Likewise.
20719 (sibcall_insn_operand): Likewise.
20720 (GOT32_symbol_operand): Likewise.
20721 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
20723 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
20725 PR rtl-optimization/83496
20726 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
20727 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
20728 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
20729 redundant insn, if any.
20730 (relax_delay_slots): Likewise.
20731 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
20733 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
20735 PR tree-optimization/83965
20736 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
20737 that grouped statements are part of a reduction chain. Return
20738 true if the statement is not marked as a reduction itself but
20739 is part of a group.
20740 (vect_recog_dot_prod_pattern): Don't check whether the statement
20741 is part of a group here.
20742 (vect_recog_sad_pattern): Likewise.
20743 (vect_recog_widen_sum_pattern): Likewise.
20745 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
20748 * final.c (rest_of_clean_state): Also look for calls inside sequences.
20750 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
20753 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
20755 (ix86_output_indirect_function_return): New prototype.
20756 (ix86_split_simple_return_pop_internal): Likewise.
20757 * config/i386/i386.c (indirect_return_via_cx): New.
20758 (indirect_return_via_cx_bnd): Likewise.
20759 (indirect_thunk_name): Handle return va CX_REG.
20760 (output_indirect_thunk_function): Create alias for
20761 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
20762 (ix86_output_indirect_jmp): Remove the bool argument.
20763 (ix86_output_indirect_function_return): New function.
20764 (ix86_split_simple_return_pop_internal): Likewise.
20765 * config/i386/i386.md (*indirect_jump): Don't pass false
20766 to ix86_output_indirect_jmp.
20767 (*tablejump_1): Likewise.
20768 (simple_return_pop_internal): Change it to define_insn_and_split.
20769 Call ix86_split_simple_return_pop_internal to split it for
20770 -mfunction-return=.
20771 (simple_return_indirect_internal): Call
20772 ix86_output_indirect_function_return instead of
20773 ix86_output_indirect_jmp.
20775 2018-02-26 Jakub Jelinek <jakub@redhat.com>
20778 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
20779 memset and value initialization afterwards.
20781 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
20783 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
20785 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20788 * common/config/aarch64/aarch64-common.c
20789 (aarch_option_optimization_table[]): Switch
20790 off fomit-frame-pointer
20792 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
20793 Chung-Ju Wu <jasonwucj@gmail.com>
20795 * config/nds32/nds32-multiple.md (load_multiple): Disallow
20797 (store_multiple): Ditto.
20799 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
20801 * config.gcc: Add --with-cpu support for nds32 target.
20802 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
20803 * config/nds32/nds32.opt: Add -mcpu= option.
20805 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
20807 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
20808 isel=yes): Warn for these deprecated options.
20810 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
20812 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
20813 ISA_2_5_MASKS_EMBEDDED.
20815 2018-02-23 Jakub Jelinek <jakub@redhat.com>
20817 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
20818 p->max as pointers rather than using iterative_hash_expr.
20820 2018-02-23 Carl Love <cel@us.ibm.com>
20822 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
20823 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
20825 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
20826 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
20827 P8V_BUILTIN_VEC_VUNSIGNED2.
20829 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
20832 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
20833 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
20835 * lra-constraints.c (curr_insn_transform): Set up
20836 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
20838 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
20839 * lra-eliminations.c (spill_pseudos): Ditto.
20840 (process_insn_for_elimination): Ditto.
20841 * lra-lives.c (reg_early_clobber_p): Use the new macros.
20842 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
20843 LRA_NON_CLOBBERED_ALT.
20845 2018-02-22 Martin Sebor <msebor@redhat.com>
20847 PR tree-optimization/84480
20848 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
20849 to maybe_diag_stxncpy_trunc. Call it.
20850 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
20851 from gimple_fold_builtin_strcpy. Print inlining stack.
20852 (handle_builtin_stxncpy): Print inlining stack.
20853 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
20855 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
20858 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
20859 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
20860 and -fcheck-pointer-bounds are used together.
20861 (indirect_thunk_prefix): New enum.
20862 (indirect_thunk_need_prefix): New function.
20863 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
20864 "_nt" instead of "_bnd" for NOTRACK prefix.
20865 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
20866 (output_indirect_thunk_function): Likewise.
20868 (ix86_code_end): Update output_indirect_thunk_function calls.
20869 (ix86_output_indirect_branch_via_reg): Replace
20870 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
20871 (ix86_output_indirect_branch_via_push): Likewise.
20872 (ix86_output_function_return): Likewise.
20873 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
20874 incompatible with -fcf-protection=branch and
20875 -fcheck-pointer-bounds.
20877 2018-02-22 Steve Ellcey <sellcey@cavium.com>
20880 * config/aarch64/aarch64.c (aarch64_print_address_internal):
20881 Change gcc_assert call to output_operand_lossage.
20883 2018-02-22 Steve Ellcey <sellcey@cavium.com>
20885 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
20887 2018-02-22 DJ Delorie <dj@redhat.com>
20888 Sebastian Perta <sebastian.perta@renesas.com>
20889 Oleg Endo <olegendo@gcc.gnu.org>
20891 * config/rx/rx.c (rx_rtx_costs): New function.
20892 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
20894 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
20896 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
20898 2018-02-22 Martin Liska <mliska@suse.cz>
20901 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
20902 Add "native" as a possible value.
20904 2018-02-22 Martin Liska <mliska@suse.cz>
20907 * config/i386/i386.c (ix86_option_override_internal):
20908 Add "native" as a possible value for -march and -mtune.
20910 2018-02-22 Jakub Jelinek <jakub@redhat.com>
20913 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
20914 to all type variants.
20916 PR tree-optimization/84503
20917 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
20918 width as info->bitpos + info->bitsize - start.
20919 (merged_store_group::merge_overlapping): Simplify width computation.
20920 (check_no_overlap): New function.
20921 (imm_store_chain_info::try_coalesce_bswap): Compute expected
20922 start + width and last_order of the group, fail if check_no_overlap
20924 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
20925 to group if check_no_overlap fails.
20927 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
20929 * config/rs6000/altivec.md: Delete contraint arguments to
20930 define_expand, define_split, and define_peephole2, and in
20931 define_insn_and_split if always unused.
20932 * config/rs6000/darwin.md: Ditto.
20933 * config/rs6000/dfp.md: Ditto.
20934 * config/rs6000/rs6000.md: Ditto.
20935 * config/rs6000/sync.md: Ditto.
20936 * config/rs6000/vector.md: Ditto.
20937 * config/rs6000/vsx.md: Ditto.
20939 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
20941 * config/rs6000/altivec.md: Write output control strings as braced
20942 blocks instead of double-quoted strings.
20943 * config/rs6000/darwin.md: Ditto.
20944 * config/rs6000/rs6000.md: Ditto.
20945 * config/rs6000/vector.md: Ditto.
20946 * config/rs6000/vsx.md: Ditto.
20948 2018-02-21 Jason Merrill <jason@redhat.com>
20950 PR c++/84314 - ICE with templates and fastcall attribute.
20951 * attribs.c (build_type_attribute_qual_variant): Remove assert.
20953 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
20955 * ipa-cp.c (determine_versionability): Fix comment typos.
20957 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
20960 * ipa-cp.c (determine_versionability): Do not version functions caling
20963 2018-02-21 Martin Liska <mliska@suse.cz>
20966 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
20967 Add "native" as a possible value.
20968 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
20969 the macro when native cpu detection is available.
20971 2018-02-21 Martin Liska <mliska@suse.cz>
20974 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
20975 Add "native" as a possible value.
20976 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
20977 when native cpu detection is available.
20979 2018-02-21 Jakub Jelinek <jakub@redhat.com>
20980 Martin Sebor <msebor@redhat.com>
20982 PR tree-optimization/84478
20983 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
20985 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
20986 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
20987 support which is conservatively correct, for 2 only stay conservative
20988 for maxlen. Formatting and comment capitalization fixes. Add STRICT
20989 argument to the 2 argument get_range_strlen, adjust 6 arg
20990 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
20992 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
20993 (gimple_fold_builtin_strlen): Pass true as last argument to
20996 2018-02-20 Martin Sebor <msebor@redhat.com>
20998 PR middle-end/84095
20999 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
21000 (builtin_memref::set_base_and_offset): Same. Handle inner references.
21001 (builtin_memref::builtin_memref): Factor out parts into
21002 set_base_and_offset and call it.
21004 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
21006 PR middle-end/84406
21007 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
21008 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
21009 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
21010 search at the associated MODE_INT.
21012 2018-02-20 Jeff Law <law@redhat.com>
21014 PR middle-end/82123
21015 PR tree-optimization/81592
21016 PR middle-end/79257
21017 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
21018 for range data rather than using global data.
21019 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
21020 range data rather than using global data.
21021 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
21022 pass it to children as needed.
21023 (struct directive::fmtresult): Similarly.
21024 (struct directive::set_width): Similarly.
21025 (struct directive::set_precision): Similarly.
21026 (format_integer, format_directive, parse_directive): Similarly.
21027 (format_none): Accept unnamed vr_values parameter.
21028 (format_percent, format_floating, format_character): Similarly.
21029 (format_string, format_plain): Similarly.
21030 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
21031 the EVRP range analyzer for range data rather than using global data.
21032 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
21033 gimple-ssa-evrp-analyze.h
21034 (class sprintf_dom_walker): Add after_dom_children member function.
21035 Add evrp_range_analyzer member.
21036 (sprintf_dom_walker::before_dom_children): Call into the EVRP
21037 range analyzer as needed.
21038 (sprintf_dom_walker::after_dom_children): New member function.
21039 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
21041 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
21042 (evrp_range_analyzer::pop_to_marker): Likewise.
21044 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
21046 PR tree-optimization/84419
21047 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
21048 with the required type if its current type is compatible but
21051 2018-02-20 Jakub Jelinek <jakub@redhat.com>
21053 PR middle-end/82004
21054 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
21055 after vectorization.
21057 2018-02-20 Martin Liska <mliska@suse.cz>
21060 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
21061 possible values if we don't have a hint.
21063 2018-02-20 Martin Liska <mliska@suse.cz>
21067 * final.c (shorten_branches): Build align_tab array with one
21069 * opts.c (finish_options): Add alignment option limit check.
21070 (MAX_CODE_ALIGN): Likewise.
21071 (MAX_CODE_ALIGN_VALUE): Likewise.
21072 * doc/invoke.texi: Document maximum allowed option value for
21073 all -falign-* options.
21075 2018-02-19 Jakub Jelinek <jakub@redhat.com>
21078 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
21079 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
21080 * var-tracking.c (emit_note_insn_var_location): Remove all references
21081 to NOTE_INSN_CALL_ARG_LOCATION.
21082 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
21083 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
21084 Use copy_rtx_if_shared.
21085 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
21086 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
21087 (dwarf2out_var_location): Remove handling of
21088 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
21090 * final.c (final_scan_insn): Remove all references to
21091 NOTE_INSN_CALL_ARG_LOCATION.
21092 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
21093 before dumping final insns.
21094 * except.c (emit_note_eh_region_end): Remove all references to
21095 NOTE_INSN_CALL_ARG_LOCATION.
21096 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
21097 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
21098 * config/arc/arc.c (hwloop_optimize): Likewise.
21099 * config/arm/arm.c (create_fix_barrier): Likewise.
21100 * config/s390/s390.c (s390_chunkify_start): Likewise.
21101 * config/sh/sh.c (find_barrier): Likewise.
21102 * config/i386/i386.c (rest_of_insert_endbranch,
21103 ix86_seh_fixup_eh_fallthru): Likewise.
21104 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
21105 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
21106 * config/frv/frv.c (frv_function_prologue): Likewise.
21107 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
21109 (note_outside_basic_block_p): Remove all references to
21110 NOTE_INSN_CALL_ARG_LOCATION.
21111 * gengtype.c (adjust_field_rtx_def): Likewise.
21112 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
21114 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
21115 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
21118 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
21121 PR tree-optimization/84452
21122 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
21123 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
21126 2018-02-19 Martin Liska <mliska@suse.cz>
21129 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
21131 2018-02-19 Martin Liska <mliska@suse.cz>
21132 Richard Sandiford <richard.sandiford@linaro.org>
21134 PR tree-optimization/82491
21135 * gimple-fold.c (get_base_constructor): Make earlier bail out
21138 2018-02-19 Carl Love <cel@us.ibm.com>
21140 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
21141 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
21143 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
21144 P8V_BUILTIN_VEC_NEG.
21146 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
21148 * config/rl78/rl78.md (movdf): New define expand.
21150 2018-02-19 Martin Liska <mliska@suse.cz>
21153 * doc/invoke.texi: Fix typo.
21154 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
21156 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
21158 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
21159 handle rs6000_single_float and rs6000_double_float specially for
21162 2018-02-16 Jeff Law <law@redhat.com>
21164 * config/rx/rx.c (add_pop_cfi_notes): New function.;
21165 (pop_regs): Use it.
21167 2018-02-16 Jakub Jelinek <jakub@redhat.com>
21170 * ipa-inline.c (inline_small_functions): Fix a typo.
21172 2018-02-16 Nathan Sidwell <nathan@acm.org>
21174 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
21176 2018-02-16 Carl Love <cel@us.ibm.com>
21178 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
21179 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
21180 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
21181 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
21182 expansion to P8V_BUILTIN_VEC_FLOAT2.
21184 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
21186 PR rtl-optimization/70023
21187 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
21188 src_regno into account.
21190 2018-02-16 Carl Love <cel@us.ibm.com>
21192 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
21193 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
21194 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
21195 * config/rs6000/rs6000.c: Remove case statements for
21196 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
21197 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
21198 and P9V_BUILTIN_VEC_VINSERT4B.
21199 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
21200 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
21201 * config/rs6000/vsx.md:
21202 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
21205 2018-02-16 Carl Love <cel@us.ibm.com>
21207 * config/rs6000/altivec.h: Add builtin names vec_extract4b
21209 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
21211 * config/rs6000/rs6000-c.c: Add the definitions for
21212 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
21213 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
21214 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
21215 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
21216 definition for insert4b and define insn *insert3b_internal.
21217 * doc/extend.texi: Add documentation for vec_extract4b.
21219 2018-02-16 Nathan Sidwell <nathan@acm.org>
21221 * doc/extend.texi (Backwards Compatibility): Mention friend
21222 injection. Note for-scope is deprecated.
21223 * doc/invoke.texi (-ffriend-injection): Deprecate.
21225 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
21227 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
21228 that moved to I2, also allow destinations that are a paradoxical
21229 subreg (instead of a normal reg).
21231 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
21234 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
21237 2018-02-16 Richard Biener <rguenther@suse.de>
21239 PR tree-optimization/84037
21240 PR tree-optimization/84016
21242 * config/i386/i386.c (ix86_builtin_vectorization_cost):
21243 Adjust vec_construct for the fact we need additional higher latency
21244 128bit inserts for AVX256 and AVX512 vector builds.
21245 (ix86_add_stmt_cost): Scale vector construction cost for
21248 2018-02-16 Richard Biener <rguenther@suse.de>
21250 PR tree-optimization/84417
21251 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
21252 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
21253 (non_rewritable_lvalue_p): Likewise, use poly-ints.
21255 2018-02-16 Martin Liska <mliska@suse.cz>
21258 * internal-fn.def (ASAN_CHECK): Set proper flags.
21259 (ASAN_MARK): Likewise.
21261 2018-02-16 Julia Koval <julia.koval@intel.com>
21263 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
21264 from PTA_CANNONLAKE.
21266 2018-02-16 Jakub Jelinek <jakub@redhat.com>
21269 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
21270 Use ++iter rather than iter++ for std::list iterators.
21271 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
21272 defer deleting them until all nodes in the forest are processed. Do
21273 free even leaf nodes. Change to_process into auto_vec.
21276 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
21277 * vec.h (vec_default_construct): Use memset instead of placement new
21278 if BROKEN_VALUE_INITIALIZATION is defined.
21279 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
21280 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
21283 PR rtl-optimization/83723
21284 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
21285 * lra.c (lra_substitute_pseudo): Likewise. If true, use
21286 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
21288 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
21290 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
21292 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
21294 PR rtl-optimization/81443
21295 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
21296 from inner REGs to paradoxical SUBREGs.
21298 2018-02-16 Richard Biener <rguenther@suse.de>
21300 PR tree-optimization/84399
21301 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
21302 For operands we can analyze at their definition make sure we can
21303 analyze them at each use as well.
21305 2018-02-16 Richard Biener <rguenther@suse.de>
21307 PR tree-optimization/84190
21308 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
21309 volatile accesses if the decl isn't volatile.
21311 2018-02-15 Jason Merrill <jason@redhat.com>
21313 PR c++/84314 - ICE with templates and fastcall attribute.
21314 * attribs.c (build_type_attribute_qual_variant): Don't clobber
21315 TYPE_CANONICAL on an existing type.
21317 2018-02-15 Jakub Jelinek <jakub@redhat.com>
21319 PR tree-optimization/84383
21320 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
21321 dstoff nor call operand_equal_p if dstbase is NULL.
21323 PR tree-optimization/84334
21324 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
21325 also a CONSTANT_CLASS_P, punt.
21327 2018-02-14 Jim Wilson <jimw@sifive.com>
21329 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
21330 first SMALL_OPERAND check. New local min_second_step. Move assert
21331 to where locals are set. Add TARGET_RVC support.
21332 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
21334 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
21336 * doc/invoke.texi: Correct -Wformat-overflow code sample.
21338 2018-02-14 Martin Sebor <msebor@redhat.com>
21340 PR tree-optimization/83698
21341 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
21342 arrays constrain the offset range to their bounds.
21343 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
21344 (builtin_access::overlap): Avoid setting the size of overlap if it's
21346 (maybe_diag_overlap): Also consider arrays when deciding what values
21347 of offsets to include in diagnostics.
21349 2018-02-14 Martin Sebor <msebor@redhat.com>
21352 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
21353 that correspond to the kind of a declaration.
21355 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
21358 * config/pa/pa.md: Load address of PIC label using the linkage table
21359 if the label is nonlocal.
21361 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
21363 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
21364 warning message if user requests -maltivec=be.
21365 * doc/invoke.texi: Document deprecation of -maltivec=be.
21367 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
21370 * config/rs6000/rs6000-c.c: Update definitions for
21371 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
21372 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
21374 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
21377 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
21378 add _get_ssp intrinsics. Remove argument from
21379 __builtin_ia32_rdssp[d|q].
21380 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
21381 * config/i386/i386-builtin.def: Remove argument from
21382 __builtin_ia32_rdssp[d|q].
21383 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
21384 ix86_expand_special_args_builtin for _rdssp[d|q].
21385 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
21386 Clear register before usage.
21387 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
21388 Add documentation for new _get_ssp and _inc_ssp intrinsics.
21390 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
21392 PR tree-optimization/84357
21393 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
21394 operand 1 of an ARRAY_REF too.
21396 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
21399 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
21400 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
21402 (set_of_reg): New struct.
21403 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
21404 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
21405 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
21407 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
21408 Split into bitclr, bitset, bitinvert patterns if appropriate.
21409 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
21410 use rx_fuse_in_memory_bitop.
21411 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
21412 to named insn, correct maximum insn length.
21414 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
21417 * machmode.def: Define a complex mode for PARTIAL_INT.
21418 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
21420 * doc/rtl.texi: Document CSPImode.
21421 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
21423 (msp430_hard_regno_nregs_with_padding): Likewise.
21425 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
21428 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
21430 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
21432 PR rtl-optimization/84169
21433 * combine.c (try_combine): New variable split_i2i3. Set it to true if
21434 we generated a parallel as new i3 and we split that to new i2 and i3
21435 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
21436 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
21437 those to i2, not i1. Partially rewrite this scan code.
21439 2018-02-13 Jakub Jelinek <jakub@redhat.com>
21442 * stor-layout.c (place_field): For variable length fields, adjust
21443 offset_align afterwards not just based on the field's alignment,
21444 but also on the size.
21446 PR middle-end/84309
21447 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
21448 of exps and logs in the use_exp2 case.
21450 2018-02-13 Jeff Law <law@redhat.com>
21452 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
21453 entry for "vector".
21455 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
21458 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
21462 * common.opt (gas-loc-support, gas-locview-support): New.
21463 (ginline-points, ginternal-reset-location-views): New.
21464 * doc/invoke.texi: Document them. Use @itemx where intended.
21465 (gvariable-location-views): Adjust.
21466 * target.def (reset_location_view): New.
21467 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
21468 (TARGET_RESET_LOCATION_VIEW): New.
21469 * doc/tm.texi: Rebuilt.
21470 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
21471 (dwarf2out_default_as_locview_support): New.
21472 (output_asm_line_debug_info): Use option variables.
21473 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
21474 (output_loc_list): Likewise.
21475 (add_high_low_attributes): Check option variables.
21476 Don't output entry view attribute in strict mode.
21477 (gen_inlined_subroutine_die): Check option variables.
21478 (dwarf2out_inline_entry): Likewise.
21479 (init_sections_and_labels): Likewise.
21480 (dwarf2out_early_finish): Likewise.
21481 (maybe_reset_location_view): New, from...
21482 (dwarf2out_var_location): ... here. Call it.
21483 * debug.h (dwarf2out_default_as_loc_support): Declare.
21484 (dwarf2out_default_as_locview_support): Declare.
21485 * hooks.c (hook_int_rtx_insn_0): New.
21486 * hooks.h (hook_int_rtx_insn_0): Declare.
21487 * toplev.c (process_options): Take -gas-loc-support and
21488 -gas-locview-support from dwarf2out. Enable
21489 -gvariable-location-views by default only with locview
21490 assembler support. Enable -ginternal-reset-location-views by
21491 default only if the target defines the corresponding hook.
21492 Enable -ginline-points by default if location views are
21493 enabled; force it disabled if statement frontiers are
21495 * tree-inline.c (expand_call_inline): Check option variables.
21496 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
21498 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
21500 PR tree-optimization/84321
21501 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
21502 handling. Also check whether the anti-range contains any values
21503 that satisfy the mask; switch to a VR_RANGE if not.
21505 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
21508 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
21510 2018-02-13 Martin Jambor <mjambor@suse.cz>
21513 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
21514 of call statements, also set location of a load to a temporary.
21516 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
21518 * config/rl78/rl78.c (add_vector_labels): New function.
21519 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
21520 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
21521 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
21522 which checks that no arguments are passed.
21523 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
21524 * doc/extend.texi: Documentation for the new attribute.
21526 2018-02-13 Andreas Schwab <schwab@suse.de>
21528 * config/riscv/linux.h (CPP_SPEC): Define.
21530 2018-02-13 Jakub Jelinek <jakub@redhat.com>
21533 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
21534 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
21535 OPTION_MASK_ISA_AES as first argument to def_builtin_const
21536 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
21537 instead of OPTION_MASK_ISA_PCLMUL as first argument to
21538 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
21539 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
21540 temporarily for AES and PCLMUL builtins.
21542 PR tree-optimization/84339
21543 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
21544 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
21547 PR middle-end/84309
21548 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
21549 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
21550 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
21552 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
21554 * omp-simd-clone.h: New file.
21555 * omp-simd-clone.c: Include omp-simd-clone.h.
21556 (expand_simd_clones): No longer static.
21557 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
21558 cgraph.h and omp-simd-clone.h.
21559 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
21560 (vect_recog_widen_shift_pattern): Formatting fix.
21561 (vect_pattern_recog_1): Don't check optab for calls.
21564 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
21565 operands[2] into a REG before using gen_lowpart on it.
21567 2018-02-12 Jeff Law <law@redhat.com>
21570 * config/sh/sh.c (find_barrier): Consider a sibling call
21573 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
21574 successfully back substituting a reg.
21576 2018-02-12 Richard Biener <rguenther@suse.de>
21578 PR tree-optimization/84037
21579 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
21580 parameter, move visited init to caller.
21581 (vect_slp_analyze_operations): Separate cost from validity
21582 check, initialize visited once for all instances.
21583 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
21585 * tree-vect-stmts.c (vect_model_simple_cost): Make early
21587 (vect_model_promotion_demotion_cost): Likewise.
21588 (vectorizable_bswap): Guard cost modeling with !slp_node
21589 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
21591 (vectorizable_call): Likewise.
21592 (vectorizable_conversion): Likewise.
21593 (vectorizable_assignment): Likewise.
21594 (vectorizable_shift): Likewise.
21595 (vectorizable_operation): Likewise.
21596 (vectorizable_store): Likewise.
21597 (vectorizable_load): Likewise.
21598 (vectorizable_condition): Likewise.
21599 (vectorizable_comparison): Likewise.
21601 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
21604 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
21605 (ASAN_MARK): Fix fnspec to account for return value, change pointer
21606 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
21608 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
21610 PR middle-end/83665
21611 * params.def (inline-min-speedup): Increase from 8 to 15.
21612 (max-inline-insns-auto): Decrease from 40 to 30.
21613 * ipa-split.c (consider_split): Add some buffer for function to
21614 be considered inlining candidate.
21615 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
21618 2018-02-12 Richard Biener <rguenther@suse.de>
21620 PR tree-optimization/84037
21621 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
21622 matched stmts if we cannot swap the non-matched ones.
21624 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
21626 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
21627 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
21628 _mm_maskz_scalef_round_ss): New intrinsics.
21629 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
21630 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
21631 __builtin_ia32_scalefss_round): Remove.
21632 (__builtin_ia32_scalefsd_mask_round,
21633 __builtin_ia32_scalefss_mask_round): New intrinsics.
21634 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
21635 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
21636 ((match_operand:VF_128 2 "<round_nimm_predicate>"
21637 "<round_constraint>")): Changed to ...
21638 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
21639 "<round_scalar_constraint>")): ... this.
21640 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
21641 %0, %1, %2<round_op3>}"): Changed to ...
21642 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
21643 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
21644 %2<round_scalar_mask_op3>}"): ... this.
21645 * config/i386/subst.md (round_scalar_nimm_predicate): New.
21647 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
21649 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
21650 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
21651 (_mm_maskz_sqrt_round_ss): New intrinsics.
21652 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
21653 (__builtin_ia32_sqrtsd_mask_round)
21654 (__builtin_ia32_sqrtss_mask_round): New builtins.
21655 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
21656 (__builtin_ia32_sqrtss_round): Remove.
21657 (__builtin_ia32_sqrtsd_mask_round)
21658 (__builtin_ia32_sqrtss_mask_round): New builtins.
21659 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
21660 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
21661 ((match_operand:VF_128 1 "vector_operand"
21662 "xBm,<round_constraint>")): Changed to ...
21663 ((match_operand:VF_128 1 "vector_operand"
21664 "xBm,<round_scalar_constraint>")): ... this.
21665 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
21666 %0, %2, %<iptr>1<round_op3>}): Changed to ...
21667 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
21668 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
21669 %<iptr>1<round_scalar_mask_op3>}): ... this.
21670 ((set_attr "prefix" "<round_prefix>")): Changed to ...
21671 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
21673 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
21676 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
21677 Cast vec_cmpeq result to correct type.
21678 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
21679 Cast vec_cmpgt result to correct type.
21681 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
21683 * final.c (final_scan_insn_1): Renamed from...
21684 (final_scan_insn): ... this. New wrapper, to recover
21685 seen from the outermost call in recursive ones.
21686 * config/sparc/sparc.c (output_return): Drop seen from call.
21687 (output_sibcall): Likewise.
21688 * config/visium/visium.c (output_branch): Likewise.
21690 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
21692 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
21695 2018-02-10 Alan Modra <amodra@gmail.com>
21698 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
21699 Specify LR as an input.
21701 2018-02-10 Jakub Jelinek <jakub@redhat.com>
21704 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
21705 remove_member_access_dummy_vars): New functions.
21706 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
21707 lower_omp_1, execute_lower_omp): Use them.
21709 PR rtl-optimization/84308
21710 * shrink-wrap.c (spread_components): Release todo vector.
21712 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
21714 PR rtl-optimization/57193
21715 * ira-color.c (struct allocno_color_data): Add member
21716 conflict_allocno_hard_prefs.
21717 (update_conflict_allocno_hard_prefs): New.
21718 (bucket_allocno_compare_func): Add a preference based on
21719 conflict_allocno_hard_prefs.
21720 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
21721 (color_allocnos): Remove a dead code. Initiate
21722 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
21724 2018-02-09 Jakub Jelinek <jakub@redhat.com>
21727 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
21728 constraint from =wa to wa. Avoid a subreg on the output operand,
21729 instead use a pseudo and subreg it in a move.
21730 (p9_xxbrd_<mode>): Changed to ...
21731 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
21732 (p9_xxbrd_v2df): New expander.
21733 (p9_xxbrw_<mode>): Changed to ...
21734 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
21735 (p9_xxbrw_v4sf): New expander.
21737 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
21739 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
21741 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
21744 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
21745 multiply in 32-bit mode.
21746 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
21747 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
21750 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
21752 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
21753 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
21754 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
21755 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
21757 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
21760 * dwarf2out.c (is_trivial_indirect_ref): New function.
21761 (dwarf2out_late_global_decl): Do not generate a location
21762 attribute for variables that have a non-trivial DECL_VALUE_EXPR
21763 and that are not defined in the current unit.
21765 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
21767 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
21768 instead of a libcall for UNORDERED.
21770 2018-02-09 Tamar Christina <tamar.christina@arm.com>
21773 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
21774 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
21776 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21779 * config/s390/s390.c (s390_set_current_function): Invoke
21780 s390_indirect_branch_settings also if fndecl didn't change.
21782 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
21784 * config/rs6000/rs6000.md (blockage): Set length to zero.
21786 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
21788 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
21790 2018-02-09 Jakub Jelinek <jakub@redhat.com>
21793 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
21794 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
21798 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
21799 PARALLEL incoming that failed vt_get_decl_and_offset check.
21801 PR middle-end/84237
21802 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
21803 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
21805 (get_variable_section): For decls in named .bss* sections pass true as
21806 second argument to bss_initializer_p.
21808 2018-02-09 Marek Polacek <polacek@redhat.com>
21809 Jakub Jelinek <jakub@redhat.com>
21812 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
21813 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
21814 Sync some changes from cxx_fold_indirect_ref.
21816 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
21818 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
21820 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
21821 (BLOCK_INLINE_ENTRY_LABEL): New.
21822 (dwarf2out_var_location): Disregard inline entry markers.
21823 (inline_entry_data): New struct.
21824 (inline_entry_data_hasher): New hashtable type.
21825 (inline_entry_data_hasher::hash): New.
21826 (inline_entry_data_hasher::equal): New.
21827 (inline_entry_data_table): New variable.
21828 (add_high_low_attributes): Add DW_AT_entry_pc and
21829 DW_AT_GNU_entry_view attributes if a pending entry is found
21830 in inline_entry_data_table. Add old entry_pc attribute only
21831 if debug nonbinding markers are disabled.
21832 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
21833 markers are enabled.
21834 (block_within_block_p, dwarf2out_inline_entry): New.
21835 (dwarf2out_finish): Check that no entries remained in
21836 inline_entry_data_table.
21837 * final.c (reemit_insn_block_notes): Handle inline entry notes.
21838 (final_scan_insn, notice_source_line): Likewise.
21839 (rest_of_clean_state): Skip inline entry markers.
21840 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
21842 * gimple.c (gimple_build_debug_inline_entry): New.
21843 * gimple.h (enum gimple_debug_subcode): Add
21844 GIMPLE_DEBUG_INLINE_ENTRY.
21845 (gimple_build_debug_inline_entry): Declare.
21846 (gimple_debug_inline_entry_p): New.
21847 (gimple_debug_nonbind_marker_p): Adjust.
21848 * insn-notes.def (INLINE_ENTRY): New.
21849 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
21850 inline entry marker notes.
21851 (print_insn): Likewise.
21852 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
21853 (INSN_DEBUG_MARKER_KIND): Likewise.
21854 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
21855 * tree-inline.c (expand_call_inline): Build and insert
21856 debug_inline_entry stmt.
21857 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
21858 inline entry blocks early, if nonbind markers are enabled.
21859 (dump_scope_block): Dump fragment info.
21860 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
21861 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
21862 (gimple_build_debug_inline_entry): New.
21863 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
21864 Enable/disable inline entry points too.
21865 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
21866 (DEBUG_INSN): Describe inline entry markers.
21868 * common.opt (gvariable-location-views): New.
21869 (gvariable-location-views=incompat5): New.
21870 * config.in: Rebuilt.
21871 * configure: Rebuilt.
21872 * configure.ac: Test assembler for view support.
21873 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
21874 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
21875 * dwarf2out.c (var_loc_view): New typedef.
21876 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
21877 (dwarf2out_locviews_in_attribute): New.
21878 (dwarf2out_locviews_in_loclist): New.
21879 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
21880 (enum dw_line_info_opcode): Add LI_adv_address.
21881 (struct dw_line_info_table): Add view.
21882 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
21883 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
21884 (zero_view_p): New variable.
21885 (ZERO_VIEW_P): New macro.
21886 (output_asm_line_debug_info): New.
21887 (struct var_loc_node): Add view.
21888 (add_AT_view_list, AT_loc_list): New.
21889 (add_var_loc_to_decl): Add view param. Test it against last.
21890 (new_loc_list): Add view params. Record them.
21891 (AT_loc_list_ptr): Handle loc and view lists.
21892 (view_list_to_loc_list_val_node): New.
21893 (print_dw_val): Handle dw_val_class_view_list.
21894 (size_of_die): Likewise.
21895 (value_format): Likewise.
21896 (loc_list_has_views): New.
21897 (gen_llsym): Set vl_symbol too.
21898 (maybe_gen_llsym, skip_loc_list_entry): New.
21899 (dwarf2out_maybe_output_loclist_view_pair): New.
21900 (output_loc_list): Output view list or entries too.
21901 (output_view_list_offset): New.
21902 (output_die): Handle dw_val_class_view_list.
21903 (output_dwarf_version): New.
21904 (output_compilation_unit_header): Use it.
21905 (output_skeleton_debug_sections): Likewise.
21906 (output_rnglists, output_line_info): Likewise.
21907 (output_pubnames, output_aranges): Update version comments.
21908 (output_one_line_info_table): Output view numbers in asm comments.
21909 (dw_loc_list): Determine current endview, pass it to new_loc_list.
21910 Call maybe_gen_llsym.
21911 (loc_list_from_tree_1): Adjust.
21912 (add_AT_location_description): Create view list attribute if
21913 needed, check it's absent otherwise.
21914 (convert_cfa_to_fb_loc_list): Adjust.
21915 (maybe_emit_file): Call output_asm_line_debug_info for test.
21916 (dwarf2out_var_location): Reset views as needed. Precompute
21917 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
21918 attribute. Set view.
21919 (new_line_info_table): Reset next view.
21920 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
21921 (dwarf2out_source_line): Likewise. Output view resets and labels to
21922 the assembler, or select appropriate line info opcodes.
21923 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
21924 (optimize_string_length): Catch it. Adjust.
21925 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
21926 dw_val_class_view_list, and remove it if no longer needed.
21927 (hash_loc_list): Hash view numbers.
21928 (loc_list_hasher::equal): Compare them.
21929 (optimize_location_lists): Check whether a view list symbol is
21930 needed, and whether the locview attribute is present, and
21931 whether they match. Remove the locview attribute if no longer
21933 (index_location_lists): Call skip_loc_list_entry for test.
21934 (dwarf2out_finish): Call output_asm_line_debug_info for test.
21935 Use output_dwarf_version.
21936 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
21937 (struct dw_val_node): Add val_view_list.
21938 * final.c (SEEN_NEXT_VIEW): New.
21939 (set_next_view_needed): New.
21940 (clear_next_view_needed): New.
21941 (maybe_output_next_view): New.
21942 (final_start_function): Rename to...
21943 (final_start_function_1): ... this. Take pointer to FIRST,
21944 add SEEN parameter. Emit param bindings in the initial view.
21945 (final_start_function): Reintroduce SEEN-less interface.
21946 (final): Rename to...
21947 (final_1): ... this. Take SEEN parameter. Output final pending
21948 next view at the end.
21949 (final): Reintroduce seen-less interface.
21950 (final_scan_insn): Output pending next view before switching
21951 sections or ending a block. Mark the next view as needed when
21952 outputting variable locations. Notify debug backend of section
21953 changes, and of location view changes.
21954 (rest_of_handle_final): Adjust.
21955 * toplev.c (process_options): Autodetect value for debug variable
21956 location views option. Warn on incompat5 without -gdwarf-5.
21957 * doc/invoke.texi (gvariable-location-views): New.
21958 (gvariable-location-views=incompat5): New.
21959 (gno-variable-location-views): New.
21961 2018-02-08 David Malcolm <dmalcolm@redhat.com>
21963 PR tree-optimization/84136
21964 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
21965 that the result of find_edge is non-NULL.
21967 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
21970 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
21971 storing integer register in SImode. Fix cost of 256 and 512
21972 byte aligned SSE register store.
21974 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
21976 * config/i386/i386.c (ix86_multiplication_cost): Fix
21977 multiplication cost for TARGET_AVX512DQ.
21979 2018-02-08 Marek Polacek <polacek@redhat.com>
21981 PR tree-optimization/84238
21982 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
21985 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
21987 PR tree-optimization/84265
21988 * tree-vect-stmts.c (vectorizable_store): Don't treat
21989 VMAT_CONTIGUOUS accesses as grouped.
21990 (vectorizable_load): Likewise.
21992 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
21994 PR tree-optimization/81635
21995 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
21996 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
21997 (test_round_for_mask): New functions.
21998 (wide_int_cc_tests): Call test_round_for_mask.
21999 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
22000 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
22001 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
22002 range returned by get_range_info.
22004 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
22007 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
22008 * symtab.c: Include builtins.h
22009 (symtab_node::output_to_lto_symbol_table_p): Move here
22010 from lto-streamer-out.c:output_symbol_p.
22011 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
22012 (output_symbol_p): Move all logic to symtab.c
22013 (produce_symtab): Update.
22015 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22017 * config/s390/s390-opts.h (enum indirect_branch): Define.
22018 * config/s390/s390-protos.h (s390_return_addr_from_memory)
22019 (s390_indirect_branch_via_thunk)
22020 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
22021 (enum s390_indirect_branch_type): Define.
22022 * config/s390/s390.c (struct s390_frame_layout, struct
22023 machine_function): Remove.
22024 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
22025 (indirect_branch_table_label_no, indirect_branch_table_name):
22027 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
22028 (enum s390_indirect_branch_option): Define.
22029 (s390_return_addr_from_memory): New function.
22030 (s390_handle_string_attribute): New function.
22031 (s390_attribute_table): Add new attribute handler.
22032 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
22033 (s390_indirect_branch_via_thunk): New function.
22034 (s390_indirect_branch_via_inline_thunk): New function.
22035 (s390_function_ok_for_sibcall): When jumping via thunk disallow
22036 sibling call optimization for non z10 compiles.
22037 (s390_emit_call): Force indirect branch target to be a single
22038 register. Add r1 clobber for non-z10 compiles.
22039 (s390_emit_epilogue): Emit return jump via return_use expander.
22040 (s390_reorg): Handle JUMP_INSNs as execute targets.
22041 (s390_option_override_internal): Perform validity checks for the
22042 new command line options.
22043 (s390_indirect_branch_attrvalue): New function.
22044 (s390_indirect_branch_settings): New function.
22045 (s390_set_current_function): Invoke s390_indirect_branch_settings.
22046 (s390_output_indirect_thunk_function): New function.
22047 (s390_code_end): Implement target hook.
22048 (s390_case_values_threshold): Implement target hook.
22049 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
22051 * config/s390/s390.h (struct s390_frame_layout)
22052 (struct machine_function): Move here from s390.c.
22053 (TARGET_INDIRECT_BRANCH_NOBP_RET)
22054 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
22055 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
22056 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
22057 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
22058 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
22059 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
22060 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
22061 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
22062 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
22063 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
22064 (mnemonic attribute): Add values which aren't recognized
22066 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
22067 pattern for branch conversion. Fix mnemonic attribute.
22068 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
22069 indirect branch via thunk if requested.
22070 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
22071 ("*indirect_jump"): Disable for branch conversion using out of
22073 ("indirect_jump_via_thunk<mode>_z10")
22074 ("indirect_jump_via_thunk<mode>")
22075 ("indirect_jump_via_inlinethunk<mode>_z10")
22076 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
22077 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
22078 ("casesi_jump_via_inlinethunk<mode>_z10")
22079 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
22080 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
22081 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
22082 ("*indirect2_jump"): Disable for branch conversion.
22083 ("casesi_jump"): Turn into expander and expand patterns for branch
22085 ("return_use"): New expander.
22086 ("*return"): Emit return via thunk and rename it to ...
22087 ("*return<mode>"): ... this one.
22088 * config/s390/s390.opt: Add new options and and enum for the
22091 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
22093 * lra-constraints.c (match_reload): Unconditionally use
22094 gen_lowpart_SUBREG, rather than selecting between that
22095 and equivalent gen_rtx_SUBREG code.
22097 2018-02-08 Richard Biener <rguenther@suse.de>
22099 PR tree-optimization/84233
22100 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
22101 changed flag instead of boguously re-using phi_inserted.
22103 2018-02-08 Martin Jambor <mjambor@suse.cz>
22105 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
22106 static local variables.
22108 2018-02-08 Richard Biener <rguenther@suse.de>
22110 PR tree-optimization/84278
22111 * tree-vect-stmts.c (vectorizable_store): When looking for
22112 smaller vector types to perform grouped strided loads/stores
22113 make sure the mode is supported by the target.
22114 (vectorizable_load): Likewise.
22116 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
22118 * config/aarch64/aarch64.c (aarch64_components_for_bb):
22119 Increase LDP/STP opportunities by adding adjacent callee-saves.
22121 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
22123 PR rtl-optimization/84068
22124 PR rtl-optimization/83459
22125 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
22127 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
22129 PR tree-optimization/84224
22130 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
22131 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
22132 non-zero arguments.
22134 2018-02-07 Iain Sandoe <iain@codesourcery.com>
22137 * config/rs6000/altivec.md (*restore_world): Remove LR use.
22138 * config/rs6000/predicates.md (restore_world_operation): Adjust op
22139 count, remove one USE.
22141 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
22143 * doc/install.texi (Configuration): Document the
22144 --with-long-double-format={ibm,ieee} PowerPC configuration
22148 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
22149 Convert from define_expand to be define_insn_and_split. Rework
22150 float/double/_Float128 conversions to QI/HI/SImode to work with
22151 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
22152 conversions to QI/HImode types did a store and then a load to
22153 truncate the value. For conversions to VSX registers, don't split
22154 the insn, instead emit the code directly. Use the code iterator
22155 any_fix to combine signed and unsigned conversions.
22156 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
22157 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
22158 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
22159 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
22160 (fix_<mode>di2_hw): Likewise.
22161 (fixuns_<mode>di2_hw): Likewise.
22162 (fix_<mode>si2_hw): Likewise.
22163 (fixuns_<mode>si2_hw): Likewise.
22164 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
22165 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
22166 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
22167 fix<uns>_trunc<SFDF:mode>si2_p8.
22168 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
22169 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22170 (fix<uns>_<mode>_mem): Likewise.
22171 (fctiw<u>z_<mode>_mem): Likewise.
22172 (fix<uns>_<mode>_mem): Likewise.
22173 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
22174 the register allocator from doing a direct move to the GPRs to do
22175 a store, and instead use the ISA 3.0 store byte/half-word from
22176 vector register instruction. For IEEE 128-bit floating point,
22177 also optimize stores of 32-bit ints.
22178 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
22180 2018-02-07 Alan Hayward <alan.hayward@arm.com>
22182 * genextract.c (push_pathstr_operand): New function to support
22184 (walk_rtx): Call push_pathstr_operand.
22185 (print_path): Support [a-zA-Z].
22187 2018-02-07 Richard Biener <rguenther@suse.de>
22189 PR tree-optimization/84037
22190 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
22191 (cse_and_gimplify_to_preheader): Declare.
22192 (vect_get_place_in_interleaving_chain): Likewise.
22193 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
22195 (_loop_vec_info::~_loop_vec_info): Delete it.
22196 (cse_and_gimplify_to_preheader): New function.
22197 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
22198 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
22199 (vectorizable_load): Likewise. For grouped stores always base
22200 the IV on the first element.
22201 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
22202 condition before gimplifying.
22204 2018-02-07 Jakub Jelinek <jakub@redhat.com>
22206 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
22207 *DIV_EXPR and *MOD_EXPR.
22209 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
22212 * config/i386/i386.c (ix86_option_override_internal): Mask out
22213 the CF_SET bit when checking -fcf-protection.
22215 2018-02-07 Tom de Vries <tom@codesourcery.com>
22218 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
22221 2018-02-07 Richard Biener <rguenther@suse.de>
22223 PR tree-optimization/84204
22224 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
22227 PR tree-optimization/84205
22228 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
22229 special-case isl_ast_op_zdiv_r.
22231 PR tree-optimization/84223
22232 * graphite-scop-detection.c (gather_bbs::before_dom_children):
22233 Only add conditions from within the region.
22234 (gather_bbs::after_dom_children): Adjust.
22236 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
22239 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
22240 * config/avr/avr.md: Only post-reload split REG-REG moves if
22241 either register is GENERAL_REG_P.
22243 2018-02-07 Jakub Jelinek <jakub@redhat.com>
22245 PR tree-optimization/84235
22246 * tree-ssa-scopedtables.c
22247 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
22248 if the subtraction is performed in floating point type where NaNs are
22249 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
22250 build 1. Formatting fix.
22252 2018-02-06 Jakub Jelinek <jakub@redhat.com>
22255 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
22256 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
22257 and skip it regardless of bb boundaries. Use CALL_P macro,
22258 don't test INSN_P (insn) together with CALL_P or JUMP_P check
22259 unnecessarily, formatting fix.
22261 2018-02-06 Michael Collison <michael.collison@arm.com>
22263 * config/arm/thumb2.md:
22264 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
22265 (*thumb_mov_notscc): Ditto.
22267 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
22270 * config/rs6000/rs6000.md (su code attribute): Use "u" for
22271 unsigned_fix, not "s".
22273 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22275 * configure.ac (gcc_fn_eh_frame_ro): New function.
22276 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
22277 correct .eh_frame permissions.
22278 * configure: Regenerate.
22280 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
22282 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
22283 irrelevant options.
22285 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22287 * config/rs6000/rs6000.c (rs6000_option_override_internal):
22288 Display warning message for -mno-speculate-indirect-jumps.
22290 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
22292 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
22294 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
22296 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
22298 PR tree-optimization/84225
22299 * tree-eh.c (find_trapping_overflow): Only call
22300 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
22302 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
22305 * config/i386/i386.c: Reimplement the check of possible options
22306 -mibt/-mshstk conbination. Change error messages.
22307 * doc/invoke.texi: Fix a typo: remove extra '='.
22309 2018-02-06 Marek Polacek <polacek@redhat.com>
22311 PR tree-optimization/84228
22312 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
22314 2018-02-06 Tamar Christina <tamar.christina@arm.com>
22317 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
22318 emitted arch directives.
22319 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
22320 __ARM_FEATURE_COPROC before changing architectures.
22322 2018-02-06 Richard Biener <rguenther@suse.de>
22324 * config/i386/i386.c (print_reg): Fix typo.
22325 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
22327 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
22329 * configure: Regenerate.
22331 2018-02-05 Martin Sebor <msebor@redhat.com>
22333 PR tree-optimization/83369
22334 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
22337 2018-02-05 Martin Liska <mliska@suse.cz>
22339 * doc/invoke.texi: Cherry-pick upstream r323995.
22341 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
22343 * ira.c (ira_init_register_move_cost): Adjust comment.
22345 2018-02-05 Martin Liska <mliska@suse.cz>
22347 PR gcov-profile/84137
22348 * doc/gcov.texi: Fix typo in documentation.
22350 2018-02-05 Martin Liska <mliska@suse.cz>
22352 PR gcov-profile/83879
22353 * doc/gcov.texi: Document necessity of --dynamic-list-data when
22354 using dlopen functionality.
22356 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
22358 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
22359 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
22360 _mm_maskz_range_ss, _mm_mask_range_round_ss,
22361 _mm_maskz_range_round_ss): New intrinsics.
22362 (__builtin_ia32_rangesd128_round)
22363 (__builtin_ia32_rangess128_round): Remove.
22364 (__builtin_ia32_rangesd128_mask_round,
22365 __builtin_ia32_rangess128_mask_round): New builtins.
22366 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
22367 __builtin_ia32_rangess128_round): Remove.
22368 (__builtin_ia32_rangesd128_mask_round,
22369 __builtin_ia32_rangess128_mask_round): New builtins.
22370 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
22371 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
22372 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
22373 "<round_saeonly_constraint>")): Changed to ...
22374 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
22375 "<round_saeonly_scalar_constraint>")): ... this.
22376 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
22377 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
22378 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
22379 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
22380 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
22382 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
22384 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
22386 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
22387 Remove all values except native, 8540 and 8548.
22389 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
22391 * config/i386/i386.c (ix86_output_function_return): Pass
22392 INVALID_REGNUM, instead of -1, as invalid register number to
22393 indirect_thunk_name and output_indirect_thunk.
22395 2018-02-02 Julia Koval <julia.koval@intel.com>
22397 * config.gcc: Add -march=icelake.
22398 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
22399 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
22400 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
22401 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
22402 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
22403 (processor_target_table): Add icelake.
22404 (ix86_option_override_internal): Handle new PTAs.
22405 (get_builtin_code_for_version): Handle icelake.
22406 (M_INTEL_COREI7_ICELAKE): New.
22407 (fold_builtin_cpu): Handle icelake.
22408 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
22409 * doc/invoke.texi: Add -march=icelake.
22411 2018-02-02 Julia Koval <julia.koval@intel.com>
22413 * config/i386/i386.c (ix86_option_override_internal): Change flags type
22414 to wide_int_bitmask.
22415 * wide-int-bitmask.h: New.
22417 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
22420 * config/i386/i386.md: Replace Pmode with word_mode in
22421 builtin_setjmp_setup and builtin_longjmp to support x32.
22423 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
22427 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
22429 (rs6000_supported_cpu_names): New static variable.
22430 (linux_cpu_translation_table): Likewise.
22431 (elf_platform) <cpu>: Define new static variable and use it.
22432 Translate kernel AT_PLATFORM name to canonical name if needed.
22433 Error if platform name is unknown.
22435 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
22438 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
22440 2018-02-01 Jeff Law <law@redhat.com>
22443 * config/i386/i386.c (release_scratch_register_on_entry): Add new
22444 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
22445 the scratch if RELEASE_VIA_POP is false.
22446 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
22447 If we have to save a temporary register, decrement SIZE appropriately.
22448 Pass new arguments to release_scratch_register_on_entry.
22449 (ix86_adjust_stack_and_probe): Likewise.
22450 (ix86_emit_probe_stack_range): Pass new arguments to
22451 release_scratch_register_on_entry.
22453 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
22455 PR rtl-optimization/84157
22456 * combine.c (change_zero_ext): Use REG_P predicate in
22457 front of HARD_REGISTER_P predicate.
22459 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
22461 * config/avr/avr.c (avr_option_override): Move disabling of
22462 -fdelete-null-pointer-checks to...
22463 * common/config/avr/avr-common.c (avr_option_optimization_table):
22466 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
22468 PR tree-optimization/81635
22469 * tree-data-ref.c (split_constant_offset_1): For types that
22470 wrap on overflow, try to use range info to prove that wrapping
22473 2018-02-01 Renlin Li <renlin.li@arm.com>
22476 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
22477 TAILCALL_ADDR_REGS.
22478 (aarch64_register_move_cost): Likewise.
22479 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
22480 TAILCALL_ADDR_REGS.
22481 (REG_CLASS_NAMES): Likewise.
22482 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
22483 TAILCALL_ADDR_REGS. Remove IP registers.
22484 * config/aarch64/aarch64.md (Ucs): Update register constraint.
22486 2018-02-01 Richard Biener <rguenther@suse.de>
22488 * domwalk.h (dom_walker::dom_walker): Add additional constructor
22489 for specifying RPO order and allow NULL for that.
22490 * domwalk.c (dom_walker::dom_walker): Likewise.
22491 (dom_walker::walk): Handle NULL RPO order.
22492 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
22494 (rewrite_update_dom_walker): Likewise.
22495 (mark_def_dom_walker): Likewise.
22497 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
22499 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
22500 (aarch64_maybe_expand_sve_subreg_move): Declare.
22501 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
22502 * config/aarch64/predicates.md (aarch64_any_register_operand): New
22504 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
22505 that are semantically a reverse operation.
22506 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
22507 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
22508 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
22510 (aarch64_can_change_mode_class): For big-endian, forbid changes
22511 between two SVE modes if they have different element sizes.
22513 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
22515 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
22516 the TImode handling for big-endian targets.
22518 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
22520 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
22521 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
22523 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
22524 Remove BSWAP handing for big-endian targets and use the form of
22525 LD1RQ appropariate for the mode.
22527 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
22529 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
22530 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
22531 duplicated element.
22533 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
22536 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
22537 check for operands that need to go through aarch64_sve_reload_be.
22539 2018-02-01 Jakub Jelinek <jakub@redhat.com>
22541 PR tree-optimization/81661
22542 PR tree-optimization/84117
22543 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
22544 * tree-eh.c: Include gimplify.h.
22545 (find_trapping_overflow, replace_trapping_overflow,
22546 rewrite_to_non_trapping_overflow): New functions.
22547 * tree-vect-loop.c: Include tree-eh.h.
22548 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
22549 * tree-data-ref.c: Include tree-eh.h.
22550 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
22552 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
22554 PR rtl-optimization/84123
22555 * combine.c (change_zero_ext): Check if hard register satisfies
22556 can_change_dest_mode before calling gen_lowpart_SUBREG.
22558 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
22561 * ira.c (ira_init_register_move_cost): Remove assert.
22563 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
22565 PR rtl-optimization/84071
22566 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
22567 * doc/tm.texi: Regenerate.
22569 2018-01-31 Richard Biener <rguenther@suse.de>
22571 PR tree-optimization/84132
22572 * tree-data-ref.c (analyze_miv_subscript): Properly
22573 check whether evolution_function_is_affine_multivariate_p
22574 before calling gcd_of_steps_may_divide_p.
22576 2018-01-31 Julia Koval <julia.koval@intel.com>
22579 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
22580 * config/i386/i386.md (rdpid_rex64) New.
22581 (rdpid): Make 32bit only.
22583 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
22586 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
22587 an IDENTIFIER_NODE for FUNCTION_TYPE's.
22589 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
22592 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
22594 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
22596 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
22598 PR rtl-optimization/84071
22599 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
22600 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
22602 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
22604 * config/arc/arc.c (arc_handle_aux_attribute): New function.
22605 (arc_attribute_table): Add 'aux' attribute.
22606 (arc_in_small_data_p): Consider aux like variables.
22607 (arc_is_aux_reg_p): New function.
22608 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
22609 (arc_get_aux_arg): New function.
22610 (prepare_move_operands): Handle aux-register access.
22611 (arc_handle_aux_attribute): New function.
22612 * doc/extend.texi (ARC Variable attributes): Add subsection.
22614 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
22616 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
22617 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
22618 (arc_attribute_table): Add 'uncached' attribute.
22619 (arc_print_operand): Print '.di' flag for uncached memory
22621 (arc_in_small_data_p): Do not consider for small data the uncached
22623 (arc_is_uncached_mem_p): New function.
22624 * config/arc/predicates.md (compact_store_memory_operand): Check
22625 for uncached memory accesses.
22626 (nonvol_nonimm_operand): Likewise.
22627 * doc/extend.texi (ARC Type Attribute): New subsection.
22629 2018-01-31 Jakub Jelinek <jakub@redhat.com>
22632 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
22633 falign-loops=): Add Optimization flag.
22635 2018-01-30 Jeff Law <law@redhat.com>
22638 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
22639 INT_REGISTERS_SAVED. Check it prior to calling
22640 get_scratch_register_on_entry.
22641 (ix86_adjust_stack_and_probe): Similarly.
22642 (ix86_emit_probe_stack_range): Similarly.
22643 (ix86_expand_prologue): Corresponding changes.
22645 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22648 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
22649 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
22651 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
22654 * lra-constraints.c (curr_insn_transform): Process AND in the
22657 2018-01-30 Jakub Jelinek <jakub@redhat.com>
22659 PR rtl-optimization/83986
22660 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
22661 dependence against last_pending_memory_flush in addition to
22662 pending_jump_insns.
22664 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
22666 PR tree-optimization/81611
22667 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
22670 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22673 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
22676 2018-01-30 Richard Biener <rguenther@suse.de>
22677 Jakub Jelinek <jakub@redhat.com>
22679 PR tree-optimization/84111
22680 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
22681 inner loops added during recursion, as they don't have up-to-date
22684 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
22687 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
22688 (can_inline_edge_by_limits_p): ... here.
22689 (can_early_inline_edge_p, check_callers,
22690 update_caller_keys, update_callee_keys, recursive_inlining,
22691 add_new_edges_to_heap, speculation_useful_p,
22692 inline_small_functions,
22693 inline_small_functions, flatten_function,
22694 inline_to_all_callers_1): Update.
22696 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
22698 * profile-count.c (profile_count::combine_with_ipa_count): Handle
22701 2018-01-30 Richard Biener <rguenther@suse.de>
22703 PR tree-optimization/83008
22704 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
22705 invariant and constant vector uses in stmts when they need
22706 more than one stmt.
22708 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22711 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
22712 * configure: Regenerate.
22714 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
22716 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
22718 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
22719 Use gen_rtx_REG rather than gen_lowpart.
22721 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
22723 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
22724 rather than 0 when creating partial subregs.
22726 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
22728 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
22731 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
22734 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
22735 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
22736 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
22737 flags. This restores the settings used before the 2017-07-24.
22738 Turning off pre increment/decrement/modify allows IVOPTS to
22739 optimize DF/SF loops where the index is an int.
22741 2018-01-29 Richard Biener <rguenther@suse.de>
22742 Kelvin Nilsen <kelvin@gcc.gnu.org>
22745 * tree-vect-stmts.c (vectorizable_call): Don't call
22746 targetm.vectorize_builtin_md_vectorized_function if callee is
22749 2018-01-22 Carl Love <cel@us.ibm.com>
22751 * doc/extend.tex: Fix typo in second arg in
22752 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
22754 2018-01-29 Richard Biener <rguenther@suse.de>
22756 PR tree-optimization/84086
22757 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
22758 (flush_ssaname_freelist): When SSA names were released reset
22759 the SCEV hash table.
22761 2018-01-29 Richard Biener <rguenther@suse.de>
22763 PR tree-optimization/84057
22764 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
22765 removed paths when removing edges.
22767 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
22769 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
22770 -mfunction-return=@var{choice}.
22772 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
22774 PR diagnostic/84034
22775 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
22776 Handle CR like TAB.
22777 (layout::print_source_line): Likewise.
22778 (test_get_line_width_without_trailing_whitespace): Add test cases.
22780 2018-01-27 Jakub Jelinek <jakub@redhat.com>
22782 PR middle-end/84040
22783 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
22786 2018-01-26 Jim Wilson <jimw@sifive.com>
22788 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
22790 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
22793 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22795 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
22796 and CMP + SUB-immediate -> SUBS.
22798 2018-01-26 Martin Sebor <msebor@redhat.com>
22800 PR tree-optimization/83896
22801 * tree-ssa-strlen.c (get_string_len): Rename...
22802 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
22803 Avoid assuming length is constant.
22804 (handle_char_store): Use HOST_WIDE_INT for string length.
22806 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
22809 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
22810 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
22812 2018-01-26 Richard Biener <rguenther@suse.de>
22814 PR rtl-optimization/84003
22815 * dse.c (record_store): Only record redundant stores when
22816 the earlier store aliases at least all accesses the later one does.
22818 2018-01-26 Jakub Jelinek <jakub@redhat.com>
22820 PR rtl-optimization/83985
22821 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
22822 REG_CFA_RESTORE insns.
22823 (delete_unmarked_insns): Don't ignore separate shrink wrapping
22824 REG_CFA_RESTORE insns here.
22827 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
22828 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
22830 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
22832 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
22833 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
22834 (arc_init): Likewise.
22835 (arc_override_options): Likewise.
22836 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
22838 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
22840 * config/arc/arc.h (TARGET_DBNZ): Define.
22841 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
22842 properly set the tune attribute.
22843 (dbnz): Use TARGET_DBNZ guard.
22844 * config/arc/arc.opt (mtune): Add core3 option.
22846 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
22848 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
22849 recognize new pic like addresses.
22850 (arc_delegitimize_address): Clean up.
22852 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
22854 * config/arc/arc-arches.def: Option mrf16 valid for all
22856 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
22857 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
22858 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
22859 * config/arc/arc-tables.opt: Regenerate.
22860 * config/arc/arc.c (arc_conditional_register_usage): Handle
22861 reduced register file case.
22862 (arc_file_start): Set must have build attributes.
22863 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
22864 mrf16 option value.
22865 * config/arc/arc.opt (mrf16): Add new option.
22866 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
22867 * config/arc/genmultilib.awk: Handle new mrf16 option.
22868 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
22869 * config/arc/t-multilib: Regenerate.
22870 * doc/invoke.texi (ARC Options): Document mrf16 option.
22872 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
22874 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
22875 * config/arc/arc.c (arc_handle_secure_attribute): New function.
22876 (arc_attribute_table): Add 'secure_call' attribute.
22877 (arc_print_operand): Print secure call operand.
22878 (arc_function_ok_for_sibcall): Don't optimize tail calls when
22880 (arc_is_secure_call_p): New function. * config/arc/arc.md
22881 (call_i): Add support for sjli instruction.
22882 (call_value_i): Likewise.
22883 * config/arc/constraints.md (Csc): New constraint.
22885 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
22886 John Eric Martin <John.Martin@emmicro-us.com>
22888 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
22889 * config/arc/arc.c (_arc_jli_section): New struct.
22890 (arc_jli_section): New type.
22891 (rc_jli_sections): New static variable.
22892 (arc_handle_jli_attribute): New function.
22893 (arc_attribute_table): Add jli_always and jli_fixed attribute.
22894 (arc_file_end): New function.
22895 (TARGET_ASM_FILE_END): Define.
22896 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
22897 (arc_add_jli_section): New function.
22898 (jli_call_scan): Likewise.
22899 (arc_reorg): Call jli_call_scan.
22900 (arc_output_addsi): Remove 'S' from printing asm operand.
22901 (arc_is_jli_call_p): New function.
22902 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
22904 (movhi_insn): Likewise.
22905 (movsi_insn): Likewise.
22906 (movsi_set_cc_insn): Likewise.
22907 (loadqi_update): Likewise.
22908 (load_zeroextendqisi_update): Likewise.
22909 (load_signextendqisi_update): Likewise.
22910 (loadhi_update): Likewise.
22911 (load_zeroextendhisi_update): Likewise.
22912 (load_signextendhisi_update): Likewise.
22913 (loadsi_update): Likewise.
22914 (loadsf_update): Likewise.
22915 (movsicc_insn): Likewise.
22916 (bset_insn): Likewise.
22917 (bxor_insn): Likewise.
22918 (bclr_insn): Likewise.
22919 (bmsk_insn): Likewise.
22920 (bicsi3_insn): Likewise.
22921 (cmpsi_cc_c_insn): Likewise.
22922 (movsi_ne): Likewise.
22923 (movsi_cond_exec): Likewise.
22924 (clrsbsi2): Likewise.
22925 (norm_f): Likewise.
22935 (call_i): Remove 'S' asm letter, add jli instruction.
22936 (call_value_i): Likewise.
22937 * config/arc/arc.op (mjli-always): New option.
22938 * config/arc/constraints.md (Cji): New constraint.
22939 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
22941 (subsf3_fpx): Likewise.
22942 (mulsf3_fpx): Likewise.
22943 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
22945 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
22946 function attrbutes.
22947 * doc/invoke.texi (ARC): Document mjli-always option.
22949 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
22951 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
22952 avoid addition with 0 and use incw and decw where possible.
22954 2018-01-26 Richard Biener <rguenther@suse.de>
22956 PR tree-optimization/81082
22957 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
22958 association if it requires casting to unsigned.
22959 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
22960 from fold_plusminus_mult_expr to catch important cases late when
22961 range info is available.
22963 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22965 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
22966 * configure.ac (hidden_linkonce): New test.
22967 * configure: Regenerate.
22968 * config.in: Regenerate.
22970 2018-01-26 Julia Koval <julia.koval@intel.com>
22972 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
22973 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
22974 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
22975 _mm_mask_bitshuffle_epi64_mask): Fix type.
22976 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
22977 USI_FTYPE_V4DI_V4DI_USI): Remove.
22978 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
22979 __builtin_ia32_vpshufbitqmb256_mask,
22980 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
22981 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
22982 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
22984 2018-01-26 Alan Modra <amodra@gmail.com>
22987 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
22988 UNSPEC_VBPERMQ. Sort other unspecs.
22990 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
22992 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
22994 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
22996 PR middle-end/83055
22997 * predict.c (drop_profile): Do not push/pop cfun; update also
22999 (handle_missing_profiles): Fix logic looking for zero profiles.
23001 2018-01-25 Jakub Jelinek <jakub@redhat.com>
23003 PR middle-end/83977
23004 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
23005 on functions with #pragma omp declare simd or functions with simd
23007 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
23008 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
23009 Remove trailing \n from warning_at calls.
23011 2018-01-25 Tom de Vries <tom@codesourcery.com>
23014 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
23015 for neutered workers.
23017 2018-01-24 Joseph Myers <joseph@codesourcery.com>
23020 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
23021 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
23023 2018-01-24 Jeff Law <law@redhat.com>
23026 * i386.c (get_probe_interval): Move to earlier point.
23027 (ix86_compute_frame_layout): If -fstack-clash-protection and
23028 the frame is larger than the probe interval, then use pushes
23029 to save registers rather than reg->mem moves.
23030 (ix86_expand_prologue): Remove conditional for int_registers_saved
23033 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
23036 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
23037 min/max for never referenced object.
23039 2018-01-24 Jakub Jelinek <jakub@redhat.com>
23041 PR middle-end/83977
23042 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
23044 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
23045 attributes from DECL_ATTRIBUTES (decl) without affecting
23046 DECL_ATTRIBUTES (current_function_decl).
23047 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
23048 functions with non-NULL DECL_ABSTRACT_ORIGIN.
23050 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
23052 PR tree-optimization/83979
23053 * fold-const.c (fold_comparison): Use constant_boolean_node
23054 instead of boolean_{true,false}_node.
23056 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
23058 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
23061 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23063 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
23064 Simplify the clause that sets the length attribute.
23065 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
23066 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
23067 clause that sets the length attribute.
23068 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
23070 2018-01-24 Tom de Vries <tom@codesourcery.com>
23073 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
23074 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
23075 Add strict parameter.
23076 (prevent_branch_around_nothing): Insert dummy insn between branch to
23077 label and label with no ptx insn inbetween.
23078 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
23080 2018-01-24 Tom de Vries <tom@codesourcery.com>
23083 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
23084 for neutered threads in warp.
23085 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
23087 2018-01-24 Richard Biener <rguenther@suse.de>
23089 PR tree-optimization/83176
23090 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
23093 2018-01-24 Richard Biener <rguenther@suse.de>
23095 PR tree-optimization/82819
23096 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
23097 code generating pluses that are no-ops in the target precision.
23099 2018-01-24 Richard Biener <rguenther@suse.de>
23101 PR middle-end/84000
23102 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
23104 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
23106 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
23107 to merge probabilities.
23108 * predict.c (probably_never_executed): Also mark as cold functions
23109 with global 0 profile and guessed local profile.
23110 * profile-count.c (profile_probability::combine_with_count): New
23112 * profile-count.h (profile_probability::operator*,
23113 profile_probability::operator*=, profile_probability::operator/,
23114 profile_probability::operator/=): Reduce precision to adjusted
23115 and set value to guessed on contradictory divisions.
23116 (profile_probability::combine_with_freq): Remove.
23117 (profile_probability::combine_wiht_count): Declare.
23118 (profile_count::force_nonzero):: Set to adjusted.
23119 (profile_count::probability_in):: Set quality to adjusted.
23120 * tree-ssa-tail-merge.c (replace_block_by): Use
23121 combine_with_count.
23123 2018-01-23 Andrew Waterman <andrew@sifive.com>
23124 Jim Wilson <jimw@sifive.com>
23126 * config/riscv/riscv.c (riscv_stack_boundary): New.
23127 (riscv_option_override): Set riscv_stack_boundary. Handle
23128 riscv_preferred_stack_boundary_arg.
23129 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
23130 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
23131 (STACK_BOUNDARY): Set to riscv_stack_boundary.
23132 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
23133 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
23134 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
23136 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
23139 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
23140 of struct ix86_frame.
23141 (ix86_expand_epilogue): Likewise. Add a local variable for
23142 the reg_save_offset field in struct ix86_frame.
23144 2018-01-23 Bin Cheng <bin.cheng@arm.com>
23146 PR tree-optimization/82604
23147 * tree-loop-distribution.c (enum partition_kind): New enum item
23148 PKIND_PARTIAL_MEMSET.
23149 (partition_builtin_p): Support above new enum item.
23150 (generate_code_for_partition): Ditto.
23151 (compute_access_range): Differentiate cases that equality can be
23152 proven at all loops, the innermost loops or no loops.
23153 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
23154 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
23155 (finalize_partitions, distribute_loop): Don't fuse partition of
23156 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
23157 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
23158 parloop is enabled.
23160 2018-01-23 Martin Liska <mliska@suse.cz>
23162 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
23163 order to ignore the predictor.
23164 (PRED_POLYMORPHIC_CALL): Likewise.
23165 (PRED_RECURSIVE_CALL): Likewise.
23167 2018-01-23 Martin Liska <mliska@suse.cz>
23169 * tree-profile.c (tree_profiling): Print function header to
23170 aware reader which function we are working on.
23171 * value-prof.c (gimple_find_values_to_profile): Do not print
23172 not interesting value histograms.
23174 2018-01-23 Martin Liska <mliska@suse.cz>
23176 * profile-count.h (enum profile_quality): Add
23177 profile_uninitialized as the first value. Do not number values
23178 as they are zero based.
23179 (profile_count::verify): Update sanity check.
23180 (profile_probability::verify): Likewise.
23182 2018-01-23 Nathan Sidwell <nathan@acm.org>
23184 * doc/invoke.texi (ffor-scope): Deprecate.
23186 2018-01-23 David Malcolm <dmalcolm@redhat.com>
23188 PR tree-optimization/83510
23189 * domwalk.c (set_all_edges_as_executable): New function.
23190 (dom_walker::dom_walker): Convert bool param
23191 "skip_unreachable_blocks" to enum reachability. Move setup of
23192 edge flags to set_all_edges_as_executable and only do it when
23193 reachability is REACHABLE_BLOCKS.
23194 * domwalk.h (enum dom_walker::reachability): New enum.
23195 (dom_walker::dom_walker): Convert bool param
23196 "skip_unreachable_blocks" to enum reachability.
23197 (set_all_edges_as_executable): New decl.
23198 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
23199 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
23201 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
23202 but converting true to REACHABLE_BLOCKS.
23203 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
23205 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
23206 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
23207 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
23209 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
23210 if check_all_array_refs will be called.
23212 2018-01-23 David Malcolm <dmalcolm@redhat.com>
23214 * tree.c (selftest::test_location_wrappers): Add more test
23217 2018-01-23 David Malcolm <dmalcolm@redhat.com>
23219 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
23220 (selftest::test_bit_in_range): Likewise.
23222 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
23225 * doc/sourcebuild.texi (vect_float): Say that the selector
23226 only describes the situation when -funsafe-math-optimizations is on.
23227 (vect_float_strict): Document.
23229 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
23231 PR tree-optimization/83965
23232 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
23233 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
23234 instead of checking only for a reduction.
23235 (vect_recog_widen_sum_pattern): Likewise.
23237 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
23239 * predict.c (probably_never_executed): Only use precise profile info.
23240 (compute_function_frequency): Skip after inlining hack since we now
23241 have quality checking.
23243 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
23245 * profile-count.h (profile_probability::very_unlikely,
23246 profile_probability::unlikely, profile_probability::even): Set
23247 precision to guessed.
23249 2018-01-23 Richard Biener <rguenther@suse.de>
23251 PR tree-optimization/83963
23252 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
23253 Properly terminate dominator walk when crossing the exit edge not
23254 when visiting its source block.
23256 2018-01-23 Jakub Jelinek <jakub@redhat.com>
23259 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
23260 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
23262 2018-01-22 Jakub Jelinek <jakub@redhat.com>
23264 PR tree-optimization/83957
23265 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
23266 semicolon after for body surrounded by braces.
23268 PR tree-optimization/83081
23269 * profile-count.h (profile_probability::split): New method.
23270 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
23271 Use profile_probability::split.
23272 (do_compare_rtx_and_jump): Fix adjustment of probabilities
23273 when splitting a single conditional jump into 2.
23275 2018-01-22 David Malcolm <dmalcolm@redhat.com>
23277 PR tree-optimization/69452
23278 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
23281 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
23283 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
23284 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
23285 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
23287 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
23289 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
23291 * config/rl78/rl78.md (movdi): New define_expand.
23292 * config/rl78/rl78.c (rl78_split_movdi): New function.
23294 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
23297 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
23299 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
23300 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
23301 128-bit to produce an UNSPEC move to get the double word with the
23302 signbit and then a shift directly to do signbit.
23303 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
23304 implementation with a new version that just does either a direct
23305 move or a regular move. Move memory interface to separate insns.
23306 Move insns so they are next to the expander.
23307 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
23308 with signbit move. Split big and little endian case.
23309 (signbit<mode>2_dm_mem_le): Likewise.
23310 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
23311 (signbit<mode>2_dm2): Likewise.
23313 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
23315 * config/rl78/rl78.md (anddi3): New define_expand.
23317 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
23319 * config/rl78/rl78.md (umindi3): New define_expand.
23321 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
23323 * config/rl78/rl78.md (smindi3): New define_expand.
23325 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
23327 * config/rl78/rl78.md (smaxdi3): New define_expand.
23329 2018-01-22 Carl Love <cel@us.ibm.com>
23331 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
23332 LVX_V1TI): Add macro expansion.
23333 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
23334 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
23335 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
23336 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
23337 Change check to determine if the instruction is a byte reversing
23338 entry. Fix typo in comment.
23339 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
23340 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
23341 Add def_builtin calls for new builtins.
23342 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
23343 Add define_insn expansion.
23345 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
23347 * config/rl78/rl78.md (umaxdi3): New define_expand.
23349 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
23351 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
23352 for non-QImode registers.
23354 2018-01-22 Richard Biener <rguenther@suse.de>
23356 PR tree-optimization/83963
23357 * graphite-scop-detection.c (scop_detection::get_sese): Delay
23358 including the loop exit block.
23359 (scop_detection::merge_sese): Likewise.
23360 (scop_detection::add_scop): Do it here instead.
23362 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23364 * doc/sourcebuild.texi (arm_softfloat): Document.
23366 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
23369 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
23370 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
23371 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
23373 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23374 David Edelsohn <dje.gcc@gmail.com>
23377 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
23378 Change "crset eq" to "crset 2".
23379 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
23380 (*call_indirect_aix<mode>_nospec): Likewise.
23381 (*call_value_indirect_aix<mode>_nospec): Likewise.
23382 (*call_indirect_elfv2<mode>_nospec): Likewise.
23383 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
23384 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
23385 change assembly output from . to $.
23386 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
23387 (indirect_jump<mode>_nospec): Change assembly output from . to $.
23388 (*tablejump<mode>_internal1_nospec): Likewise.
23390 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
23393 * config/sh/sh_optimize_sett_clrt.cc:
23394 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
23396 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
23398 PR tree-optimization/83940
23399 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
23400 offset_dt to vect_constant_def rather than vect_unknown_def_type.
23401 (vect_check_load_store_mask): Add a mask_dt_out parameter and
23402 use it to pass back the definition type.
23403 (vect_check_store_rhs): Likewise rhs_dt_out.
23404 (vect_build_gather_load_calls): Add a mask_dt argument and use
23405 it instead of a call to vect_is_simple_use.
23406 (vectorizable_store): Update calls to vect_check_load_store_mask
23407 and vect_check_store_rhs. Use the dt returned by the latter instead
23408 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
23409 instead of calls to vect_is_simple_use. Pass the scalar rather
23410 than the vector operand to vect_is_simple_use when handling
23411 second and subsequent copies of an rhs value.
23412 (vectorizable_load): Update calls to vect_check_load_store_mask
23413 and vect_build_gather_load_calls. Use the cached mask_dt and
23414 gs_info.offset_dt instead of calls to vect_is_simple_use.
23416 2018-01-20 Jakub Jelinek <jakub@redhat.com>
23418 PR middle-end/83945
23419 * tree-emutls.c: Include gimplify.h.
23420 (lower_emutls_2): New function.
23421 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
23422 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
23423 it before further processing.
23426 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
23427 UINTVAL (trueop1) instead of INTVAL (op1).
23429 2018-01-19 Jakub Jelinek <jakub@redhat.com>
23433 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
23434 INCOMING_FRAME_SP_OFFSET if not defined.
23435 (scan_trace): Add ENTRY argument. If true and
23436 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
23437 emit a note to adjust the CFA offset.
23438 (create_cfi_notes): Adjust scan_trace callers.
23439 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
23440 INCOMING_FRAME_SP_OFFSET in the CIE.
23441 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
23442 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
23444 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
23445 * doc/tm.texi: Regenerated.
23447 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23449 PR rtl-optimization/83147
23450 * lra-constraints.c (remove_inheritance_pseudos): Use
23451 lra_substitute_pseudo_within_insn.
23453 2018-01-19 Tom de Vries <tom@codesourcery.com>
23454 Cesar Philippidis <cesar@codesourcery.com>
23457 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
23459 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
23462 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
23463 spaces for function labels.
23465 2018-01-19 Martin Liska <mliska@suse.cz>
23467 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
23468 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
23469 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
23470 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
23471 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
23472 (PRED_CONST_RETURN): Change from 69 to 65.
23473 (PRED_NULL_RETURN): Change from 91 to 71.
23474 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
23475 (PRED_LOOP_GUARD): Change from 66 to 73.
23477 2018-01-19 Martin Liska <mliska@suse.cz>
23479 * predict.c (predict_insn_def): Add new assert.
23480 (struct branch_predictor): Change type to signed integer.
23481 (test_prediction_value_range): Amend test to cover
23482 PROB_UNINITIALIZED.
23483 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
23484 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
23485 (PRED_LOOP_ITERATIONS_MAX): Likewise.
23486 (PRED_LOOP_IV_COMPARE): Likewise.
23487 * predict.h (PROB_UNINITIALIZED): Define new constant.
23489 2018-01-19 Martin Liska <mliska@suse.cz>
23491 * predict.c (dump_prediction): Add new format for
23492 analyze_brprob.py script which is enabled with -details
23494 * profile-count.h (precise_p): New function.
23496 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
23498 PR tree-optimization/83922
23499 * tree-vect-loop.c (vect_verify_full_masking): Return false if
23500 there are no statements that need masking.
23501 (vect_active_double_reduction_p): New function.
23502 (vect_analyze_loop_operations): Use it when handling phis that
23503 are not in the loop header.
23505 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
23507 PR tree-optimization/83914
23508 * tree-vect-loop.c (vectorizable_induction): Don't convert
23509 init_expr or apply the peeling adjustment for inductions
23510 that are nested within the vectorized loop.
23512 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23514 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
23517 2018-01-18 Jakub Jelinek <jakub@redhat.com>
23521 * function.h (gimplify_parameters): Add gimple_seq * argument.
23522 * function.c: Include gimple.h and options.h.
23523 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
23524 for the added local temporaries if needed.
23525 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
23526 if there are any parameter cleanups, wrap whole body into a
23527 try/finally with the cleanups.
23529 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
23532 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
23533 Use GET_MODE_CLASS for scalar floating point.
23535 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
23539 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
23540 Fix call of call_cgraph_insertion_hooks.
23542 2018-01-18 Martin Sebor <msebor@redhat.com>
23544 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
23546 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
23549 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
23552 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
23555 * common.opt: (-ffile-prefix-map): New option.
23556 * opts.c (common_handle_option): Defer it.
23557 * opts-global.c (handle_common_deferred_options): Handle it.
23558 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
23559 * file-prefix-map.h: New file.
23560 (remap_debug_filename, add_debug_prefix_map): ...here.
23561 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
23562 * final.c (debug_prefix_map, add_debug_prefix_map
23563 remap_debug_filename): Move to...
23564 * file-prefix-map.c: New file.
23565 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
23566 generalize, get rid of alloca(), use strrchr() instead of strchr().
23567 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
23568 Implement in terms of add_prefix_map().
23569 (remap_macro_filename, remap_debug_filename): Implement in term of
23571 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
23572 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
23573 * dbxout.c: Include file-prefix-map.h.
23574 * varasm.c: Likewise.
23575 * vmsdbgout.c: Likewise.
23576 * xcoffout.c: Likewise.
23577 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
23578 * doc/cppopts.texi (-fmacro-prefix-map): Document.
23579 * doc/invoke.texi (-ffile-prefix-map): Document.
23580 (-fdebug-prefix-map): Update description.
23582 2018-01-18 Martin Liska <mliska@suse.cz>
23584 * config/i386/i386.c (indirect_thunk_name): Document that also
23586 (output_indirect_thunk): Document why both instructions
23587 (pause and lfence) are generated.
23589 2018-01-18 Richard Biener <rguenther@suse.de>
23591 PR tree-optimization/83887
23592 * graphite-scop-detection.c
23593 (scop_detection::get_nearest_dom_with_single_entry): Remove.
23594 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
23595 (scop_detection::merge_sese): Re-implement with a flood-fill
23596 algorithm that properly finds a SESE region if it exists.
23598 2018-01-18 Jakub Jelinek <jakub@redhat.com>
23601 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
23602 pointer_diff optimizations use view_convert instead of convert.
23604 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23606 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
23607 Generate different code for -mno-speculate-indirect-jumps.
23608 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
23609 (*call_indirect_aix<mode>): Disable for
23610 -mno-speculate-indirect-jumps.
23611 (*call_indirect_aix<mode>_nospec): New define_insn.
23612 (*call_value_indirect_aix<mode>): Disable for
23613 -mno-speculate-indirect-jumps.
23614 (*call_value_indirect_aix<mode>_nospec): New define_insn.
23615 (*sibcall_nonlocal_sysv<mode>): Generate different code for
23616 -mno-speculate-indirect-jumps.
23617 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
23619 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
23621 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
23622 long double type, set the flags for noting the default long double
23623 type, even if we don't pass or return a long double type.
23625 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
23628 * ipa-inline.c (flatten_function): Do not overwrite final inlining
23631 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
23633 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
23634 support for merge[hl].
23635 (fold_mergehl_helper): New helper function.
23636 (tree-vector-builder.h): New #include for tree_vector_builder usage.
23637 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
23638 (altivec_vmrglw_direct): Add xxmrglw insn.
23640 2018-01-17 Andrew Waterman <andrew@sifive.com>
23642 * config/riscv/riscv.c (riscv_conditional_register_usage): If
23643 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
23645 2018-01-17 David Malcolm <dmalcolm@redhat.com>
23648 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
23649 call the lto_location_cache before reading the
23650 DECL_SOURCE_LOCATION of the types.
23652 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
23653 Richard Sandiford <richard.sandiford@linaro.org>
23655 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
23656 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
23657 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
23658 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
23659 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
23661 * config/aarch64/constraints.md (aarch64_movti_operand):
23663 * config/aarch64/predicates.md (Uti): Add new constraint.
23665 2018-01-17 Carl Love <cel@us.ibm.com>
23667 * config/rs6000/vsx.md (define_expand xl_len_r,
23668 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
23669 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
23671 (define_expand, define_insn): Move the shift left from the
23672 define_insn to the define_expand for lxvl and stxvl instructions.
23673 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
23674 and XL_LEN_R definitions to PURE.
23676 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
23678 * config/i386/i386.c (indirect_thunk_name): Declare regno
23679 as unsigned int. Compare regno with INVALID_REGNUM.
23680 (output_indirect_thunk): Ditto.
23681 (output_indirect_thunk_function): Ditto.
23682 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
23683 in the call to output_indirect_thunk_function.
23685 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
23687 PR middle-end/83884
23688 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
23689 rather than the size of inner_type to determine the stack slot size
23690 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
23692 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
23695 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
23698 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
23700 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
23701 endian Linux systems to optionally enable multilibs for selecting
23702 the long double type if the user configured an explicit type.
23703 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
23704 have no long double multilibs if not defined.
23705 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
23706 warn if the user used -mabi={ieee,ibm}longdouble and we built
23707 multilibs for long double.
23708 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
23709 appropriate multilib option.
23710 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
23712 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
23713 for building long double multilibs.
23714 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
23716 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
23718 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
23721 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
23723 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
23726 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
23729 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
23732 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
23734 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
23735 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
23737 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
23739 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
23740 different rtl trees depending on TARGET_64BIT.
23741 (rs6000_gen_lvx): Likewise.
23743 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
23745 * config/visium/visium.md (nop): Tweak comment.
23746 (hazard_nop): Likewise.
23748 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23750 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
23751 -mspeculate-indirect-jumps.
23752 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
23753 for -mno-speculate-indirect-jumps.
23754 (*call_indirect_elfv2<mode>_nospec): New define_insn.
23755 (*call_value_indirect_elfv2<mode>): Disable for
23756 -mno-speculate-indirect-jumps.
23757 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
23758 (indirect_jump): Emit different RTL for
23759 -mno-speculate-indirect-jumps.
23760 (*indirect_jump<mode>): Disable for
23761 -mno-speculate-indirect-jumps.
23762 (*indirect_jump<mode>_nospec): New define_insn.
23763 (tablejump): Emit different RTL for
23764 -mno-speculate-indirect-jumps.
23765 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
23766 (tablejumpsi_nospec): New define_expand.
23767 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
23768 (tablejumpdi_nospec): New define_expand.
23769 (*tablejump<mode>_internal1): Disable for
23770 -mno-speculate-indirect-jumps.
23771 (*tablejump<mode>_internal1_nospec): New define_insn.
23772 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
23775 2018-01-16 Artyom Skrobov tyomitch@gmail.com
23777 * caller-save.c (insert_save): Drop unnecessary parameter. All
23780 2018-01-16 Jakub Jelinek <jakub@redhat.com>
23781 Richard Biener <rguenth@suse.de>
23784 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
23785 return early, inline manually is_gimple_sizepos. Make sure if we
23786 call gimplify_expr we don't end up with a gimple constant.
23787 * tree.c (variably_modified_type_p): Don't return true for
23788 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
23789 * gimplify.h (is_gimple_sizepos): Remove.
23791 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
23793 PR tree-optimization/83857
23794 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
23795 vectorizable_live_operation for pure SLP statements.
23796 (vectorizable_live_operation): Handle PHIs.
23798 2018-01-16 Richard Biener <rguenther@suse.de>
23800 PR tree-optimization/83867
23801 * tree-vect-stmts.c (vect_transform_stmt): Precompute
23802 nested_in_vect_loop_p since the scalar stmt may get invalidated.
23804 2018-01-16 Jakub Jelinek <jakub@redhat.com>
23807 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
23808 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
23809 If off is not INTEGER_CST, issue a may not be aligned warning
23810 rather than isn't aligned. Use isn%'t rather than isn't.
23811 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
23813 <case MULT_EXPR>: Improve the case when bottom and one of the
23814 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
23815 operand, in that case check if the other operand is multiple of
23816 bottom divided by the INTEGER_CST operand.
23818 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
23821 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
23822 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
23823 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
23824 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
23825 * config/pa/pa.c (pa_function_arg_advance): Likewise.
23826 (pa_function_arg, pa_arg_partial_bytes): Likewise.
23827 (pa_function_arg_size): New function.
23829 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
23831 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
23832 in a separate statement.
23834 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
23836 PR tree-optimization/83847
23837 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
23838 group gathers and scatters.
23840 2018-01-16 Jakub Jelinek <jakub@redhat.com>
23842 PR rtl-optimization/86620
23843 * params.def (max-sched-ready-insns): Bump minimum value to 1.
23845 PR rtl-optimization/83213
23846 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
23847 to last if both are JUMP_INSNs.
23849 PR tree-optimization/83843
23850 * gimple-ssa-store-merging.c
23851 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
23852 store_immediate_info for bswap/nop orig_stores.
23854 2018-01-15 Andrew Waterman <andrew@sifive.com>
23856 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
23858 <UDIV>: Increase cost if !TARGET_DIV.
23860 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
23862 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
23863 (define_attr "cr_logical_3op"): New.
23864 (cceq_ior_compare): Adjust.
23865 (cceq_ior_compare_complement): Adjust.
23866 (*cceq_rev_compare): Adjust.
23867 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
23868 (is_cracked_insn): Adjust.
23869 (insn_must_be_first_in_group): Adjust.
23870 * config/rs6000/40x.md: Adjust.
23871 * config/rs6000/440.md: Adjust.
23872 * config/rs6000/476.md: Adjust.
23873 * config/rs6000/601.md: Adjust.
23874 * config/rs6000/603.md: Adjust.
23875 * config/rs6000/6xx.md: Adjust.
23876 * config/rs6000/7450.md: Adjust.
23877 * config/rs6000/7xx.md: Adjust.
23878 * config/rs6000/8540.md: Adjust.
23879 * config/rs6000/cell.md: Adjust.
23880 * config/rs6000/e300c2c3.md: Adjust.
23881 * config/rs6000/e500mc.md: Adjust.
23882 * config/rs6000/e500mc64.md: Adjust.
23883 * config/rs6000/e5500.md: Adjust.
23884 * config/rs6000/e6500.md: Adjust.
23885 * config/rs6000/mpc.md: Adjust.
23886 * config/rs6000/power4.md: Adjust.
23887 * config/rs6000/power5.md: Adjust.
23888 * config/rs6000/power6.md: Adjust.
23889 * config/rs6000/power7.md: Adjust.
23890 * config/rs6000/power8.md: Adjust.
23891 * config/rs6000/power9.md: Adjust.
23892 * config/rs6000/rs64.md: Adjust.
23893 * config/rs6000/titan.md: Adjust.
23895 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
23897 * config/i386/predicates.md (indirect_branch_operand): Rewrite
23898 ix86_indirect_branch_register logic.
23900 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
23902 * config/i386/constraints.md (Bs): Update
23903 ix86_indirect_branch_register check. Don't check
23904 ix86_indirect_branch_register with GOT_memory_operand.
23906 * config/i386/predicates.md (GOT_memory_operand): Don't check
23907 ix86_indirect_branch_register here.
23908 (GOT32_symbol_operand): Likewise.
23910 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
23912 * config/i386/predicates.md (constant_call_address_operand):
23913 Rewrite ix86_indirect_branch_register logic.
23914 (sibcall_insn_operand): Likewise.
23916 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
23918 * config/i386/constraints.md (Bs): Replace
23919 ix86_indirect_branch_thunk_register with
23920 ix86_indirect_branch_register.
23922 * config/i386/i386.md (indirect_jump): Likewise.
23923 (tablejump): Likewise.
23924 (*sibcall_memory): Likewise.
23925 (*sibcall_value_memory): Likewise.
23926 Peepholes of indirect call and jump via memory: Likewise.
23927 * config/i386/i386.opt: Likewise.
23928 * config/i386/predicates.md (indirect_branch_operand): Likewise.
23929 (GOT_memory_operand): Likewise.
23930 (call_insn_operand): Likewise.
23931 (sibcall_insn_operand): Likewise.
23932 (GOT32_symbol_operand): Likewise.
23934 2018-01-15 Jakub Jelinek <jakub@redhat.com>
23936 PR middle-end/83837
23937 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
23938 type rather than type addr's type points to.
23939 (expand_omp_atomic_mutex): Likewise.
23940 (expand_omp_atomic): Likewise.
23942 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
23945 * config/i386/i386.c (output_indirect_thunk_function): Use
23946 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
23947 for __x86_return_thunk.
23949 2018-01-15 Richard Biener <rguenther@suse.de>
23951 PR middle-end/83850
23952 * expmed.c (extract_bit_field_1): Fix typo.
23954 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23957 * config/arm/iterators.md (VF): New mode iterator.
23958 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
23959 Remove integer-related logic from pattern.
23960 (neon_vabd<mode>_3): Likewise.
23962 2018-01-15 Jakub Jelinek <jakub@redhat.com>
23964 PR middle-end/82694
23965 * common.opt (fstrict-overflow): No longer an alias.
23966 (fwrapv-pointer): New option.
23967 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
23968 also for pointer types based on flag_wrapv_pointer.
23969 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
23970 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
23971 opts->x_flag_wrapv got set.
23972 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
23973 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
23974 POINTER_TYPE_OVERFLOW_UNDEFINED.
23975 * match.pd: Likewise in address comparison pattern.
23976 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
23978 2018-01-15 Richard Biener <rguenther@suse.de>
23981 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
23982 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
23983 Reset type names to their identifier if their TYPE_DECL doesn't
23984 have linkage (and thus is used for ODR and devirt).
23985 (save_debug_info_for_decl): Remove.
23986 (save_debug_info_for_type): Likewise.
23987 (add_tree_to_fld_list): Adjust.
23988 * tree-pretty-print.c (dump_generic_node): Make dumping of
23989 type names more robust.
23991 2018-01-15 Richard Biener <rguenther@suse.de>
23993 * BASE-VER: Bump to 8.0.1.
23995 2018-01-14 Martin Sebor <msebor@redhat.com>
23998 * builtins.c (check_access): Avoid warning when the no-warning bit
24001 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
24003 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
24004 * ira-color (allocno_hard_regs_compare): Likewise.
24006 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
24009 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
24010 Use .pushsection/.popsection.
24012 2018-01-14 Martin Sebor <msebor@redhat.com>
24015 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
24017 2018-01-14 Jakub Jelinek <jakub@redhat.com>
24019 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
24020 entry from extra_headers.
24021 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
24022 extra_headers, make the list bitwise identical to the i?86-*-* one.
24024 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
24026 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
24027 -mcmodel=large with -mindirect-branch=thunk,
24028 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
24029 -mfunction-return=thunk-extern.
24030 * doc/invoke.texi: Document -mcmodel=large is incompatible with
24031 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
24032 -mfunction-return=thunk and -mfunction-return=thunk-extern.
24034 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
24036 * config/i386/i386.c (print_reg): Print the name of the full
24037 integer register without '%'.
24038 (ix86_print_operand): Handle 'V'.
24039 * doc/extend.texi: Document 'V' modifier.
24041 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
24043 * config/i386/constraints.md (Bs): Disallow memory operand for
24044 -mindirect-branch-register.
24046 * config/i386/predicates.md (indirect_branch_operand): Likewise.
24047 (GOT_memory_operand): Likewise.
24048 (call_insn_operand): Likewise.
24049 (sibcall_insn_operand): Likewise.
24050 (GOT32_symbol_operand): Likewise.
24051 * config/i386/i386.md (indirect_jump): Call convert_memory_address
24052 for -mindirect-branch-register.
24053 (tablejump): Likewise.
24054 (*sibcall_memory): Likewise.
24055 (*sibcall_value_memory): Likewise.
24056 Disallow peepholes of indirect call and jump via memory for
24057 -mindirect-branch-register.
24058 (*call_pop): Replace m with Bw.
24059 (*call_value_pop): Likewise.
24060 (*sibcall_pop_memory): Replace m with Bs.
24061 * config/i386/i386.opt (mindirect-branch-register): New option.
24062 * doc/invoke.texi: Document -mindirect-branch-register option.
24064 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
24066 * config/i386/i386-protos.h (ix86_output_function_return): New.
24067 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
24068 set function_return_type.
24069 (indirect_thunk_name): Add ret_p to indicate thunk for function
24071 (output_indirect_thunk_function): Pass false to
24072 indirect_thunk_name.
24073 (ix86_output_indirect_branch_via_reg): Likewise.
24074 (ix86_output_indirect_branch_via_push): Likewise.
24075 (output_indirect_thunk_function): Create alias for function
24076 return thunk if regno < 0.
24077 (ix86_output_function_return): New function.
24078 (ix86_handle_fndecl_attribute): Handle function_return.
24079 (ix86_attribute_table): Add function_return.
24080 * config/i386/i386.h (machine_function): Add
24081 function_return_type.
24082 * config/i386/i386.md (simple_return_internal): Use
24083 ix86_output_function_return.
24084 (simple_return_internal_long): Likewise.
24085 * config/i386/i386.opt (mfunction-return=): New option.
24086 (indirect_branch): Mention -mfunction-return=.
24087 * doc/extend.texi: Document function_return function attribute.
24088 * doc/invoke.texi: Document -mfunction-return= option.
24090 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
24092 * config/i386/i386-opts.h (indirect_branch): New.
24093 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
24094 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
24095 with local indirect jump when converting indirect call and jump.
24096 (ix86_set_indirect_branch_type): New.
24097 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
24098 (indirectlabelno): New.
24099 (indirect_thunk_needed): Likewise.
24100 (indirect_thunk_bnd_needed): Likewise.
24101 (indirect_thunks_used): Likewise.
24102 (indirect_thunks_bnd_used): Likewise.
24103 (INDIRECT_LABEL): Likewise.
24104 (indirect_thunk_name): Likewise.
24105 (output_indirect_thunk): Likewise.
24106 (output_indirect_thunk_function): Likewise.
24107 (ix86_output_indirect_branch_via_reg): Likewise.
24108 (ix86_output_indirect_branch_via_push): Likewise.
24109 (ix86_output_indirect_branch): Likewise.
24110 (ix86_output_indirect_jmp): Likewise.
24111 (ix86_code_end): Call output_indirect_thunk_function if needed.
24112 (ix86_output_call_insn): Call ix86_output_indirect_branch if
24114 (ix86_handle_fndecl_attribute): Handle indirect_branch.
24115 (ix86_attribute_table): Add indirect_branch.
24116 * config/i386/i386.h (machine_function): Add indirect_branch_type
24117 and has_local_indirect_jump.
24118 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
24120 (tablejump): Likewise.
24121 (*indirect_jump): Use ix86_output_indirect_jmp.
24122 (*tablejump_1): Likewise.
24123 (simple_return_indirect_internal): Likewise.
24124 * config/i386/i386.opt (mindirect-branch=): New option.
24125 (indirect_branch): New.
24128 (thunk-inline): Likewise.
24129 (thunk-extern): Likewise.
24130 * doc/extend.texi: Document indirect_branch function attribute.
24131 * doc/invoke.texi: Document -mindirect-branch= option.
24133 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
24136 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
24138 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
24140 * ipa-inline.c (want_inline_small_function_p): Return false if
24141 inlining has already failed with CIF_FINAL_ERROR.
24142 (update_caller_keys): Call want_inline_small_function_p before
24144 (update_callee_keys): Likewise.
24146 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
24148 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
24150 (rs6000_quadword_masked_address_p): Likewise.
24151 (quad_aligned_load_p): Likewise.
24152 (quad_aligned_store_p): Likewise.
24153 (const_load_sequence_p): Add comment to describe the outer-most loop.
24154 (mimic_memory_attributes_and_flags): New function.
24155 (rs6000_gen_stvx): Likewise.
24156 (replace_swapped_aligned_store): Likewise.
24157 (rs6000_gen_lvx): Likewise.
24158 (replace_swapped_aligned_load): Likewise.
24159 (replace_swapped_load_constant): Capitalize argument name in
24160 comment describing this function.
24161 (rs6000_analyze_swaps): Add a third pass to search for vector loads
24162 and stores that access quad-word aligned addresses and replace
24163 with stvx or lvx instructions when appropriate.
24164 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
24165 New function prototype.
24166 (rs6000_quadword_masked_address_p): Likewise.
24167 (rs6000_gen_lvx): Likewise.
24168 (rs6000_gen_stvx): Likewise.
24169 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
24170 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
24171 when memory address is aligned.
24172 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
24173 this split to select lvx instruction when memory address is aligned.
24174 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
24175 instruction when memory address is aligned.
24176 (*vsx_le_perm_load_v16qi): Likewise.
24177 (four unnamed splitters): Modify to select the stvx instruction
24178 when memory is aligned.
24180 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
24182 * predict.c (determine_unlikely_bbs): Handle correctly BBs
24183 which appears in the queue multiple times.
24185 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24186 Alan Hayward <alan.hayward@arm.com>
24187 David Sherwood <david.sherwood@arm.com>
24189 * tree-vectorizer.h (vec_lower_bound): New structure.
24190 (_loop_vec_info): Add check_nonzero and lower_bounds.
24191 (LOOP_VINFO_CHECK_NONZERO): New macro.
24192 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
24193 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
24194 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
24195 fields. Make seg_len the distance travelled, not including the
24197 (dr_direction_indicator): Declare.
24198 (dr_zero_step_indicator): Likewise.
24199 (dr_known_forward_stride_p): Likewise.
24200 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
24202 (runtime_alias_check_p): Allow runtime alias checks with
24204 (operator ==): Compare access_size and align.
24205 (prune_runtime_alias_test_list): Rework for new distinction between
24206 the access_size and seg_len.
24207 (create_intersect_range_checks_index): Likewise. Cope with polynomial
24209 (get_segment_min_max): New function.
24210 (create_intersect_range_checks): Use it.
24211 (dr_step_indicator): New function.
24212 (dr_direction_indicator): Likewise.
24213 (dr_zero_step_indicator): Likewise.
24214 (dr_known_forward_stride_p): Likewise.
24215 * tree-loop-distribution.c (data_ref_segment_size): Return
24216 DR_STEP * (niters - 1).
24217 (compute_alias_check_pairs): Update call to the dr_with_seg_len
24219 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
24220 (vect_preserves_scalar_order_p): New function, split out from...
24221 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
24222 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
24223 (vect_vfa_access_size): New function.
24224 (vect_vfa_align): Likewise.
24225 (vect_compile_time_alias): Take access_size_a and access_b arguments.
24226 (dump_lower_bound): New function.
24227 (vect_check_lower_bound): Likewise.
24228 (vect_small_gap_p): Likewise.
24229 (vectorizable_with_step_bound_p): Likewise.
24230 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
24231 depencies if the vectorization factor is 1. Convert the checks
24232 for nonzero steps into checks on the bounds of DR_STEP. Try using
24233 a bunds check for variable steps if the minimum required step is
24234 relatively small. Update calls to the dr_with_seg_len
24235 constructor and to vect_compile_time_alias.
24236 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
24238 (vect_loop_versioning): Call it.
24239 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
24241 (vect_estimate_min_profitable_iters): Account for any bounds checks.
24243 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24244 Alan Hayward <alan.hayward@arm.com>
24245 David Sherwood <david.sherwood@arm.com>
24247 * doc/sourcebuild.texi (vect_scatter_store): Document.
24248 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
24250 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
24252 * genopinit.c (main): Add supports_vec_scatter_store and
24253 supports_vec_scatter_store_cached to target_optabs.
24254 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
24255 IFN_MASK_SCATTER_STORE.
24256 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
24258 * internal-fn.h (internal_store_fn_p): Declare.
24259 (internal_fn_stored_value_index): Likewise.
24260 * internal-fn.c (scatter_store_direct): New macro.
24261 (expand_scatter_store_optab_fn): New function.
24262 (direct_scatter_store_optab_supported_p): New macro.
24263 (internal_store_fn_p): New function.
24264 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
24265 IFN_MASK_SCATTER_STORE.
24266 (internal_fn_mask_index): Likewise.
24267 (internal_fn_stored_value_index): New function.
24268 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
24269 for scatter stores.
24270 * optabs-query.h (supports_vec_scatter_store_p): Declare.
24271 * optabs-query.c (supports_vec_scatter_store_p): New function.
24272 * tree-vectorizer.h (vect_get_store_rhs): Declare.
24273 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
24274 true for scatter stores.
24275 (vect_gather_scatter_fn_p): Handle scatter stores too.
24276 (vect_check_gather_scatter): Consider using scatter stores if
24277 supports_vec_scatter_store_p.
24278 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
24279 scatter stores too.
24280 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
24281 internal_fn_stored_value_index.
24282 (check_load_store_masking): Handle scatter stores too.
24283 (vect_get_store_rhs): Make public.
24284 (vectorizable_call): Use internal_store_fn_p.
24285 (vectorizable_store): Handle scatter store internal functions.
24286 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
24287 when deciding whether the end of the group has been reached.
24288 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
24289 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
24290 (mask_scatter_store<mode>): New insns.
24292 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24293 Alan Hayward <alan.hayward@arm.com>
24294 David Sherwood <david.sherwood@arm.com>
24296 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
24297 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
24298 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
24300 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
24301 Use vect_truncate_gather_scatter_offset if we can't treat the
24302 operation as a normal gather load or scatter store.
24303 (get_group_load_store_type): Take the gather_scatter_info
24304 as argument. Try using a gather load or scatter store for
24305 single-element groups.
24306 (get_load_store_type): Update calls to get_group_load_store_type
24307 and vect_use_strided_gather_scatters_p.
24309 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24310 Alan Hayward <alan.hayward@arm.com>
24311 David Sherwood <david.sherwood@arm.com>
24313 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
24314 optional tree argument.
24315 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
24317 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
24318 but continue to use the current value as a fallback.
24319 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
24320 to compare the updates.
24321 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
24322 (get_load_store_type): Use it when handling a strided access.
24323 (vect_get_strided_load_store_ops): New function.
24324 (vect_get_data_ptr_increment): Likewise.
24325 (vectorizable_load): Handle strided gather loads. Always pass
24326 a step to vect_create_data_ref_ptr and bump_vector_ptr.
24328 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24329 Alan Hayward <alan.hayward@arm.com>
24330 David Sherwood <david.sherwood@arm.com>
24332 * doc/md.texi (gather_load@var{m}): Document.
24333 (mask_gather_load@var{m}): Likewise.
24334 * genopinit.c (main): Add supports_vec_gather_load and
24335 supports_vec_gather_load_cached to target_optabs.
24336 * optabs-tree.c (init_tree_optimization_optabs): Use
24337 ggc_cleared_alloc to allocate target_optabs.
24338 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
24339 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
24341 * internal-fn.h (internal_load_fn_p): Declare.
24342 (internal_gather_scatter_fn_p): Likewise.
24343 (internal_fn_mask_index): Likewise.
24344 (internal_gather_scatter_fn_supported_p): Likewise.
24345 * internal-fn.c (gather_load_direct): New macro.
24346 (expand_gather_load_optab_fn): New function.
24347 (direct_gather_load_optab_supported_p): New macro.
24348 (direct_internal_fn_optab): New function.
24349 (internal_load_fn_p): Likewise.
24350 (internal_gather_scatter_fn_p): Likewise.
24351 (internal_fn_mask_index): Likewise.
24352 (internal_gather_scatter_fn_supported_p): Likewise.
24353 * optabs-query.c (supports_at_least_one_mode_p): New function.
24354 (supports_vec_gather_load_p): Likewise.
24355 * optabs-query.h (supports_vec_gather_load_p): Declare.
24356 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
24357 and memory_type field.
24358 (NUM_PATTERNS): Bump to 15.
24359 * tree-vect-data-refs.c: Include internal-fn.h.
24360 (vect_gather_scatter_fn_p): New function.
24361 (vect_describe_gather_scatter_call): Likewise.
24362 (vect_check_gather_scatter): Try using internal functions for
24363 gather loads. Recognize existing calls to a gather load function.
24364 (vect_analyze_data_refs): Consider using gather loads if
24365 supports_vec_gather_load_p.
24366 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
24367 (vect_get_gather_scatter_offset_type): Likewise.
24368 (vect_convert_mask_for_vectype): Likewise.
24369 (vect_add_conversion_to_patterm): Likewise.
24370 (vect_try_gather_scatter_pattern): Likewise.
24371 (vect_recog_gather_scatter_pattern): New pattern recognizer.
24372 (vect_vect_recog_func_ptrs): Add it.
24373 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
24374 internal_fn_mask_index and internal_gather_scatter_fn_p.
24375 (check_load_store_masking): Take the gather_scatter_info as an
24376 argument and handle gather loads.
24377 (vect_get_gather_scatter_ops): New function.
24378 (vectorizable_call): Check internal_load_fn_p.
24379 (vectorizable_load): Likewise. Handle gather load internal
24381 (vectorizable_store): Update call to check_load_store_masking.
24382 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
24383 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
24384 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
24385 (aarch64_gather_scale_operand_d): New predicates.
24386 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
24387 (mask_gather_load<mode>): New insns.
24389 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24390 Alan Hayward <alan.hayward@arm.com>
24391 David Sherwood <david.sherwood@arm.com>
24393 * optabs.def (fold_left_plus_optab): New optab.
24394 * doc/md.texi (fold_left_plus_@var{m}): Document.
24395 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
24396 * internal-fn.c (fold_left_direct): Define.
24397 (expand_fold_left_optab_fn): Likewise.
24398 (direct_fold_left_optab_supported_p): Likewise.
24399 * fold-const-call.c (fold_const_fold_left): New function.
24400 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
24401 * tree-parloops.c (valid_reduction_p): New function.
24402 (gather_scalar_reductions): Use it.
24403 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
24404 (vect_finish_replace_stmt): Declare.
24405 * tree-vect-loop.c (fold_left_reduction_fn): New function.
24406 (needs_fold_left_reduction_p): New function, split out from...
24407 (vect_is_simple_reduction): ...here. Accept reductions that
24408 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
24409 (vect_force_simple_reduction): Also store the reduction type in
24410 the assignment's STMT_VINFO_REDUC_TYPE.
24411 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
24412 (merge_with_identity): New function.
24413 (vect_expand_fold_left): Likewise.
24414 (vectorize_fold_left_reduction): Likewise.
24415 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
24416 scalar phi in place for it. Check for target support and reject
24417 cases that would reassociate the operation. Defer the transform
24418 phase to vectorize_fold_left_reduction.
24419 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
24420 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
24421 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
24423 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24425 * tree-if-conv.c (predicate_mem_writes): Remove redundant
24426 call to ifc_temp_var.
24428 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24429 Alan Hayward <alan.hayward@arm.com>
24430 David Sherwood <david.sherwood@arm.com>
24432 * target.def (legitimize_address_displacement): Take the original
24433 offset as a poly_int.
24434 * targhooks.h (default_legitimize_address_displacement): Update
24436 * targhooks.c (default_legitimize_address_displacement): Likewise.
24437 * doc/tm.texi: Regenerate.
24438 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
24439 as an argument, moving assert of ad->disp == ad->disp_term to...
24440 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
24441 Try calling targetm.legitimize_address_displacement before expanding
24442 the address rather than afterwards, and adjust for the new interface.
24443 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
24444 Match the new hook interface. Handle SVE addresses.
24445 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
24446 new hook interface.
24448 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24450 * Makefile.in (OBJS): Add early-remat.o.
24451 * target.def (select_early_remat_modes): New hook.
24452 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
24453 * doc/tm.texi: Regenerate.
24454 * targhooks.h (default_select_early_remat_modes): Declare.
24455 * targhooks.c (default_select_early_remat_modes): New function.
24456 * timevar.def (TV_EARLY_REMAT): New timevar.
24457 * passes.def (pass_early_remat): New pass.
24458 * tree-pass.h (make_pass_early_remat): Declare.
24459 * early-remat.c: New file.
24460 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
24462 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
24464 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24465 Alan Hayward <alan.hayward@arm.com>
24466 David Sherwood <david.sherwood@arm.com>
24468 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
24469 vfm1 with a bound_epilog parameter.
24470 (vect_do_peeling): Update calls accordingly, and move the prologue
24471 call earlier in the function. Treat the base bound_epilog as 0 for
24472 fully-masked loops and retain vf - 1 for other loops. Add 1 to
24473 this base when peeling for gaps.
24474 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
24475 with fully-masked loops.
24476 (vect_estimate_min_profitable_iters): Handle the single peeled
24477 iteration in that case.
24479 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24480 Alan Hayward <alan.hayward@arm.com>
24481 David Sherwood <david.sherwood@arm.com>
24483 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
24484 single-element interleaving even if the size is not a power of 2.
24485 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
24486 accesses for single-element interleaving if the group size is
24489 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24490 Alan Hayward <alan.hayward@arm.com>
24491 David Sherwood <david.sherwood@arm.com>
24493 * doc/md.texi (fold_extract_last_@var{m}): Document.
24494 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
24495 * optabs.def (fold_extract_last_optab): New optab.
24496 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
24497 * internal-fn.c (fold_extract_direct): New macro.
24498 (expand_fold_extract_optab_fn): Likewise.
24499 (direct_fold_extract_optab_supported_p): Likewise.
24500 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
24501 * tree-vect-loop.c (vect_model_reduction_cost): Handle
24502 EXTRACT_LAST_REDUCTION.
24503 (get_initial_def_for_reduction): Do not create an initial vector
24504 for EXTRACT_LAST_REDUCTION reductions.
24505 (vectorizable_reduction): Leave the scalar phi in place for
24506 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
24507 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
24508 epilogue code for EXTRACT_LAST_REDUCTION and defer the
24509 transform phase to vectorizable_condition.
24510 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
24512 (vect_finish_stmt_generation): ...here.
24513 (vect_finish_replace_stmt): New function.
24514 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
24515 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
24517 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
24519 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24520 Alan Hayward <alan.hayward@arm.com>
24521 David Sherwood <david.sherwood@arm.com>
24523 * doc/md.texi (extract_last_@var{m}): Document.
24524 * optabs.def (extract_last_optab): New optab.
24525 * internal-fn.def (EXTRACT_LAST): New internal function.
24526 * internal-fn.c (cond_unary_direct): New macro.
24527 (expand_cond_unary_optab_fn): Likewise.
24528 (direct_cond_unary_optab_supported_p): Likewise.
24529 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
24530 loops using EXTRACT_LAST.
24531 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
24532 (extract_last_<mode>): ...this optab.
24533 (vec_extract<mode><Vel>): Update accordingly.
24535 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24536 Alan Hayward <alan.hayward@arm.com>
24537 David Sherwood <david.sherwood@arm.com>
24539 * target.def (empty_mask_is_expensive): New hook.
24540 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
24541 * doc/tm.texi: Regenerate.
24542 * targhooks.h (default_empty_mask_is_expensive): Declare.
24543 * targhooks.c (default_empty_mask_is_expensive): New function.
24544 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
24545 if the target says that empty masks are expensive.
24546 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
24548 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
24550 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24551 Alan Hayward <alan.hayward@arm.com>
24552 David Sherwood <david.sherwood@arm.com>
24554 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
24555 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
24556 (vect_use_loop_mask_for_alignment_p): New function.
24557 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
24558 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
24559 niters_skip argument. Make sure that the first niters_skip elements
24560 of the first iteration are inactive.
24561 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
24562 Update call to vect_set_loop_masks_directly.
24563 (get_misalign_in_elems): New function, split out from...
24564 (vect_gen_prolog_loop_niters): ...here.
24565 (vect_update_init_of_dr): Take a code argument that specifies whether
24566 the adjustment should be added or subtracted.
24567 (vect_update_init_of_drs): Likewise.
24568 (vect_prepare_for_masked_peels): New function.
24569 (vect_do_peeling): Skip prologue peeling if we're using a mask
24570 instead. Update call to vect_update_inits_of_drs.
24571 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
24573 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
24574 alignment. Do not include the number of peeled iterations in
24575 the minimum threshold in that case.
24576 (vectorizable_induction): Adjust the start value down by
24577 LOOP_VINFO_MASK_SKIP_NITERS iterations.
24578 (vect_transform_loop): Call vect_prepare_for_masked_peels.
24579 Take the number of skipped iterations into account when calculating
24581 * tree-vect-stmts.c (vect_gen_while_not): New function.
24583 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24584 Alan Hayward <alan.hayward@arm.com>
24585 David Sherwood <david.sherwood@arm.com>
24587 * doc/sourcebuild.texi (vect_fully_masked): Document.
24588 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
24589 default value to 0.
24590 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
24592 (vect_analyze_loop_2): ...here. Don't check the vectorization
24593 factor against the number of loop iterations if the loop is
24596 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24597 Alan Hayward <alan.hayward@arm.com>
24598 David Sherwood <david.sherwood@arm.com>
24600 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
24601 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
24602 (dump_groups): Update accordingly.
24603 (iv_use::mem_type): New member variable.
24604 (address_p): New function.
24605 (record_use): Add a mem_type argument and initialize the new
24607 (record_group_use): Add a mem_type argument. Use address_p.
24608 Remove obsolete null checks of base_object. Update call to record_use.
24609 (find_interesting_uses_op): Update call to record_group_use.
24610 (find_interesting_uses_cond): Likewise.
24611 (find_interesting_uses_address): Likewise.
24612 (get_mem_type_for_internal_fn): New function.
24613 (find_address_like_use): Likewise.
24614 (find_interesting_uses_stmt): Try find_address_like_use before
24615 calling find_interesting_uses_op.
24616 (addr_offset_valid_p): Use the iv mem_type field as the type
24617 of the addressed memory.
24618 (add_autoinc_candidates): Likewise.
24619 (get_address_cost): Likewise.
24620 (split_small_address_groups_p): Use address_p.
24621 (split_address_groups): Likewise.
24622 (add_iv_candidate_for_use): Likewise.
24623 (autoinc_possible_for_pair): Likewise.
24624 (rewrite_groups): Likewise.
24625 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
24626 (determine_group_iv_cost): Update after split of USE_ADDRESS.
24627 (get_alias_ptr_type_for_ptr_address): New function.
24628 (rewrite_use_address): Rewrite address uses in calls that were
24629 identified by find_address_like_use.
24631 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24632 Alan Hayward <alan.hayward@arm.com>
24633 David Sherwood <david.sherwood@arm.com>
24635 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
24637 * gimple-expr.h (is_gimple_addressable: Likewise.
24638 * gimple-expr.c (is_gimple_address): Likewise.
24639 * internal-fn.c (expand_call_mem_ref): New function.
24640 (expand_mask_load_optab_fn): Use it.
24641 (expand_mask_store_optab_fn): Likewise.
24643 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24644 Alan Hayward <alan.hayward@arm.com>
24645 David Sherwood <david.sherwood@arm.com>
24647 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
24648 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
24649 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
24650 (cond_umax@var{mode}): Document.
24651 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
24652 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
24653 (cond_umin_optab, cond_umax_optab): New optabs.
24654 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
24655 (COND_IOR, COND_XOR): New internal functions.
24656 * internal-fn.h (get_conditional_internal_fn): Declare.
24657 * internal-fn.c (cond_binary_direct): New macro.
24658 (expand_cond_binary_optab_fn): Likewise.
24659 (direct_cond_binary_optab_supported_p): Likewise.
24660 (get_conditional_internal_fn): New function.
24661 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
24662 Cope with reduction statements that are vectorized as calls rather
24664 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
24665 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
24666 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
24667 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
24668 (UNSPEC_COND_EOR): New unspecs.
24669 (optab): Add mappings for them.
24670 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
24671 (sve_int_op, sve_fp_op): New int attributes.
24673 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24674 Alan Hayward <alan.hayward@arm.com>
24675 David Sherwood <david.sherwood@arm.com>
24677 * optabs.def (while_ult_optab): New optab.
24678 * doc/md.texi (while_ult@var{m}@var{n}): Document.
24679 * internal-fn.def (WHILE_ULT): New internal function.
24680 * internal-fn.h (direct_internal_fn_supported_p): New override
24681 that takes two types as argument.
24682 * internal-fn.c (while_direct): New macro.
24683 (expand_while_optab_fn): New function.
24684 (convert_optab_supported_p): Likewise.
24685 (direct_while_optab_supported_p): New macro.
24686 * wide-int.h (wi::udiv_ceil): New function.
24687 * tree-vectorizer.h (rgroup_masks): New structure.
24688 (vec_loop_masks): New typedef.
24689 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
24690 and fully_masked_p.
24691 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
24692 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
24693 (vect_max_vf): New function.
24694 (slpeel_make_loop_iterate_ntimes): Delete.
24695 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
24696 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
24697 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
24698 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
24699 internal-fn.h, stor-layout.h and optabs-query.h.
24700 (vect_set_loop_mask): New function.
24701 (add_preheader_seq): Likewise.
24702 (add_header_seq): Likewise.
24703 (interleave_supported_p): Likewise.
24704 (vect_maybe_permute_loop_masks): Likewise.
24705 (vect_set_loop_masks_directly): Likewise.
24706 (vect_set_loop_condition_masked): Likewise.
24707 (vect_set_loop_condition_unmasked): New function, split out from
24708 slpeel_make_loop_iterate_ntimes.
24709 (slpeel_make_loop_iterate_ntimes): Rename to..
24710 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
24711 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
24712 (vect_do_peeling): Update call accordingly.
24713 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
24715 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
24716 mask_compare_type, can_fully_mask_p and fully_masked_p.
24717 (release_vec_loop_masks): New function.
24718 (_loop_vec_info): Use it to free the loop masks.
24719 (can_produce_all_loop_masks_p): New function.
24720 (vect_get_max_nscalars_per_iter): Likewise.
24721 (vect_verify_full_masking): Likewise.
24722 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
24723 retries, and free the mask rgroups before retrying. Check loop-wide
24724 reasons for disallowing fully-masked loops. Make the final decision
24725 about whether use a fully-masked loop or not.
24726 (vect_estimate_min_profitable_iters): Do not assume that peeling
24727 for the number of iterations will be needed for fully-masked loops.
24728 (vectorizable_reduction): Disable fully-masked loops.
24729 (vectorizable_live_operation): Likewise.
24730 (vect_halve_mask_nunits): New function.
24731 (vect_double_mask_nunits): Likewise.
24732 (vect_record_loop_mask): Likewise.
24733 (vect_get_loop_mask): Likewise.
24734 (vect_transform_loop): Handle the case in which the final loop
24735 iteration might handle a partial vector. Call vect_set_loop_condition
24736 instead of slpeel_make_loop_iterate_ntimes.
24737 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
24738 (check_load_store_masking): New function.
24739 (prepare_load_store_mask): Likewise.
24740 (vectorizable_store): Handle fully-masked loops.
24741 (vectorizable_load): Likewise.
24742 (supportable_widening_operation): Use vect_halve_mask_nunits for
24744 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
24745 (vect_gen_while): New function.
24746 * config/aarch64/aarch64.md (umax<mode>3): New expander.
24747 (aarch64_uqdec<mode>): New insn.
24749 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24750 Alan Hayward <alan.hayward@arm.com>
24751 David Sherwood <david.sherwood@arm.com>
24753 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
24754 (reduc_xor_scal_optab): New optabs.
24755 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
24756 (reduc_xor_scal_@var{m}): Document.
24757 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
24758 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
24759 internal functions.
24760 * fold-const-call.c (fold_const_call): Handle them.
24761 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
24762 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
24763 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
24764 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
24765 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
24766 (UNSPEC_XORV): New unspecs.
24767 (optab): Add entries for them.
24768 (BITWISEV): New int iterator.
24769 (bit_reduc_op): New int attributes.
24771 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24772 Alan Hayward <alan.hayward@arm.com>
24773 David Sherwood <david.sherwood@arm.com>
24775 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
24776 * internal-fn.def (VEC_SHL_INSERT): New internal function.
24777 * optabs.def (vec_shl_insert_optab): New optab.
24778 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
24779 (duplicate_and_interleave): Likewise.
24780 * tree-vect-loop.c: Include internal-fn.h.
24781 (neutral_op_for_slp_reduction): New function, split out from
24782 get_initial_defs_for_reduction.
24783 (get_initial_def_for_reduction): Handle option 2 for variable-length
24784 vectors by loading the neutral value into a vector and then shifting
24785 the initial value into element 0.
24786 (get_initial_defs_for_reduction): Replace the code argument with
24787 the neutral value calculated by neutral_op_for_slp_reduction.
24788 Use gimple_build_vector for constant-length vectors.
24789 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
24790 but the first group_size elements have a neutral value.
24791 Use duplicate_and_interleave otherwise.
24792 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
24793 Update call to get_initial_defs_for_reduction. Handle SLP
24794 reductions for variable-length vectors by creating one vector
24795 result for each scalar result, with the elements associated
24796 with other scalar results stubbed out with the neutral value.
24797 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
24798 Require IFN_VEC_SHL_INSERT for double reductions on
24799 variable-length vectors, or SLP reductions that have
24800 a neutral value. Require can_duplicate_and_interleave_p
24801 support for variable-length unchained SLP reductions if there
24802 is no neutral value, such as for MIN/MAX reductions. Also require
24803 the number of vector elements to be a multiple of the number of
24804 SLP statements when doing variable-length unchained SLP reductions.
24805 Update call to vect_create_epilog_for_reduction.
24806 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
24807 and remove initial values.
24808 (duplicate_and_interleave): Make public.
24809 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
24810 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
24812 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24813 Alan Hayward <alan.hayward@arm.com>
24814 David Sherwood <david.sherwood@arm.com>
24816 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
24817 (can_duplicate_and_interleave_p): New function.
24818 (vect_get_and_check_slp_defs): Take the vector of statements
24819 rather than just the current one. Remove excess parentheses.
24820 Restriction rejectinon of vect_constant_def and vect_external_def
24821 for variable-length vectors to boolean types, or types for which
24822 can_duplicate_and_interleave_p is false.
24823 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
24824 (duplicate_and_interleave): New function.
24825 (vect_get_constant_vectors): Use gimple_build_vector for
24826 constant-length vectors and suitable variable-length constant
24827 vectors. Use duplicate_and_interleave for other variable-length
24828 vectors. Don't defer the update when inserting new statements.
24830 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24831 Alan Hayward <alan.hayward@arm.com>
24832 David Sherwood <david.sherwood@arm.com>
24834 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
24835 min_profitable_iters doesn't go negative.
24837 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24838 Alan Hayward <alan.hayward@arm.com>
24839 David Sherwood <david.sherwood@arm.com>
24841 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
24842 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
24843 * optabs.def (vec_mask_load_lanes_optab): New optab.
24844 (vec_mask_store_lanes_optab): Likewise.
24845 * internal-fn.def (MASK_LOAD_LANES): New internal function.
24846 (MASK_STORE_LANES): Likewise.
24847 * internal-fn.c (mask_load_lanes_direct): New macro.
24848 (mask_store_lanes_direct): Likewise.
24849 (expand_mask_load_optab_fn): Handle masked operations.
24850 (expand_mask_load_lanes_optab_fn): New macro.
24851 (expand_mask_store_optab_fn): Handle masked operations.
24852 (expand_mask_store_lanes_optab_fn): New macro.
24853 (direct_mask_load_lanes_optab_supported_p): Likewise.
24854 (direct_mask_store_lanes_optab_supported_p): Likewise.
24855 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
24857 (vect_load_lanes_supported): Likewise.
24858 * tree-vect-data-refs.c (strip_conversion): New function.
24859 (can_group_stmts_p): Likewise.
24860 (vect_analyze_data_ref_accesses): Use it instead of checking
24861 for a pair of assignments.
24862 (vect_store_lanes_supported): Take a masked_p parameter.
24863 (vect_load_lanes_supported): Likewise.
24864 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
24865 vect_store_lanes_supported and vect_load_lanes_supported.
24866 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
24867 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
24868 parameter. Don't allow gaps for masked accesses.
24869 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
24870 and vect_load_lanes_supported.
24871 (get_load_store_type): Take a masked_p parameter and update
24872 call to get_group_load_store_type.
24873 (vectorizable_store): Update call to get_load_store_type.
24874 Handle IFN_MASK_STORE_LANES.
24875 (vectorizable_load): Update call to get_load_store_type.
24876 Handle IFN_MASK_LOAD_LANES.
24878 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24879 Alan Hayward <alan.hayward@arm.com>
24880 David Sherwood <david.sherwood@arm.com>
24882 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
24884 * config/aarch64/aarch64-protos.h
24885 (aarch64_sve_struct_memory_operand_p): Declare.
24886 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
24887 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
24888 (VPRED, vpred): Handle SVE structure modes.
24889 * config/aarch64/constraints.md (Utx): New constraint.
24890 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
24891 (aarch64_sve_struct_nonimmediate_operand): New predicates.
24892 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
24893 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
24894 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
24895 structure modes. Split into pieces after RA.
24896 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
24897 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
24899 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
24900 SVE structure modes.
24901 (aarch64_classify_address): Likewise.
24902 (sizetochar): Move earlier in file.
24903 (aarch64_print_operand): Handle SVE register lists.
24904 (aarch64_array_mode): New function.
24905 (aarch64_sve_struct_memory_operand_p): Likewise.
24906 (TARGET_ARRAY_MODE): Redefine.
24908 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24909 Alan Hayward <alan.hayward@arm.com>
24910 David Sherwood <david.sherwood@arm.com>
24912 * target.def (array_mode): New target hook.
24913 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
24914 * doc/tm.texi: Regenerate.
24915 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
24916 * hooks.c (hook_optmode_mode_uhwi_none): New function.
24917 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
24918 targetm.array_mode.
24919 * stor-layout.c (mode_for_array): Likewise. Support polynomial
24922 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24923 Alan Hayward <alan.hayward@arm.com>
24924 David Sherwood <david.sherwood@arm.com>
24926 * fold-const.c (fold_binary_loc): Check the argument types
24927 rather than the result type when testing for a vector operation.
24929 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24931 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
24932 * doc/tm.texi: Regenerate.
24934 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24935 Alan Hayward <alan.hayward@arm.com>
24936 David Sherwood <david.sherwood@arm.com>
24938 * doc/invoke.texi (-msve-vector-bits=): Document new option.
24939 (sve): Document new AArch64 extension.
24940 * doc/md.texi (w): Extend the description of the AArch64
24941 constraint to include SVE vectors.
24942 (Upl, Upa): Document new AArch64 predicate constraints.
24943 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
24945 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
24946 (msve-vector-bits=): New option.
24947 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
24948 SVE when these are disabled.
24949 (sve): New extension.
24950 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
24951 modes. Adjust their number of units based on aarch64_sve_vg.
24952 (MAX_BITSIZE_MODE_ANY_MODE): Define.
24953 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
24954 aarch64_addr_query_type.
24955 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
24956 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
24957 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
24958 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
24959 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
24960 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
24961 (aarch64_simd_imm_zero_p): Delete.
24962 (aarch64_check_zero_based_sve_index_immediate): Declare.
24963 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
24964 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
24965 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
24966 (aarch64_sve_float_mul_immediate_p): Likewise.
24967 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
24968 rather than an rtx.
24969 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
24970 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
24971 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
24972 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
24973 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
24974 (aarch64_regmode_natural_size): Likewise.
24975 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
24976 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
24978 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
24979 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
24980 for VG and the SVE predicate registers.
24981 (V_ALIASES): Add a "z"-prefixed alias.
24982 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
24983 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
24984 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
24985 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
24986 (REG_CLASS_NAMES): Add entries for them.
24987 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
24988 and the predicate registers.
24989 (aarch64_sve_vg): Declare.
24990 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
24991 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
24992 (REGMODE_NATURAL_SIZE): Define.
24993 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
24995 * config/aarch64/aarch64.c: Include cfgrtl.h.
24996 (simd_immediate_info): Add a constructor for series vectors,
24997 and an associated step field.
24998 (aarch64_sve_vg): New variable.
24999 (aarch64_dbx_register_number): Handle VG and the predicate registers.
25000 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
25001 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
25002 (VEC_ANY_DATA, VEC_STRUCT): New constants.
25003 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
25004 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
25005 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
25006 (aarch64_get_mask_mode): New functions.
25007 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
25008 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
25009 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
25010 predicate modes and predicate registers. Explicitly restrict
25011 GPRs to modes of 16 bytes or smaller. Only allow FP registers
25012 to store a vector mode if it is recognized by
25013 aarch64_classify_vector_mode.
25014 (aarch64_regmode_natural_size): New function.
25015 (aarch64_hard_regno_caller_save_mode): Return the original mode
25017 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
25018 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
25019 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
25020 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
25022 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
25023 does not overlap dest if the function is frame-related. Handle
25025 (aarch64_split_add_offset): New function.
25026 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
25027 them aarch64_add_offset.
25028 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
25029 and update call to aarch64_sub_sp.
25030 (aarch64_add_cfa_expression): New function.
25031 (aarch64_expand_prologue): Pass extra temporary registers to the
25032 functions above. Handle the case in which we need to emit new
25033 DW_CFA_expressions for registers that were originally saved
25034 relative to the stack pointer, but now have to be expressed
25035 relative to the frame pointer.
25036 (aarch64_output_mi_thunk): Pass extra temporary registers to the
25038 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
25039 IP0 and IP1 values for SVE frames.
25040 (aarch64_expand_vec_series): New function.
25041 (aarch64_expand_sve_widened_duplicate): Likewise.
25042 (aarch64_expand_sve_const_vector): Likewise.
25043 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
25044 Handle SVE constants. Use emit_move_insn to move a force_const_mem
25045 into the register, rather than emitting a SET directly.
25046 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
25047 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
25048 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
25049 (offset_9bit_signed_scaled_p): New functions.
25050 (aarch64_replicate_bitmask_imm): New function.
25051 (aarch64_bitmask_imm): Use it.
25052 (aarch64_cannot_force_const_mem): Reject expressions involving
25053 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
25054 (aarch64_classify_index): Handle SVE indices, by requiring
25055 a plain register index with a scale that matches the element size.
25056 (aarch64_classify_address): Handle SVE addresses. Assert that
25057 the mode of the address is VOIDmode or an integer mode.
25058 Update call to aarch64_classify_symbol.
25059 (aarch64_classify_symbolic_expression): Update call to
25060 aarch64_classify_symbol.
25061 (aarch64_const_vec_all_in_range_p): New function.
25062 (aarch64_print_vector_float_operand): Likewise.
25063 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
25064 "vN" for FP registers with SVE modes. Handle (const ...) vectors
25065 and the FP immediates 1.0 and 0.5.
25066 (aarch64_print_address_internal): Handle SVE addresses.
25067 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
25068 (aarch64_regno_regclass): Handle predicate registers.
25069 (aarch64_secondary_reload): Handle big-endian reloads of SVE
25071 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
25072 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
25073 (aarch64_convert_sve_vector_bits): New function.
25074 (aarch64_override_options): Use it to handle -msve-vector-bits=.
25075 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
25076 rather than an rtx.
25077 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
25078 Handle SVE vector and predicate modes. Accept VL-based constants
25079 that need only one temporary register, and VL offsets that require
25080 no temporary registers.
25081 (aarch64_conditional_register_usage): Mark the predicate registers
25082 as fixed if SVE isn't available.
25083 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
25084 Return true for SVE vector and predicate modes.
25085 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
25086 rather than an unsigned int. Handle SVE modes.
25087 (aarch64_preferred_simd_mode): Update call accordingly. Handle
25089 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
25091 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
25092 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
25093 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
25094 (aarch64_sve_float_mul_immediate_p): New functions.
25095 (aarch64_sve_valid_immediate): New function.
25096 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
25097 Explicitly reject structure modes. Check for INDEX constants.
25098 Handle PTRUE and PFALSE constants.
25099 (aarch64_check_zero_based_sve_index_immediate): New function.
25100 (aarch64_simd_imm_zero_p): Delete.
25101 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
25102 vector modes. Accept constants in the range of CNT[BHWD].
25103 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
25104 ask for an Advanced SIMD mode.
25105 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
25106 (aarch64_simd_vector_alignment): Handle SVE predicates.
25107 (aarch64_vectorize_preferred_vector_alignment): New function.
25108 (aarch64_simd_vector_alignment_reachable): Use it instead of
25110 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
25111 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
25113 (MAX_VECT_LEN): Delete.
25114 (expand_vec_perm_d): Add a vec_flags field.
25115 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
25116 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
25117 (aarch64_evpc_ext): Don't apply a big-endian lane correction
25119 (aarch64_evpc_rev): Rename to...
25120 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
25121 (aarch64_evpc_rev_global): New function.
25122 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
25123 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
25125 (aarch64_evpc_sve_tbl): New function.
25126 (aarch64_expand_vec_perm_const_1): Update after rename of
25127 aarch64_evpc_rev. Handle SVE permutes too, trying
25128 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
25129 than aarch64_evpc_tbl.
25130 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
25131 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
25132 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
25133 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
25134 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
25135 (aarch64_expand_sve_vcond): New functions.
25136 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
25137 of aarch64_vector_mode_p.
25138 (aarch64_dwarf_poly_indeterminate_value): New function.
25139 (aarch64_compute_pressure_classes): Likewise.
25140 (aarch64_can_change_mode_class): Likewise.
25141 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
25142 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
25143 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
25144 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
25145 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
25146 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
25147 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
25148 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
25150 (Dn, Dl, Dr): Accept const as well as const_vector.
25151 (Dz): Likewise. Compare against CONST0_RTX.
25152 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
25153 of "vector" where appropriate.
25154 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
25155 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
25156 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
25157 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
25158 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
25159 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
25160 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
25161 (v_int_equiv): Extend to SVE modes.
25162 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
25164 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
25165 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
25166 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
25167 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
25168 (SVE_COND_FP_CMP): New int iterators.
25169 (perm_hilo): Handle the new unpack unspecs.
25170 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
25172 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
25173 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
25174 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
25175 (aarch64_equality_operator, aarch64_constant_vector_operand)
25176 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
25177 (aarch64_sve_nonimmediate_operand): Likewise.
25178 (aarch64_sve_general_operand): Likewise.
25179 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
25180 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
25181 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
25182 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
25183 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
25184 (aarch64_sve_float_arith_immediate): Likewise.
25185 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
25186 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
25187 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
25188 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
25189 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
25190 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
25191 (aarch64_sve_float_arith_operand): Likewise.
25192 (aarch64_sve_float_arith_with_sub_operand): Likewise.
25193 (aarch64_sve_float_mul_operand): Likewise.
25194 (aarch64_sve_vec_perm_operand): Likewise.
25195 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
25196 (aarch64_mov_operand): Accept const_poly_int and const_vector.
25197 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
25198 as well as const_vector.
25199 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
25200 in file. Use CONST0_RTX and CONSTM1_RTX.
25201 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
25202 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
25203 Use aarch64_simd_imm_zero.
25204 * config/aarch64/aarch64-sve.md: New file.
25205 * config/aarch64/aarch64.md: Include it.
25206 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
25207 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
25208 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
25209 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
25210 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
25211 (sve): New attribute.
25212 (enabled): Disable instructions with the sve attribute unless
25214 (movqi, movhi): Pass CONST_POLY_INT operaneds through
25215 aarch64_expand_mov_immediate.
25216 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
25217 CNT[BHSD] immediates.
25218 (movti): Split CONST_POLY_INT moves into two halves.
25219 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
25220 Split additions that need a temporary here if the destination
25221 is the stack pointer.
25222 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
25223 (*add<mode>3_poly_1): New instruction.
25224 (set_clobber_cc): New expander.
25226 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25228 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
25229 parameter and use it instead of GET_MODE_SIZE (innermode). Use
25230 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
25231 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
25232 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
25233 Change innermode from fixed_mode_size to machine_mode.
25234 (simplify_subreg): Update call accordingly. Handle a constant-sized
25235 subreg of a variable-length CONST_VECTOR.
25237 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25238 Alan Hayward <alan.hayward@arm.com>
25239 David Sherwood <david.sherwood@arm.com>
25241 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
25242 (add_offset_to_base): New function, split out from...
25243 (create_mem_ref): ...here. When handling a scale other than 1,
25244 check first whether the address is valid without the offset.
25245 Add it into the base if so, leaving the index and scale as-is.
25247 2018-01-12 Jakub Jelinek <jakub@redhat.com>
25250 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
25251 fold_for_warn before checking if arg2 is INTEGER_CST.
25253 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
25255 * config/rs6000/predicates.md (load_multiple_operation): Delete.
25256 (store_multiple_operation): Delete.
25257 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
25258 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
25259 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
25260 guarded by TARGET_STRING.
25261 (rs6000_output_load_multiple): Delete.
25262 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
25263 OPTION_MASK_STRING / TARGET_STRING handling.
25264 (print_operand) <'N', 'O'>: Add comment that these are unused now.
25265 (const rs6000_opt_masks) <"string">: Change mask to 0.
25266 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
25267 (MASK_STRING): Delete.
25268 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
25270 (load_multiple): Delete.
25277 (store_multiple): Delete.
25284 (movmemsi_8reg): Delete.
25285 (corresponding unnamed define_insn): Delete.
25286 (movmemsi_6reg): Delete.
25287 (corresponding unnamed define_insn): Delete.
25288 (movmemsi_4reg): Delete.
25289 (corresponding unnamed define_insn): Delete.
25290 (movmemsi_2reg): Delete.
25291 (corresponding unnamed define_insn): Delete.
25292 (movmemsi_1reg): Delete.
25293 (corresponding unnamed define_insn): Delete.
25294 * config/rs6000/rs6000.opt (mno-string): New.
25295 (mstring): Replace by deprecation warning stub.
25296 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
25298 2018-01-12 Jakub Jelinek <jakub@redhat.com>
25300 * regrename.c (regrename_do_replace): If replacing the same
25301 reg multiple times, try to reuse last created gen_raw_REG.
25304 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
25305 main to workaround a bug in GDB.
25307 2018-01-12 Tom de Vries <tom@codesourcery.com>
25310 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
25312 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
25314 PR rtl-optimization/80481
25315 * ira-color.c (get_cap_member): New function.
25316 (allocnos_conflict_by_live_ranges_p): Use it.
25317 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
25318 (setup_slot_coalesced_allocno_live_ranges): Ditto.
25320 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
25323 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
25324 (*saddl_se_1): Ditto.
25325 (*ssubsi_1): Ditto.
25326 (*ssubl_se_1): Ditto.
25328 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
25330 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
25331 rather than wi::to_widest for DR_INITs.
25332 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
25333 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
25334 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
25336 (vect_analyze_group_access_1): Note that here.
25338 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
25340 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
25341 polynomial type sizes.
25343 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
25345 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
25346 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
25347 (gimple_add_tmp_var): Likewise.
25349 2018-01-12 Martin Liska <mliska@suse.cz>
25351 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
25352 (gimple_alloc_sizes): Likewise.
25353 (dump_gimple_statistics): Use PRIu64 in printf format.
25354 * gimple.h: Change uint64_t to int.
25356 2018-01-12 Martin Liska <mliska@suse.cz>
25358 * tree-core.h: Use uint64_t instead of int.
25359 * tree.c (tree_node_counts): Likewise.
25360 (tree_node_sizes): Likewise.
25361 (dump_tree_statistics): Use PRIu64 in printf format.
25363 2018-01-12 Martin Liska <mliska@suse.cz>
25365 * Makefile.in: As qsort_chk is implemented in vec.c, add
25366 vec.o to linkage of gencfn-macros.
25367 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
25368 passing the info to record_node_allocation_statistics.
25369 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
25371 * ggc-common.c (struct ggc_usage): Add operator== and use
25372 it in operator< and compare function.
25373 * mem-stats.h (struct mem_usage): Likewise.
25374 * vec.c (struct vec_usage): Remove operator< and compare
25375 function. Can be simply inherited.
25377 2018-01-12 Martin Jambor <mjambor@suse.cz>
25380 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
25381 * tree-ssa-math-opts.c: Include domwalk.h.
25382 (convert_mult_to_fma_1): New function.
25383 (fma_transformation_info): New type.
25384 (fma_deferring_state): Likewise.
25385 (cancel_fma_deferring): New function.
25386 (result_of_phi): Likewise.
25387 (last_fma_candidate_feeds_initial_phi): Likewise.
25388 (convert_mult_to_fma): Added deferring logic, split actual
25389 transformation to convert_mult_to_fma_1.
25390 (math_opts_dom_walker): New type.
25391 (math_opts_dom_walker::after_dom_children): New method, body moved
25392 here from pass_optimize_widening_mul::execute, added deferring logic
25394 (pass_optimize_widening_mul::execute): Moved most of code to
25395 math_opts_dom_walker::after_dom_children.
25396 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
25397 * config/i386/i386.c (ix86_option_override_internal): Added
25398 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
25400 2018-01-12 Richard Biener <rguenther@suse.de>
25403 * dwarf2out.c (gen_variable_die): Do not reset old_die for
25404 inline instance vars.
25406 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
25409 * config/rx/rx.c (rx_is_restricted_memory_address):
25410 Handle SUBREG case.
25412 2018-01-12 Richard Biener <rguenther@suse.de>
25414 PR tree-optimization/80846
25415 * target.def (split_reduction): New target hook.
25416 * targhooks.c (default_split_reduction): New function.
25417 * targhooks.h (default_split_reduction): Declare.
25418 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
25419 target requests first reduce vectors by combining low and high
25421 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
25422 (get_vectype_for_scalar_type_and_size): Export.
25423 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
25424 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
25425 * doc/tm.texi: Regenerate.
25426 * config/i386/i386.c (ix86_split_reduction): Implement
25427 TARGET_VECTORIZE_SPLIT_REDUCTION.
25429 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
25432 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
25433 in PIC mode except for TARGET_VXWORKS_RTP.
25434 * config/sparc/sparc.c: Include cfgrtl.h.
25435 (TARGET_INIT_PIC_REG): Define.
25436 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
25437 (sparc_pic_register_p): New predicate.
25438 (sparc_legitimate_address_p): Use it.
25439 (sparc_legitimize_pic_address): Likewise.
25440 (sparc_delegitimize_address): Likewise.
25441 (sparc_mode_dependent_address_p): Likewise.
25442 (gen_load_pcrel_sym): Remove 4th parameter.
25443 (load_got_register): Adjust call to above. Remove obsolete stuff.
25444 (sparc_expand_prologue): Do not call load_got_register here.
25445 (sparc_flat_expand_prologue): Likewise.
25446 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
25447 (sparc_use_pseudo_pic_reg): New function.
25448 (sparc_init_pic_reg): Likewise.
25449 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
25450 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
25452 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
25454 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
25455 Add item for branch_cost.
25457 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
25459 PR rtl-optimization/83565
25460 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
25461 not extend the result to a larger mode for rotate operations.
25462 (num_sign_bit_copies1): Likewise.
25464 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25467 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
25469 Use values-Xc.o for -pedantic.
25470 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
25472 2018-01-12 Martin Liska <mliska@suse.cz>
25475 * ipa-devirt.c (final_warning_record::grow_type_warnings):
25477 (possible_polymorphic_call_targets): Use it.
25478 (ipa_devirt): Likewise.
25480 2018-01-12 Martin Liska <mliska@suse.cz>
25482 * profile-count.h (enum profile_quality): Use 0 as invalid
25483 enum value of profile_quality.
25485 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
25487 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
25488 -mext-string options.
25490 2018-01-12 Richard Biener <rguenther@suse.de>
25492 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
25493 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
25494 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
25496 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
25498 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
25500 * configure.ac (--with-long-double-format): Add support for the
25501 configuration option to change the default long double format on
25503 * config.gcc (powerpc*-linux*-*): Likewise.
25504 * configure: Regenerate.
25505 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
25506 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
25507 used without modification.
25509 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25511 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
25512 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
25513 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
25514 MISC_BUILTIN_SPEC_BARRIER.
25515 (rs6000_init_builtins): Likewise.
25516 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
25518 (speculation_barrier): New define_insn.
25519 * doc/extend.texi: Document __builtin_speculation_barrier.
25521 2018-01-11 Jakub Jelinek <jakub@redhat.com>
25524 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
25525 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
25526 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
25528 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
25529 integral modes instead of "ss" and "sd".
25530 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
25531 vectors with 32-bit and 64-bit elements.
25532 (vecdupssescalarmodesuffix): New mode attribute.
25533 (vec_dup<mode>): Use it.
25535 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
25538 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
25539 frame if argument is passed on stack.
25541 2018-01-11 Jakub Jelinek <jakub@redhat.com>
25544 * ree.c (combine_reaching_defs): Optimize also
25545 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
25546 reg2=any_extend(exp); reg1=reg2;, formatting fix.
25548 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
25550 PR middle-end/83189
25551 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
25553 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
25555 PR middle-end/83718
25556 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
25557 after they are computed.
25559 2018-01-11 Bin Cheng <bin.cheng@arm.com>
25561 PR tree-optimization/83695
25562 * gimple-loop-linterchange.cc
25563 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
25564 reset cached scev information after interchange.
25565 (pass_linterchange::execute): Remove call to scev_reset_htab.
25567 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25569 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
25570 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
25571 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
25572 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
25573 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
25574 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
25575 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
25576 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
25577 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
25578 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
25579 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
25580 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
25581 (V_lane_reg): Likewise.
25582 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
25584 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
25585 (vfmal_lane_low<mode>_intrinsic,
25586 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
25587 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
25588 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
25589 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
25590 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
25591 vfmsl_lane_high<mode>_intrinsic): New define_insns.
25593 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25595 * config/arm/arm-cpus.in (fp16fml): New feature.
25596 (ALL_SIMD): Add fp16fml.
25597 (armv8.2-a): Add fp16fml as an option.
25598 (armv8.3-a): Likewise.
25599 (armv8.4-a): Add fp16fml as part of fp16.
25600 * config/arm/arm.h (TARGET_FP16FML): Define.
25601 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
25603 * config/arm/arm-modes.def (V2HF): Define.
25604 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
25605 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
25606 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
25607 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
25608 vfmsl_low, vfmsl_high): New set of builtins.
25609 * config/arm/iterators.md (PLUSMINUS): New code iterator.
25610 (vfml_op): New code attribute.
25611 (VFMLHALVES): New int iterator.
25612 (VFML, VFMLSEL): New mode attributes.
25613 (V_reg): Define mapping for V2HF.
25614 (V_hi, V_lo): New mode attributes.
25615 (VF_constraint): Likewise.
25616 (vfml_half, vfml_half_selector): New int attributes.
25617 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
25619 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
25620 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
25622 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
25623 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
25624 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
25625 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
25627 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
25628 Document new effective target and option set.
25630 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25632 * config/arm/arm-cpus.in (armv8_4): New feature.
25633 (ARMv8_4a): New fgroup.
25634 (armv8.4-a): New arch.
25635 * config/arm/arm-tables.opt: Regenerate.
25636 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
25637 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
25638 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
25639 Add matching rules for -march=armv8.4-a and extensions.
25640 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
25642 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
25645 * config/rx/rx.md (BW): New mode attribute.
25646 (sync_lock_test_and_setsi): Add mode suffix to insn output.
25648 2018-01-11 Richard Biener <rguenther@suse.de>
25650 PR tree-optimization/83435
25651 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
25652 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
25653 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
25655 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
25656 Alan Hayward <alan.hayward@arm.com>
25657 David Sherwood <david.sherwood@arm.com>
25659 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
25661 (aarch64_classify_address): Initialize it. Track polynomial offsets.
25662 (aarch64_print_address_internal): Use it to check for a zero offset.
25664 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
25665 Alan Hayward <alan.hayward@arm.com>
25666 David Sherwood <david.sherwood@arm.com>
25668 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
25669 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
25670 Return a poly_int64 rather than a HOST_WIDE_INT.
25671 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
25672 rather than a HOST_WIDE_INT.
25673 * config/aarch64/aarch64.h (aarch64_frame): Protect with
25674 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
25675 hard_fp_offset, frame_size, initial_adjust, callee_offset and
25676 final_offset from HOST_WIDE_INT to poly_int64.
25677 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
25678 to_constant when getting the number of units in an Advanced SIMD
25680 (aarch64_builtin_vectorized_function): Check for a constant number
25682 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
25684 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
25685 attribute instead of GET_MODE_NUNITS.
25686 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
25687 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
25688 GET_MODE_SIZE for fixed-size registers.
25689 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
25690 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
25691 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
25692 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
25693 (aarch64_print_operand, aarch64_print_address_internal)
25694 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
25695 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
25696 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
25697 Handle polynomial GET_MODE_SIZE.
25698 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
25699 wider than SImode without modification.
25700 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
25701 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
25702 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
25703 passing and returning SVE modes.
25704 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
25705 rather than GEN_INT.
25706 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
25707 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
25708 (aarch64_allocate_and_probe_stack_space): Likewise.
25709 (aarch64_layout_frame): Cope with polynomial offsets.
25710 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
25711 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
25712 polynomial offsets.
25713 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
25714 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
25715 poly_int64 rather than a HOST_WIDE_INT.
25716 (aarch64_get_separate_components, aarch64_process_components)
25717 (aarch64_expand_prologue, aarch64_expand_epilogue)
25718 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
25719 (aarch64_anchor_offset): New function, split out from...
25720 (aarch64_legitimize_address): ...here.
25721 (aarch64_builtin_vectorization_cost): Handle polynomial
25722 TYPE_VECTOR_SUBPARTS.
25723 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
25725 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
25726 number of elements from the PARALLEL rather than the mode.
25727 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
25728 rather than GET_MODE_BITSIZE.
25729 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
25730 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
25731 (aarch64_expand_vec_perm_const_1): Handle polynomial
25732 d->perm.length () and d->perm elements.
25733 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
25734 Apply to_constant to d->perm elements.
25735 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
25736 polynomial CONST_VECTOR_NUNITS.
25737 (aarch64_move_pointer): Take amount as a poly_int64 rather
25739 (aarch64_progress_pointer): Avoid temporary variable.
25740 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
25741 the mode attribute instead of GET_MODE.
25743 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
25744 Alan Hayward <alan.hayward@arm.com>
25745 David Sherwood <david.sherwood@arm.com>
25747 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
25748 x exists before using it.
25749 (aarch64_add_constant_internal): Rename to...
25750 (aarch64_add_offset_1): ...this. Replace regnum with separate
25751 src and dest rtxes. Handle the case in which they're different,
25752 including when the offset is zero. Replace scratchreg with an rtx.
25753 Use 2 additions if there is no spare register into which we can
25754 move a 16-bit constant.
25755 (aarch64_add_constant): Delete.
25756 (aarch64_add_offset): Replace reg with separate src and dest
25757 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
25758 Use aarch64_add_offset_1.
25759 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
25760 an rtx rather than an int. Take the delta as a poly_int64
25761 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
25762 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
25763 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
25764 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
25765 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
25766 and aarch64_add_sp.
25767 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
25768 aarch64_add_constant.
25770 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
25772 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
25773 Use scalar_float_mode.
25775 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
25777 * config/aarch64/aarch64-simd.md
25778 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
25779 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
25780 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
25781 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
25782 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
25783 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
25784 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
25785 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
25786 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
25787 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
25789 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25792 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
25793 targ_options->x_arm_arch_string is non NULL.
25795 2018-01-11 Tamar Christina <tamar.christina@arm.com>
25797 * config/aarch64/aarch64.h
25798 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
25800 2018-01-11 Sudakshina Das <sudi.das@arm.com>
25803 * expmed.c (emit_store_flag_force): Swap if const op0
25804 and change VOIDmode to mode of op0.
25806 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
25808 PR rtl-optimization/83761
25809 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
25810 than bytes to mode_for_size.
25812 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
25814 PR middle-end/83189
25815 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
25816 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
25819 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
25821 PR middle-end/83575
25822 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
25823 when in layout mode.
25824 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
25825 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
25828 2018-01-10 Michael Collison <michael.collison@arm.com>
25830 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
25831 * config/aarch64/aarch64-option-extension.def: Add
25832 AARCH64_OPT_EXTENSION of 'fp16fml'.
25833 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
25834 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
25835 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
25836 * config/aarch64/constraints.md (Ui7): New constraint.
25837 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
25838 (VFMLA_SEL_W): Ditto.
25841 (VFMLA16_LOW): New int iterator.
25842 (VFMLA16_HIGH): Ditto.
25843 (UNSPEC_FMLAL): New unspec.
25844 (UNSPEC_FMLSL): Ditto.
25845 (UNSPEC_FMLAL2): Ditto.
25846 (UNSPEC_FMLSL2): Ditto.
25847 (f16mac): New code attribute.
25848 * config/aarch64/aarch64-simd-builtins.def
25849 (aarch64_fmlal_lowv2sf): Ditto.
25850 (aarch64_fmlsl_lowv2sf): Ditto.
25851 (aarch64_fmlalq_lowv4sf): Ditto.
25852 (aarch64_fmlslq_lowv4sf): Ditto.
25853 (aarch64_fmlal_highv2sf): Ditto.
25854 (aarch64_fmlsl_highv2sf): Ditto.
25855 (aarch64_fmlalq_highv4sf): Ditto.
25856 (aarch64_fmlslq_highv4sf): Ditto.
25857 (aarch64_fmlal_lane_lowv2sf): Ditto.
25858 (aarch64_fmlsl_lane_lowv2sf): Ditto.
25859 (aarch64_fmlal_laneq_lowv2sf): Ditto.
25860 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
25861 (aarch64_fmlalq_lane_lowv4sf): Ditto.
25862 (aarch64_fmlsl_lane_lowv4sf): Ditto.
25863 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
25864 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
25865 (aarch64_fmlal_lane_highv2sf): Ditto.
25866 (aarch64_fmlsl_lane_highv2sf): Ditto.
25867 (aarch64_fmlal_laneq_highv2sf): Ditto.
25868 (aarch64_fmlsl_laneq_highv2sf): Ditto.
25869 (aarch64_fmlalq_lane_highv4sf): Ditto.
25870 (aarch64_fmlsl_lane_highv4sf): Ditto.
25871 (aarch64_fmlalq_laneq_highv4sf): Ditto.
25872 (aarch64_fmlsl_laneq_highv4sf): Ditto.
25873 * config/aarch64/aarch64-simd.md:
25874 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
25875 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
25876 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
25877 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
25878 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
25879 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
25880 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
25881 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
25882 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
25883 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
25884 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
25885 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
25886 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
25887 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
25888 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
25889 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
25890 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
25891 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
25892 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
25893 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
25894 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
25895 (vfmlsl_low_u32): Ditto.
25896 (vfmlalq_low_u32): Ditto.
25897 (vfmlslq_low_u32): Ditto.
25898 (vfmlal_high_u32): Ditto.
25899 (vfmlsl_high_u32): Ditto.
25900 (vfmlalq_high_u32): Ditto.
25901 (vfmlslq_high_u32): Ditto.
25902 (vfmlal_lane_low_u32): Ditto.
25903 (vfmlsl_lane_low_u32): Ditto.
25904 (vfmlal_laneq_low_u32): Ditto.
25905 (vfmlsl_laneq_low_u32): Ditto.
25906 (vfmlalq_lane_low_u32): Ditto.
25907 (vfmlslq_lane_low_u32): Ditto.
25908 (vfmlalq_laneq_low_u32): Ditto.
25909 (vfmlslq_laneq_low_u32): Ditto.
25910 (vfmlal_lane_high_u32): Ditto.
25911 (vfmlsl_lane_high_u32): Ditto.
25912 (vfmlal_laneq_high_u32): Ditto.
25913 (vfmlsl_laneq_high_u32): Ditto.
25914 (vfmlalq_lane_high_u32): Ditto.
25915 (vfmlslq_lane_high_u32): Ditto.
25916 (vfmlalq_laneq_high_u32): Ditto.
25917 (vfmlslq_laneq_high_u32): Ditto.
25918 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
25919 (AARCH64_FL_FOR_ARCH8_4): New.
25920 (AARCH64_ISA_F16FML): New ISA flag.
25921 (TARGET_F16FML): New feature flag for fp16fml.
25922 (doc/invoke.texi): Document new fp16fml option.
25924 2018-01-10 Michael Collison <michael.collison@arm.com>
25926 * config/aarch64/aarch64-builtins.c:
25927 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
25928 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
25929 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
25930 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
25931 (AARCH64_ISA_SHA3): New ISA flag.
25932 (TARGET_SHA3): New feature flag for sha3.
25933 * config/aarch64/iterators.md (sha512_op): New int attribute.
25934 (CRYPTO_SHA512): New int iterator.
25935 (UNSPEC_SHA512H): New unspec.
25936 (UNSPEC_SHA512H2): Ditto.
25937 (UNSPEC_SHA512SU0): Ditto.
25938 (UNSPEC_SHA512SU1): Ditto.
25939 * config/aarch64/aarch64-simd-builtins.def
25940 (aarch64_crypto_sha512hqv2di): New builtin.
25941 (aarch64_crypto_sha512h2qv2di): Ditto.
25942 (aarch64_crypto_sha512su0qv2di): Ditto.
25943 (aarch64_crypto_sha512su1qv2di): Ditto.
25944 (aarch64_eor3qv8hi): Ditto.
25945 (aarch64_rax1qv2di): Ditto.
25946 (aarch64_xarqv2di): Ditto.
25947 (aarch64_bcaxqv8hi): Ditto.
25948 * config/aarch64/aarch64-simd.md:
25949 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
25950 (aarch64_crypto_sha512su0qv2di): Ditto.
25951 (aarch64_crypto_sha512su1qv2di): Ditto.
25952 (aarch64_eor3qv8hi): Ditto.
25953 (aarch64_rax1qv2di): Ditto.
25954 (aarch64_xarqv2di): Ditto.
25955 (aarch64_bcaxqv8hi): Ditto.
25956 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
25957 (vsha512h2q_u64): Ditto.
25958 (vsha512su0q_u64): Ditto.
25959 (vsha512su1q_u64): Ditto.
25960 (veor3q_u16): Ditto.
25961 (vrax1q_u64): Ditto.
25962 (vxarq_u64): Ditto.
25963 (vbcaxq_u16): Ditto.
25964 * config/arm/types.md (crypto_sha512): New type attribute.
25965 (crypto_sha3): Ditto.
25966 (doc/invoke.texi): Document new sha3 option.
25968 2018-01-10 Michael Collison <michael.collison@arm.com>
25970 * config/aarch64/aarch64-builtins.c:
25971 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
25972 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
25973 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
25974 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
25975 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
25976 (AARCH64_ISA_SM4): New ISA flag.
25977 (TARGET_SM4): New feature flag for sm4.
25978 * config/aarch64/aarch64-simd-builtins.def
25979 (aarch64_sm3ss1qv4si): Ditto.
25980 (aarch64_sm3tt1aq4si): Ditto.
25981 (aarch64_sm3tt1bq4si): Ditto.
25982 (aarch64_sm3tt2aq4si): Ditto.
25983 (aarch64_sm3tt2bq4si): Ditto.
25984 (aarch64_sm3partw1qv4si): Ditto.
25985 (aarch64_sm3partw2qv4si): Ditto.
25986 (aarch64_sm4eqv4si): Ditto.
25987 (aarch64_sm4ekeyqv4si): Ditto.
25988 * config/aarch64/aarch64-simd.md:
25989 (aarch64_sm3ss1qv4si): Ditto.
25990 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
25991 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
25992 (aarch64_sm4eqv4si): Ditto.
25993 (aarch64_sm4ekeyqv4si): Ditto.
25994 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
25995 (sm3part_op): Ditto.
25996 (CRYPTO_SM3TT): Ditto.
25997 (CRYPTO_SM3PART): Ditto.
25998 (UNSPEC_SM3SS1): New unspec.
25999 (UNSPEC_SM3TT1A): Ditto.
26000 (UNSPEC_SM3TT1B): Ditto.
26001 (UNSPEC_SM3TT2A): Ditto.
26002 (UNSPEC_SM3TT2B): Ditto.
26003 (UNSPEC_SM3PARTW1): Ditto.
26004 (UNSPEC_SM3PARTW2): Ditto.
26005 (UNSPEC_SM4E): Ditto.
26006 (UNSPEC_SM4EKEY): Ditto.
26007 * config/aarch64/constraints.md (Ui2): New constraint.
26008 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
26009 * config/arm/types.md (crypto_sm3): New type attribute.
26010 (crypto_sm4): Ditto.
26011 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
26012 (vsm3tt1aq_u32): Ditto.
26013 (vsm3tt1bq_u32): Ditto.
26014 (vsm3tt2aq_u32): Ditto.
26015 (vsm3tt2bq_u32): Ditto.
26016 (vsm3partw1q_u32): Ditto.
26017 (vsm3partw2q_u32): Ditto.
26018 (vsm4eq_u32): Ditto.
26019 (vsm4ekeyq_u32): Ditto.
26020 (doc/invoke.texi): Document new sm4 option.
26022 2018-01-10 Michael Collison <michael.collison@arm.com>
26024 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
26025 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
26026 (AARCH64_FL_FOR_ARCH8_4): New.
26027 (AARCH64_FL_V8_4): New flag.
26028 (doc/invoke.texi): Document new armv8.4-a option.
26030 2018-01-10 Michael Collison <michael.collison@arm.com>
26032 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
26033 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
26034 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
26035 * config/aarch64/aarch64-option-extension.def: Add
26036 AARCH64_OPT_EXTENSION of 'sha2'.
26037 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
26038 (crypto): Disable sha2 and aes if crypto disabled.
26039 (crypto): Enable aes and sha2 if enabled.
26040 (simd): Disable sha2 and aes if simd disabled.
26041 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
26043 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
26044 (TARGET_SHA2): New feature flag for sha2.
26045 (TARGET_AES): New feature flag for aes.
26046 * config/aarch64/aarch64-simd.md:
26047 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
26048 conditional on TARGET_AES.
26049 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
26050 (aarch64_crypto_sha1hsi): Make pattern conditional
26052 (aarch64_crypto_sha1hv4si): Ditto.
26053 (aarch64_be_crypto_sha1hv4si): Ditto.
26054 (aarch64_crypto_sha1su1v4si): Ditto.
26055 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
26056 (aarch64_crypto_sha1su0v4si): Ditto.
26057 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
26058 (aarch64_crypto_sha256su0v4si): Ditto.
26059 (aarch64_crypto_sha256su1v4si): Ditto.
26060 (doc/invoke.texi): Document new aes and sha2 options.
26062 2018-01-10 Martin Sebor <msebor@redhat.com>
26064 PR tree-optimization/83781
26065 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
26068 2018-01-11 Martin Sebor <msebor@gmail.com>
26069 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26071 PR tree-optimization/83501
26072 PR tree-optimization/81703
26074 * tree-ssa-strlen.c (get_string_cst): Rename...
26075 (get_string_len): ...to this. Handle global constants.
26076 (handle_char_store): Adjust.
26078 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
26079 Jim Wilson <jimw@sifive.com>
26081 * config/riscv/riscv-protos.h (riscv_output_return): New.
26082 * config/riscv/riscv.c (struct machine_function): New naked_p field.
26083 (riscv_attribute_table, riscv_output_return),
26084 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
26085 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
26086 (riscv_compute_frame_info): Only compute frame->mask if not a naked
26088 (riscv_expand_prologue): Add early return for naked function.
26089 (riscv_expand_epilogue): Likewise.
26090 (riscv_function_ok_for_sibcall): Return false for naked function.
26091 (riscv_set_current_function): New.
26092 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
26093 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
26094 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
26095 * doc/extend.texi (RISC-V Function Attributes): New.
26097 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
26099 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
26100 check for 128-bit long double before checking TCmode.
26101 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
26102 128-bit long doubles before checking TFmode or TCmode.
26103 (FLOAT128_IBM_P): Likewise.
26105 2018-01-10 Martin Sebor <msebor@redhat.com>
26107 PR tree-optimization/83671
26108 * builtins.c (c_strlen): Unconditionally return zero for the empty
26110 Use -Warray-bounds for warnings.
26111 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
26112 for non-constant array indices with COMPONENT_REF, arrays of
26113 arrays, and pointers to arrays.
26114 (gimple_fold_builtin_strlen): Determine and set length range for
26115 non-constant character arrays.
26117 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
26119 PR middle-end/81897
26120 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
26123 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
26125 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
26127 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
26130 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
26131 VECTOR_MEM_ALTIVEC_OR_VSX_P.
26132 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
26133 indexed_or_indirect_operand predicate.
26134 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
26135 (*vsx_le_perm_load_v8hi): Likewise.
26136 (*vsx_le_perm_load_v16qi): Likewise.
26137 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
26138 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
26139 (*vsx_le_perm_store_v8hi): Likewise.
26140 (*vsx_le_perm_store_v16qi): Likewise.
26141 (eight unnamed splitters): Likewise.
26143 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
26145 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
26146 * config/rs6000/emmintrin.h: Likewise.
26147 * config/rs6000/mmintrin.h: Likewise.
26148 * config/rs6000/xmmintrin.h: Likewise.
26150 2018-01-10 David Malcolm <dmalcolm@redhat.com>
26153 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
26155 * tree.c (tree_nop_conversion): Return true for location wrapper
26157 (maybe_wrap_with_location): New function.
26158 (selftest::check_strip_nops): New function.
26159 (selftest::test_location_wrappers): New function.
26160 (selftest::tree_c_tests): Call it.
26161 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
26162 (maybe_wrap_with_location): New decl.
26163 (EXPR_LOCATION_WRAPPER_P): New macro.
26164 (location_wrapper_p): New inline function.
26165 (tree_strip_any_location_wrapper): New inline function.
26167 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
26170 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
26171 stack_realign_offset for the largest alignment of stack slot
26173 (ix86_find_max_used_stack_alignment): New function.
26174 (ix86_finalize_stack_frame_flags): Use it. Set
26175 max_used_stack_alignment if we don't realign stack.
26176 * config/i386/i386.h (machine_function): Add
26177 max_used_stack_alignment.
26179 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
26181 * config/arm/arm.opt (-mbranch-cost): New option.
26182 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
26185 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
26188 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
26189 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
26191 2018-01-10 Richard Biener <rguenther@suse.de>
26194 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
26195 early out so it also covers the case where we have a non-NULL
26198 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
26200 PR tree-optimization/83753
26201 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
26202 for non-strided grouped accesses if the number of elements is 1.
26204 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
26207 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
26208 * i386.h (TARGET_USE_GATHER): Define.
26209 * x86-tune.def (X86_TUNE_USE_GATHER): New.
26211 2018-01-10 Martin Liska <mliska@suse.cz>
26214 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
26215 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
26217 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
26218 CLEANUP_NO_PARTITIONING is not set.
26220 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
26222 * doc/rtl.texi: Remove documentation of (const ...) wrappers
26223 for vectors, as a partial revert of r254296.
26224 * rtl.h (const_vec_p): Delete.
26225 (const_vec_duplicate_p): Don't test for vector CONSTs.
26226 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
26227 * expmed.c (make_tree): Likewise.
26230 * common.md (E, F): Use CONSTANT_P instead of checking for
26232 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
26233 checking for CONST_VECTOR.
26235 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
26237 PR middle-end/83575
26238 * predict.c (force_edge_cold): Handle in more sane way edges
26239 with no prediction.
26241 2018-01-09 Carl Love <cel@us.ibm.com>
26243 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
26245 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
26246 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
26247 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
26248 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
26249 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
26250 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
26251 * config/rs6000/rs6000-protos.h: Add extern defition for
26252 rs6000_generate_float2_double_code.
26253 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
26255 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
26256 (float2_v2df): Add define_expand.
26258 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
26261 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
26262 op_mode in the force_to_mode call.
26264 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
26266 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
26267 instead of checking each element individually.
26268 (aarch64_evpc_uzp): Likewise.
26269 (aarch64_evpc_zip): Likewise.
26270 (aarch64_evpc_ext): Likewise.
26271 (aarch64_evpc_rev): Likewise.
26272 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
26273 instead of checking each element individually. Return true without
26275 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
26276 whether all selected elements come from the same input, instead of
26277 checking each element individually. Remove calls to gen_rtx_REG,
26278 start_sequence and end_sequence and instead assert that no rtl is
26281 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
26283 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
26284 order of HIGH and CONST checks.
26286 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
26288 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
26289 if the destination isn't an SSA_NAME.
26291 2018-01-09 Richard Biener <rguenther@suse.de>
26293 PR tree-optimization/83668
26294 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
26296 (canonicalize_loop_form): ... here, renamed from ...
26297 (canonicalize_loop_closed_ssa_form): ... this and amended to
26298 swap successor edges for loop exit blocks to make us use
26299 the RPO order we need for initial schedule generation.
26301 2018-01-09 Joseph Myers <joseph@codesourcery.com>
26303 PR tree-optimization/64811
26304 * match.pd: When optimizing comparisons with Inf, avoid
26305 introducing or losing exceptions from comparisons with NaN.
26307 2018-01-09 Martin Liska <mliska@suse.cz>
26310 * asan.c (shadow_mem_size): Add gcc_assert.
26312 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
26314 Don't save registers in main().
26317 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
26318 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
26319 * config/avr/avr.c (avr_set_current_function): Don't error if
26320 naked, OS_task or OS_main are specified at the same time.
26321 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
26323 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
26325 * common/config/avr/avr-common.c (avr_option_optimization_table):
26326 Switch on -mmain-is-OS_task for optimizing compilations.
26328 2018-01-09 Richard Biener <rguenther@suse.de>
26330 PR tree-optimization/83572
26331 * graphite.c: Include cfganal.h.
26332 (graphite_transform_loops): Connect infinite loops to exit
26333 and remove fake edges at the end.
26335 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
26337 * ipa-inline.c (edge_badness): Revert accidental checkin.
26339 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
26342 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
26343 symbols; not inline clones.
26345 2018-01-09 Jakub Jelinek <jakub@redhat.com>
26348 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
26349 hard registers. Formatting fixes.
26351 PR preprocessor/83722
26352 * gcc.c (try_generate_repro): Pass
26353 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
26354 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
26357 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
26358 Kito Cheng <kito.cheng@gmail.com>
26360 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
26361 (riscv_leaf_function_p): Delete.
26362 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
26364 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26366 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
26368 (do_ifelse): New function.
26369 (do_isel): New function.
26370 (do_sub3): New function.
26371 (do_add3): New function.
26372 (do_load_mask_compare): New function.
26373 (do_overlap_load_compare): New function.
26374 (expand_compare_loop): New function.
26375 (expand_block_compare): Call expand_compare_loop() when appropriate.
26376 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
26377 option description.
26378 (-mblock-compare-inline-loop-limit): New option.
26380 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26383 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
26384 Reverse order of second and third operands in first alternative.
26385 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
26386 of first and second elements in UNSPEC_VPERMR vector.
26387 (altivec_expand_vec_perm_le): Likewise.
26389 2018-01-08 Jeff Law <law@redhat.com>
26391 PR rtl-optimizatin/81308
26392 * tree-switch-conversion.c (cfg_altered): New file scoped static.
26393 (process_switch): If group_case_labels makes a change, then set
26395 (pass_convert_switch::execute): If a switch is converted, then
26396 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
26398 PR rtl-optimization/81308
26399 * recog.c (split_all_insns): Conditionally cleanup the CFG after
26402 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
26404 PR target/83663 - Revert r255946
26405 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
26406 generation for cases where splatting a value is not useful.
26407 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
26408 across a vec_duplicate and a paradoxical subreg forming a vector
26409 mode to a vec_concat.
26411 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26413 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
26414 -march=armv8.3-a variants.
26415 * config/arm/t-multilib: Likewise.
26416 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
26418 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26420 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
26422 (cceq_ior_compare_complement): Give it a name so I can use it, and
26423 change boolean_or_operator predicate to boolean_operator so it can
26424 be used to generate a crand.
26425 (eqne): New code iterator.
26426 (bd/bd_neg): New code_attrs.
26427 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
26428 a single define_insn.
26429 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
26430 decrement (bdnzt/bdnzf/bdzt/bdzf).
26431 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
26432 with the new names of the branch decrement patterns, and added the
26433 names of the branch decrement conditional patterns.
26435 2018-01-08 Richard Biener <rguenther@suse.de>
26437 PR tree-optimization/83563
26438 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
26441 2018-01-08 Richard Biener <rguenther@suse.de>
26443 PR middle-end/83713
26444 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
26446 2018-01-08 Richard Biener <rguenther@suse.de>
26448 PR tree-optimization/83685
26449 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
26450 references to abnormals.
26452 2018-01-08 Richard Biener <rguenther@suse.de>
26455 * dwarf2out.c (output_indirect_strings): Handle empty
26456 skeleton_debug_str_hash.
26457 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
26459 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
26461 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
26462 (emit_store_direct): Likewise.
26463 (arc_trampoline_adjust_address): Likewise.
26464 (arc_asm_trampoline_template): New function.
26465 (arc_initialize_trampoline): Use asm_trampoline_template.
26466 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
26467 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
26468 * config/arc/arc.md (flush_icache): Delete pattern.
26470 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
26472 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
26473 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
26476 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
26479 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
26480 by not USED_FOR_TARGET.
26481 (make_pass_resolve_sw_modes): Likewise.
26483 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
26485 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
26488 2018-01-08 Richard Biener <rguenther@suse.de>
26490 PR middle-end/83580
26491 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
26493 2018-01-08 Richard Biener <rguenther@suse.de>
26495 PR middle-end/83517
26496 * match.pd ((t * 2) / 2) -> t): Add missing :c.
26498 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
26500 PR middle-end/81897
26501 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
26502 basic blocks with a small number of successors.
26503 (convert_control_dep_chain_into_preds): Improve handling of
26505 (dump_predicates): Split apart into...
26506 (dump_pred_chain): ...here...
26507 (dump_pred_info): ...and here.
26508 (can_one_predicate_be_invalidated_p): Add debugging printfs.
26509 (can_chain_union_be_invalidated_p): Improve check for invalidation
26511 (uninit_uses_cannot_happen): Avoid unnecessary if
26512 convert_control_dep_chain_into_preds yielded nothing.
26514 2018-01-06 Martin Sebor <msebor@redhat.com>
26516 PR tree-optimization/83640
26517 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
26518 subtracting negative offset from size.
26519 (builtin_access::overlap): Adjust offset bounds of the access to fall
26520 within the size of the object if possible.
26522 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
26524 PR rtl-optimization/83699
26525 * expmed.c (extract_bit_field_1): Restrict the vector usage of
26526 extract_bit_field_as_subreg to cases in which the extracted
26527 value is also a vector.
26529 * lra-constraints.c (process_alt_operands): Test for the equivalence
26530 substitutions when detecting a possible reload cycle.
26532 2018-01-06 Jakub Jelinek <jakub@redhat.com>
26535 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
26536 by default if flag_selective_schedling{,2}. Formatting fixes.
26538 PR rtl-optimization/83682
26539 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
26540 if it has non-VECTOR_MODE element mode.
26541 (vec_duplicate_p): Likewise.
26543 PR middle-end/83694
26544 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
26545 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
26547 2018-01-05 Jakub Jelinek <jakub@redhat.com>
26550 * config/i386/i386-builtin.def
26551 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
26552 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
26553 Require also OPTION_MASK_ISA_AVX512F in addition to
26554 OPTION_MASK_ISA_GFNI.
26555 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
26556 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
26557 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
26558 to OPTION_MASK_ISA_GFNI.
26559 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
26560 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
26561 OPTION_MASK_ISA_AVX512BW.
26562 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
26563 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
26564 addition to OPTION_MASK_ISA_GFNI.
26565 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
26566 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
26567 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
26568 to OPTION_MASK_ISA_GFNI.
26569 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
26570 a requirement for all ISAs rather than any of them with a few
26572 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
26574 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
26575 bitmasks to be enabled with 3 exceptions, instead of requiring any
26576 enabled ISA with lots of exceptions.
26577 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
26578 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
26579 Change avx512bw in isa attribute to avx512f.
26580 * config/i386/sgxintrin.h: Add license boilerplate.
26581 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
26582 to __AVX512F__ and __AVX512VL to __AVX512VL__.
26583 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
26584 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
26586 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
26587 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
26588 temporarily sse2 rather than sse if not enabled already.
26591 * config/i386/sse.md (VI248_VLBW): Rename to ...
26592 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
26593 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
26594 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
26595 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
26596 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
26597 mode iterator instead of VI248_VLBW.
26599 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
26601 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
26602 (record_modified): Skip clobbers; add debug output.
26603 (param_change_prob): Use sreal frequencies.
26605 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
26607 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
26608 punt for user-aligned variables.
26610 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
26612 * tree-chrec.c (chrec_contains_symbols): Return true for
26615 2018-01-05 Sudakshina Das <sudi.das@arm.com>
26618 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
26619 of (x|y) == x for BICS pattern.
26621 2018-01-05 Jakub Jelinek <jakub@redhat.com>
26623 PR tree-optimization/83605
26624 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
26625 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
26628 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
26630 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
26631 * config/epiphany/rtems.h: New file.
26633 2018-01-04 Jakub Jelinek <jakub@redhat.com>
26634 Uros Bizjak <ubizjak@gmail.com>
26637 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
26638 QIreg_operand instead of register_operand predicate.
26639 * config/i386/i386.c (ix86_rop_should_change_byte_p,
26640 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
26641 comments instead of -fmitigate[-_]rop.
26643 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26646 * cgraphunit.c (symbol_table::compile): Switch to text_section
26647 before calling assembly_start debug hook.
26648 * run-rtl-passes.c (run_rtl_passes): Likewise.
26651 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
26653 * tree-vrp.c (extract_range_from_binary_expr_1): Check
26654 range_int_cst_p rather than !symbolic_range_p before calling
26655 extract_range_from_multiplicative_op_1.
26657 2018-01-04 Jeff Law <law@redhat.com>
26659 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
26660 redundant test in assertion.
26662 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
26664 * doc/rtl.texi: Document machine_mode wrapper classes.
26666 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
26668 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
26669 using tree_to_uhwi.
26671 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
26673 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
26674 the VEC_PERM_EXPR fold to fail.
26676 2018-01-04 Jakub Jelinek <jakub@redhat.com>
26679 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
26680 to switched_sections.
26682 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
26685 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
26686 test for d.testing.
26688 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
26691 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
26692 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
26694 2018-01-04 Jakub Jelinek <jakub@redhat.com>
26697 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
26698 is BLKmode and bitpos not zero or mode change is needed.
26700 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
26703 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
26706 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
26709 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
26710 instead of MULT rtx. Update all corresponding splitters.
26711 (*saddl_se): Ditto.
26712 (*ssub<modesuffix>): Ditto.
26713 (*ssubl_se): Ditto.
26714 (*cmp_sadd_di): Update split patterns.
26715 (*cmp_sadd_si): Ditto.
26716 (*cmp_sadd_sidi): Ditto.
26717 (*cmp_ssub_di): Ditto.
26718 (*cmp_ssub_si): Ditto.
26719 (*cmp_ssub_sidi): Ditto.
26720 * config/alpha/predicates.md (const23_operand): New predicate.
26721 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
26722 Look for ASHIFT, not MULT inner operand.
26723 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
26725 2018-01-04 Martin Liska <mliska@suse.cz>
26727 PR gcov-profile/83669
26728 * gcov.c (output_intermediate_file): Add version to intermediate
26730 * doc/gcov.texi: Document new field 'version' in intermediate
26731 file format. Fix location of '-k' option of gcov command.
26733 2018-01-04 Martin Liska <mliska@suse.cz>
26736 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
26738 2018-01-04 Jakub Jelinek <jakub@redhat.com>
26740 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
26742 2018-01-03 Martin Sebor <msebor@redhat.com>
26744 PR tree-optimization/83655
26745 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
26746 checking calls with invalid arguments.
26748 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26750 * tree-vect-stmts.c (vect_get_store_rhs): New function.
26751 (vectorizable_mask_load_store): Delete.
26752 (vectorizable_call): Return false for masked loads and stores.
26753 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
26754 instead of gimple_assign_rhs1.
26755 (vectorizable_load): Handle IFN_MASK_LOAD.
26756 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
26758 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26760 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
26762 (vectorizable_mask_load_store): ...here.
26763 (vectorizable_load): ...and here.
26765 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26767 * tree-vect-stmts.c (vect_build_all_ones_mask)
26768 (vect_build_zero_merge_argument): New functions, split out from...
26769 (vectorizable_load): ...here.
26771 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26773 * tree-vect-stmts.c (vect_check_store_rhs): New function,
26775 (vectorizable_mask_load_store): ...here.
26776 (vectorizable_store): ...and here.
26778 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26780 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
26782 (vectorizable_mask_load_store): ...here.
26784 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26786 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
26787 (vect_model_store_cost): Take a vec_load_store_type instead of a
26789 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
26790 (vect_model_store_cost): Take a vec_load_store_type instead of a
26792 (vectorizable_mask_load_store): Update accordingly.
26793 (vectorizable_store): Likewise.
26794 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
26796 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26798 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
26799 IFN_MASK_LOAD calls here rather than...
26800 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
26802 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26803 Alan Hayward <alan.hayward@arm.com>
26804 David Sherwood <david.sherwood@arm.com>
26806 * expmed.c (extract_bit_field_1): For vector extracts,
26807 fall back to extract_bit_field_as_subreg if vec_extract
26810 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26811 Alan Hayward <alan.hayward@arm.com>
26812 David Sherwood <david.sherwood@arm.com>
26814 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
26815 they are variable or constant sized.
26816 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
26817 slots for constant-sized data.
26819 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26820 Alan Hayward <alan.hayward@arm.com>
26821 David Sherwood <david.sherwood@arm.com>
26823 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
26824 handling COND_EXPRs with boolean comparisons, try to find a better
26825 basis for the mask type than the boolean itself.
26827 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26829 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
26830 is calculated and how it can be overridden.
26831 * genmodes.c (max_bitsize_mode_any_mode): New variable.
26832 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
26834 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
26837 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26838 Alan Hayward <alan.hayward@arm.com>
26839 David Sherwood <david.sherwood@arm.com>
26841 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
26842 Remove the mode argument.
26843 (aarch64_simd_valid_immediate): Remove the mode and inverse
26845 * config/aarch64/iterators.md (bitsize): New iterator.
26846 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
26847 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
26848 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
26849 aarch64_simd_valid_immediate.
26850 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
26851 (aarch64_reg_or_bic_imm): Likewise.
26852 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
26853 with an insn_type enum and msl with a modifier_type enum.
26854 Replace element_width with a scalar_mode. Change the shift
26855 to unsigned int. Add constructors for scalar_float_mode and
26856 scalar_int_mode elements.
26857 (aarch64_vect_float_const_representable_p): Delete.
26858 (aarch64_can_const_movi_rtx_p)
26859 (aarch64_simd_scalar_immediate_valid_for_move)
26860 (aarch64_simd_make_constant): Update call to
26861 aarch64_simd_valid_immediate.
26862 (aarch64_advsimd_valid_immediate_hs): New function.
26863 (aarch64_advsimd_valid_immediate): Likewise.
26864 (aarch64_simd_valid_immediate): Remove mode and inverse
26865 arguments. Rewrite to use the above. Use const_vec_duplicate_p
26866 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
26867 and aarch64_float_const_representable_p on the result.
26868 (aarch64_output_simd_mov_immediate): Remove mode argument.
26869 Update call to aarch64_simd_valid_immediate and use of
26870 simd_immediate_info.
26871 (aarch64_output_scalar_simd_mov_immediate): Update call
26874 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26875 Alan Hayward <alan.hayward@arm.com>
26876 David Sherwood <david.sherwood@arm.com>
26878 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
26879 (mode_nunits): Likewise CONST_MODE_NUNITS.
26880 * machmode.def (ADJUST_NUNITS): Document.
26881 * genmodes.c (mode_data::need_nunits_adj): New field.
26882 (blank_mode): Update accordingly.
26883 (adj_nunits): New variable.
26884 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
26886 (emit_mode_size_inline): Set need_bytesize_adj for all modes
26887 listed in adj_nunits.
26888 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
26889 listed in adj_nunits. Don't emit case statements for such modes.
26890 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
26891 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
26892 nothing if adj_nunits is nonnull.
26893 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
26894 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
26895 (emit_mode_fbit): Update use of print_maybe_const_decl.
26896 (emit_move_size): Likewise. Treat the array as non-const
26898 (emit_mode_adjustments): Handle adj_nunits.
26900 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26902 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
26903 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
26904 (VECTOR_MODES): Use it.
26905 (make_vector_modes): Take the prefix as an argument.
26907 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26908 Alan Hayward <alan.hayward@arm.com>
26909 David Sherwood <david.sherwood@arm.com>
26911 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
26912 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
26913 for MODE_VECTOR_BOOL.
26914 * machmode.def (VECTOR_BOOL_MODE): Document.
26915 * genmodes.c (VECTOR_BOOL_MODE): New macro.
26916 (make_vector_bool_mode): New function.
26917 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
26919 * lto-streamer-in.c (lto_input_mode_table): Likewise.
26920 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
26922 * stor-layout.c (int_mode_for_mode): Likewise.
26923 * tree.c (build_vector_type_for_mode): Likewise.
26924 * varasm.c (output_constant_pool_2): Likewise.
26925 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
26926 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
26927 for MODE_VECTOR_BOOL.
26928 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
26929 of mode class checks.
26930 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
26931 instead of a list of mode class checks.
26932 (expand_vector_scalar_condition): Likewise.
26933 (type_for_widest_vector_mode): Handle BImode as an inner mode.
26935 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26936 Alan Hayward <alan.hayward@arm.com>
26937 David Sherwood <david.sherwood@arm.com>
26939 * machmode.h (mode_size): Change from unsigned short to
26941 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
26942 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
26943 or if measurement_type is not polynomial.
26944 (fixed_size_mode::includes_p): Check for constant-sized modes.
26945 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
26946 return a poly_uint16 rather than an unsigned short.
26947 (emit_mode_size): Change the type of mode_size from unsigned short
26948 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
26949 (emit_mode_adjustments): Cope with polynomial vector sizes.
26950 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
26952 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
26954 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
26955 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
26956 * caller-save.c (setup_save_areas): Likewise.
26957 (replace_reg_with_saved_mem): Likewise.
26958 * calls.c (emit_library_call_value_1): Likewise.
26959 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
26960 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
26961 (gen_lowpart_for_combine): Likewise.
26962 * convert.c (convert_to_integer_1): Likewise.
26963 * cse.c (equiv_constant, cse_insn): Likewise.
26964 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
26965 (cselib_subst_to_values): Likewise.
26966 * dce.c (word_dce_process_block): Likewise.
26967 * df-problems.c (df_word_lr_mark_ref): Likewise.
26968 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
26969 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
26970 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
26971 (rtl_for_decl_location): Likewise.
26972 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
26973 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
26974 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
26975 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
26976 (expand_expr_real_1): Likewise.
26977 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
26978 (pad_below): Likewise.
26979 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
26980 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
26981 * ira.c (get_subreg_tracking_sizes): Likewise.
26982 * ira-build.c (ira_create_allocno_objects): Likewise.
26983 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
26984 (ira_sort_regnos_for_alter_reg): Likewise.
26985 * ira-costs.c (record_operand_costs): Likewise.
26986 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
26987 (resolve_simple_move): Likewise.
26988 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
26989 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
26990 (lra_constraints): Likewise.
26991 (CONST_POOL_OK_P): Reject variable-sized modes.
26992 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
26993 (add_pseudo_to_slot, lra_spill): Likewise.
26994 * omp-low.c (omp_clause_aligned_alignment): Likewise.
26995 * optabs-query.c (get_best_extraction_insn): Likewise.
26996 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
26997 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
26998 (expand_mult_highpart, valid_multiword_target_p): Likewise.
26999 * recog.c (offsettable_address_addr_space_p): Likewise.
27000 * regcprop.c (maybe_mode_change): Likewise.
27001 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
27002 * regrename.c (build_def_use): Likewise.
27003 * regstat.c (dump_reg_info): Likewise.
27004 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
27005 (find_reloads, find_reloads_subreg_address): Likewise.
27006 * reload1.c (eliminate_regs_1): Likewise.
27007 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
27008 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
27009 (simplify_binary_operation_1, simplify_subreg): Likewise.
27010 * targhooks.c (default_function_arg_padding): Likewise.
27011 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
27012 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
27013 (verify_gimple_assign_ternary): Likewise.
27014 * tree-inline.c (estimate_move_cost): Likewise.
27015 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
27016 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
27017 (get_address_cost_ainc): Likewise.
27018 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
27019 (vect_supportable_dr_alignment): Likewise.
27020 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
27021 (vectorizable_reduction): Likewise.
27022 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
27023 (vectorizable_operation, vectorizable_load): Likewise.
27024 * tree.c (build_same_sized_truth_vector_type): Likewise.
27025 * valtrack.c (cleanup_auto_inc_dec): Likewise.
27026 * var-tracking.c (emit_note_insn_var_location): Likewise.
27027 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
27028 (ADDR_VEC_ALIGN): Likewise.
27030 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27031 Alan Hayward <alan.hayward@arm.com>
27032 David Sherwood <david.sherwood@arm.com>
27034 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
27036 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
27037 or if measurement_type is polynomial.
27038 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
27039 * combine.c (make_extraction): Likewise.
27040 * dse.c (find_shift_sequence): Likewise.
27041 * dwarf2out.c (mem_loc_descriptor): Likewise.
27042 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
27043 (extract_bit_field, extract_low_bits): Likewise.
27044 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
27045 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
27046 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
27047 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
27048 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
27049 * reload.c (find_reloads): Likewise.
27050 * reload1.c (alter_reg): Likewise.
27051 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
27052 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
27053 * tree-if-conv.c (predicate_mem_writes): Likewise.
27054 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
27055 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
27056 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
27057 * valtrack.c (dead_debug_insert_temp): Likewise.
27058 * varasm.c (mergeable_constant_section): Likewise.
27059 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
27061 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27062 Alan Hayward <alan.hayward@arm.com>
27063 David Sherwood <david.sherwood@arm.com>
27065 * expr.c (expand_assignment): Cope with polynomial mode sizes
27066 when assigning to a CONCAT.
27068 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27069 Alan Hayward <alan.hayward@arm.com>
27070 David Sherwood <david.sherwood@arm.com>
27072 * machmode.h (mode_precision): Change from unsigned short to
27074 (mode_to_precision): Return a poly_uint16 rather than an unsigned
27076 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
27077 or if measurement_type is not polynomial.
27078 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
27079 in which the mode is already known to be a scalar_int_mode.
27080 * genmodes.c (emit_mode_precision): Change the type of mode_precision
27081 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
27083 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
27084 for GET_MODE_PRECISION.
27085 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
27086 for GET_MODE_PRECISION.
27087 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
27089 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
27090 (expand_field_assignment, make_extraction): Likewise.
27091 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
27092 (get_last_value): Likewise.
27093 * convert.c (convert_to_integer_1): Likewise.
27094 * cse.c (cse_insn): Likewise.
27095 * expr.c (expand_expr_real_1): Likewise.
27096 * lra-constraints.c (simplify_operand_subreg): Likewise.
27097 * optabs-query.c (can_atomic_load_p): Likewise.
27098 * optabs.c (expand_atomic_load): Likewise.
27099 (expand_atomic_store): Likewise.
27100 * ree.c (combine_reaching_defs): Likewise.
27101 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
27102 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
27103 * tree.h (type_has_mode_precision_p): Likewise.
27104 * ubsan.c (instrument_si_overflow): Likewise.
27106 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27107 Alan Hayward <alan.hayward@arm.com>
27108 David Sherwood <david.sherwood@arm.com>
27110 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
27111 polynomial numbers of units.
27112 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
27113 (valid_vector_subparts_p): New function.
27114 (build_vector_type): Remove temporary shim and take the number
27115 of units as a poly_uint64 rather than an int.
27116 (build_opaque_vector_type): Take the number of units as a
27117 poly_uint64 rather than an int.
27118 * tree.c (build_vector_from_ctor): Handle polynomial
27119 TYPE_VECTOR_SUBPARTS.
27120 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
27121 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
27122 (build_vector_from_val): If the number of units is variable,
27123 use build_vec_duplicate_cst for constant operands and
27124 VEC_DUPLICATE_EXPR otherwise.
27125 (make_vector_type): Remove temporary is_constant ().
27126 (build_vector_type, build_opaque_vector_type): Take the number of
27127 units as a poly_uint64 rather than an int.
27128 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
27130 * cfgexpand.c (expand_debug_expr): Likewise.
27131 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
27132 (store_constructor, expand_expr_real_1): Likewise.
27133 (const_scalar_mask_from_tree): Likewise.
27134 * fold-const-call.c (fold_const_reduction): Likewise.
27135 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
27136 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
27137 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
27138 (fold_relational_const): Likewise.
27139 (native_interpret_vector): Likewise. Change the size from an
27140 int to an unsigned int.
27141 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
27142 TYPE_VECTOR_SUBPARTS.
27143 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
27144 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
27145 duplicating a non-constant operand into a variable-length vector.
27146 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
27147 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
27148 * ipa-icf.c (sem_variable::equals): Likewise.
27149 * match.pd: Likewise.
27150 * omp-simd-clone.c (simd_clone_subparts): Likewise.
27151 * print-tree.c (print_node): Likewise.
27152 * stor-layout.c (layout_type): Likewise.
27153 * targhooks.c (default_builtin_vectorization_cost): Likewise.
27154 * tree-cfg.c (verify_gimple_comparison): Likewise.
27155 (verify_gimple_assign_binary): Likewise.
27156 (verify_gimple_assign_ternary): Likewise.
27157 (verify_gimple_assign_single): Likewise.
27158 * tree-pretty-print.c (dump_generic_node): Likewise.
27159 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
27160 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
27161 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
27162 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
27163 (vect_shift_permute_load_chain): Likewise.
27164 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
27165 (expand_vector_condition, optimize_vector_constructor): Likewise.
27166 (lower_vec_perm, get_compute_type): Likewise.
27167 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
27168 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
27169 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
27170 (vect_recog_mask_conversion_pattern): Likewise.
27171 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
27172 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
27173 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
27174 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
27175 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
27176 (vectorizable_shift, vectorizable_operation, vectorizable_store)
27177 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
27178 (supportable_widening_operation): Likewise.
27179 (supportable_narrowing_operation): Likewise.
27180 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
27182 * varasm.c (output_constant): Likewise.
27184 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27185 Alan Hayward <alan.hayward@arm.com>
27186 David Sherwood <david.sherwood@arm.com>
27188 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
27189 so that both the length == 3 and length != 3 cases set up their
27190 own permute vectors. Add comments explaining why we know the
27191 number of elements is constant.
27192 (vect_permute_load_chain): Likewise.
27194 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27195 Alan Hayward <alan.hayward@arm.com>
27196 David Sherwood <david.sherwood@arm.com>
27198 * machmode.h (mode_nunits): Change from unsigned char to
27200 (ONLY_FIXED_SIZE_MODES): New macro.
27201 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
27202 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
27203 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
27205 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
27206 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
27207 or if measurement_type is not polynomial.
27208 * genmodes.c (ZERO_COEFFS): New macro.
27209 (emit_mode_nunits_inline): Make mode_nunits_inline return a
27211 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
27212 Use ZERO_COEFFS when emitting initializers.
27213 * data-streamer.h (bp_pack_poly_value): New function.
27214 (bp_unpack_poly_value): Likewise.
27215 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
27216 for GET_MODE_NUNITS.
27217 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
27218 for GET_MODE_NUNITS.
27219 * tree.c (make_vector_type): Remove temporary shim and make
27220 the real function take the number of units as a poly_uint64
27221 rather than an int.
27222 (build_vector_type_for_mode): Handle polynomial nunits.
27223 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
27224 * emit-rtl.c (const_vec_series_p_1): Likewise.
27225 (gen_rtx_CONST_VECTOR): Likewise.
27226 * fold-const.c (test_vec_duplicate_folding): Likewise.
27227 * genrecog.c (validate_pattern): Likewise.
27228 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
27229 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
27230 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
27231 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
27232 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
27233 * rtlanal.c (subreg_get_info): Likewise.
27234 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
27235 (vect_grouped_load_supported): Likewise.
27236 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
27237 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
27238 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
27239 (simplify_const_unary_operation, simplify_binary_operation_1)
27240 (simplify_const_binary_operation, simplify_ternary_operation)
27241 (test_vector_ops_duplicate, test_vector_ops): Likewise.
27242 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
27243 instead of CONST_VECTOR_NUNITS.
27244 * varasm.c (output_constant_pool_2): Likewise.
27245 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
27246 explicit-encoded elements in the XVEC for variable-length vectors.
27248 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27250 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
27252 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27253 Alan Hayward <alan.hayward@arm.com>
27254 David Sherwood <david.sherwood@arm.com>
27256 * coretypes.h (fixed_size_mode): Declare.
27257 (fixed_size_mode_pod): New typedef.
27258 * builtins.h (target_builtins::x_apply_args_mode)
27259 (target_builtins::x_apply_result_mode): Change type to
27260 fixed_size_mode_pod.
27261 * builtins.c (apply_args_size, apply_result_size, result_vector)
27262 (expand_builtin_apply_args_1, expand_builtin_apply)
27263 (expand_builtin_return): Update accordingly.
27265 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27267 * cse.c (hash_rtx_cb): Hash only the encoded elements.
27268 * cselib.c (cselib_hash_rtx): Likewise.
27269 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
27270 CONST_VECTOR encoding.
27272 2018-01-03 Jakub Jelinek <jakub@redhat.com>
27273 Jeff Law <law@redhat.com>
27276 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
27277 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
27278 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
27279 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
27282 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
27283 explicitly probe *sp in a noreturn function if there were any callee
27284 register saves or frame pointer is needed.
27286 2018-01-03 Jakub Jelinek <jakub@redhat.com>
27289 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
27290 BLKmode for ternary, binary or unary expressions.
27293 * var-tracking.c (delete_vta_debug_insn): New inline function.
27294 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
27295 insns from get_insns () to NULL instead of each bb separately.
27296 Use delete_vta_debug_insn. No longer static.
27297 (vt_debug_insns_local, variable_tracking_main_1): Adjust
27298 delete_vta_debug_insns callers.
27299 * rtl.h (delete_vta_debug_insns): Declare.
27300 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
27301 instead of variable_tracking_main.
27303 2018-01-03 Martin Sebor <msebor@redhat.com>
27305 PR tree-optimization/83603
27306 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
27307 arguments past the endof the argument list in functions declared
27308 without a prototype.
27309 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
27310 Avoid checking when arguments are null.
27312 2018-01-03 Martin Sebor <msebor@redhat.com>
27315 * doc/extend.texi (attribute const): Fix a typo.
27316 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
27317 issuing -Wsuggest-attribute for void functions.
27319 2018-01-03 Martin Sebor <msebor@redhat.com>
27321 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
27322 offset_int::from instead of wide_int::to_shwi.
27323 (maybe_diag_overlap): Remove assertion.
27324 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
27325 * gimple-ssa-sprintf.c (format_directive): Same.
27326 (parse_directive): Same.
27327 (sprintf_dom_walker::compute_format_length): Same.
27328 (try_substitute_return_value): Same.
27330 2018-01-03 Jeff Law <law@redhat.com>
27332 PR middle-end/83654
27333 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
27334 non-constant residual for zero at runtime and avoid probing in
27335 that case. Reorganize code for trailing problem to mirror handling
27338 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27340 PR tree-optimization/83501
27341 * tree-ssa-strlen.c (get_string_cst): New.
27342 (handle_char_store): Call get_string_cst.
27344 2018-01-03 Martin Liska <mliska@suse.cz>
27346 PR tree-optimization/83593
27347 * tree-ssa-strlen.c: Include tree-cfg.h.
27348 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
27349 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
27350 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
27352 (strlen_dom_walker::before_dom_children): Call
27353 gimple_purge_dead_eh_edges. Dump tranformation with details
27355 (strlen_dom_walker::before_dom_children): Update call by adding
27356 new argument cleanup_eh.
27357 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
27359 2018-01-03 Martin Liska <mliska@suse.cz>
27362 * cif-code.def (VARIADIC_THUNK): New enum value.
27363 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
27366 2018-01-03 Jan Beulich <jbeulich@suse.com>
27368 * sse.md (mov<mode>_internal): Tighten condition for when to use
27369 vmovdqu<ssescalarsize> for TI and OI modes.
27371 2018-01-03 Jakub Jelinek <jakub@redhat.com>
27373 Update copyright years.
27375 2018-01-03 Martin Liska <mliska@suse.cz>
27378 * ipa-visibility.c (function_and_variable_visibility): Skip
27379 functions with noipa attribure.
27381 2018-01-03 Jakub Jelinek <jakub@redhat.com>
27383 * gcc.c (process_command): Update copyright notice dates.
27384 * gcov-dump.c (print_version): Ditto.
27385 * gcov.c (print_version): Ditto.
27386 * gcov-tool.c (print_version): Ditto.
27387 * gengtype.c (create_file): Ditto.
27388 * doc/cpp.texi: Bump @copying's copyright year.
27389 * doc/cppinternals.texi: Ditto.
27390 * doc/gcc.texi: Ditto.
27391 * doc/gccint.texi: Ditto.
27392 * doc/gcov.texi: Ditto.
27393 * doc/install.texi: Ditto.
27394 * doc/invoke.texi: Ditto.
27396 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27398 * vector-builder.h (vector_builder::m_full_nelts): Change from
27399 unsigned int to poly_uint64.
27400 (vector_builder::full_nelts): Update prototype accordingly.
27401 (vector_builder::new_vector): Likewise.
27402 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
27403 (vector_builder::operator ==): Likewise.
27404 (vector_builder::finalize): Likewise.
27405 * int-vector-builder.h (int_vector_builder::int_vector_builder):
27406 Take the number of elements as a poly_uint64 rather than an
27408 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
27409 from unsigned int to poly_uint64.
27410 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
27411 (vec_perm_indices::new_vector): Likewise.
27412 (vec_perm_indices::length): Likewise.
27413 (vec_perm_indices::nelts_per_input): Likewise.
27414 (vec_perm_indices::input_nelts): Likewise.
27415 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
27416 number of elements per input as a poly_uint64 rather than an
27417 unsigned int. Use the original encoding for variable-length
27418 vectors, rather than clamping each individual element.
27419 For the second and subsequent elements in each pattern,
27420 clamp the step and base before clamping their sum.
27421 (vec_perm_indices::series_p): Handle polynomial element counts.
27422 (vec_perm_indices::all_in_range_p): Likewise.
27423 (vec_perm_indices_to_tree): Likewise.
27424 (vec_perm_indices_to_rtx): Likewise.
27425 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
27426 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
27427 (tree_vector_builder::new_binary_operation): Handle polynomial
27428 element counts. Return false if we need to know the number
27429 of elements at compile time.
27430 * fold-const.c (fold_vec_perm): Punt if the number of elements
27431 isn't known at compile time.
27433 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27435 * vec-perm-indices.h (vec_perm_builder): Change element type
27436 from HOST_WIDE_INT to poly_int64.
27437 (vec_perm_indices::element_type): Update accordingly.
27438 (vec_perm_indices::clamp): Handle polynomial element_types.
27439 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
27440 (vec_perm_indices::all_in_range_p): Likewise.
27441 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
27443 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
27444 polynomial vec_perm_indices element types.
27445 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
27446 * fold-const.c (fold_vec_perm): Likewise.
27447 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
27448 * tree-vect-generic.c (lower_vec_perm): Likewise.
27449 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
27450 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
27451 element type to HOST_WIDE_INT.
27453 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27454 Alan Hayward <alan.hayward@arm.com>
27455 David Sherwood <david.sherwood@arm.com>
27457 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
27458 rather than an int. Use plus_constant.
27459 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
27460 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
27462 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27463 Alan Hayward <alan.hayward@arm.com>
27464 David Sherwood <david.sherwood@arm.com>
27466 * calls.c (emit_call_1, expand_call): Change struct_value_size from
27467 a HOST_WIDE_INT to a poly_int64.
27469 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27470 Alan Hayward <alan.hayward@arm.com>
27471 David Sherwood <david.sherwood@arm.com>
27473 * calls.c (load_register_parameters): Cope with polynomial
27474 mode sizes. Require a constant size for BLKmode parameters
27475 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
27476 forces a parameter to be padded at the lsb end in order to
27477 fill a complete number of words, require the parameter size
27478 to be ordered wrt UNITS_PER_WORD.
27480 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27481 Alan Hayward <alan.hayward@arm.com>
27482 David Sherwood <david.sherwood@arm.com>
27484 * reload1.c (spill_stack_slot_width): Change element type
27485 from unsigned int to poly_uint64_pod.
27486 (alter_reg): Treat mode sizes as polynomial.
27488 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27489 Alan Hayward <alan.hayward@arm.com>
27490 David Sherwood <david.sherwood@arm.com>
27492 * reload.c (complex_word_subreg_p): New function.
27493 (reload_inner_reg_of_subreg, push_reload): Use it.
27495 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27496 Alan Hayward <alan.hayward@arm.com>
27497 David Sherwood <david.sherwood@arm.com>
27499 * lra-constraints.c (process_alt_operands): Reject matched
27500 operands whose sizes aren't ordered.
27501 (match_reload): Refer to this check here.
27503 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27504 Alan Hayward <alan.hayward@arm.com>
27505 David Sherwood <david.sherwood@arm.com>
27507 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
27508 that the mode size is in the set {1, 2, 4, 8, 16}.
27510 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27511 Alan Hayward <alan.hayward@arm.com>
27512 David Sherwood <david.sherwood@arm.com>
27514 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
27515 Use plus_constant instead of gen_rtx_PLUS.
27517 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27518 Alan Hayward <alan.hayward@arm.com>
27519 David Sherwood <david.sherwood@arm.com>
27521 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
27522 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
27523 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
27524 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
27525 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
27526 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
27527 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
27528 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
27529 * config/i386/i386.c (ix86_push_rounding): ...this new function.
27530 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
27532 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
27533 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
27534 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
27535 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
27536 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
27537 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
27538 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
27539 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
27540 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
27541 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
27543 * expr.c (emit_move_resolve_push): Treat the input and result
27544 of PUSH_ROUNDING as a poly_int64.
27545 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
27546 (emit_push_insn): Likewise.
27547 * lra-eliminations.c (mark_not_eliminable): Likewise.
27548 * recog.c (push_operand): Likewise.
27549 * reload1.c (elimination_effects): Likewise.
27550 * rtlanal.c (nonzero_bits1): Likewise.
27551 * calls.c (store_one_arg): Likewise. Require the padding to be
27552 known at compile time.
27554 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27555 Alan Hayward <alan.hayward@arm.com>
27556 David Sherwood <david.sherwood@arm.com>
27558 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
27559 Use plus_constant instead of gen_rtx_PLUS.
27561 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27562 Alan Hayward <alan.hayward@arm.com>
27563 David Sherwood <david.sherwood@arm.com>
27565 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
27566 rather than an int.
27568 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27569 Alan Hayward <alan.hayward@arm.com>
27570 David Sherwood <david.sherwood@arm.com>
27572 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
27573 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
27574 via stack temporaries. Treat the mode size as polynomial too.
27576 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27577 Alan Hayward <alan.hayward@arm.com>
27578 David Sherwood <david.sherwood@arm.com>
27580 * expr.c (expand_expr_real_2): When handling conversions involving
27581 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
27582 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
27583 as a poly_uint64 too.
27585 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27586 Alan Hayward <alan.hayward@arm.com>
27587 David Sherwood <david.sherwood@arm.com>
27589 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
27591 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27592 Alan Hayward <alan.hayward@arm.com>
27593 David Sherwood <david.sherwood@arm.com>
27595 * combine.c (can_change_dest_mode): Handle polynomial
27596 REGMODE_NATURAL_SIZE.
27597 * expmed.c (store_bit_field_1): Likewise.
27598 * expr.c (store_constructor): Likewise.
27599 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
27600 and polynomial REGMODE_NATURAL_SIZE.
27601 (gen_lowpart_common): Likewise.
27602 * reginfo.c (record_subregs_of_mode): Likewise.
27603 * rtlanal.c (read_modify_subreg_p): Likewise.
27605 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27606 Alan Hayward <alan.hayward@arm.com>
27607 David Sherwood <david.sherwood@arm.com>
27609 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
27610 numbers of elements.
27612 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27613 Alan Hayward <alan.hayward@arm.com>
27614 David Sherwood <david.sherwood@arm.com>
27616 * match.pd: Cope with polynomial numbers of vector elements.
27618 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27619 Alan Hayward <alan.hayward@arm.com>
27620 David Sherwood <david.sherwood@arm.com>
27622 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
27623 in a POINTER_PLUS_EXPR.
27625 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27626 Alan Hayward <alan.hayward@arm.com>
27627 David Sherwood <david.sherwood@arm.com>
27629 * omp-simd-clone.c (simd_clone_subparts): New function.
27630 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
27631 (ipa_simd_modify_function_body): Likewise.
27633 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27634 Alan Hayward <alan.hayward@arm.com>
27635 David Sherwood <david.sherwood@arm.com>
27637 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
27638 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
27639 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
27640 (expand_vector_condition, vector_element): Likewise.
27641 (subparts_gt): New function.
27642 (get_compute_type): Use subparts_gt.
27643 (count_type_subparts): Delete.
27644 (expand_vector_operations_1): Use subparts_gt instead of
27645 count_type_subparts.
27647 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27648 Alan Hayward <alan.hayward@arm.com>
27649 David Sherwood <david.sherwood@arm.com>
27651 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
27652 (vect_compile_time_alias): ...this new function. Do the calculation
27653 on poly_ints rather than trees.
27654 (vect_prune_runtime_alias_test_list): Update call accordingly.
27656 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27657 Alan Hayward <alan.hayward@arm.com>
27658 David Sherwood <david.sherwood@arm.com>
27660 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
27662 (vect_schedule_slp_instance): Likewise.
27664 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27665 Alan Hayward <alan.hayward@arm.com>
27666 David Sherwood <david.sherwood@arm.com>
27668 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
27669 constant and extern definitions for variable-length vectors.
27670 (vect_get_constant_vectors): Note that the number of units
27671 is known to be constant.
27673 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27674 Alan Hayward <alan.hayward@arm.com>
27675 David Sherwood <david.sherwood@arm.com>
27677 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
27678 of units as polynomial. Choose between WIDE and NARROW based
27681 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27682 Alan Hayward <alan.hayward@arm.com>
27683 David Sherwood <david.sherwood@arm.com>
27685 * tree-vect-stmts.c (simd_clone_subparts): New function.
27686 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
27688 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27689 Alan Hayward <alan.hayward@arm.com>
27690 David Sherwood <david.sherwood@arm.com>
27692 * tree-vect-stmts.c (vectorizable_call): Treat the number of
27693 vectors as polynomial. Use build_index_vector for
27694 IFN_GOMP_SIMD_LANE.
27696 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27697 Alan Hayward <alan.hayward@arm.com>
27698 David Sherwood <david.sherwood@arm.com>
27700 * tree-vect-stmts.c (get_load_store_type): Treat the number of
27701 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
27702 for variable-length vectors.
27703 (vectorizable_mask_load_store): Treat the number of units as
27704 polynomial, asserting that it is constant if the condition has
27705 already been enforced.
27706 (vectorizable_store, vectorizable_load): Likewise.
27708 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27709 Alan Hayward <alan.hayward@arm.com>
27710 David Sherwood <david.sherwood@arm.com>
27712 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
27713 of units as polynomial. Punt if we can't tell at compile time
27714 which vector contains the final result.
27716 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27717 Alan Hayward <alan.hayward@arm.com>
27718 David Sherwood <david.sherwood@arm.com>
27720 * tree-vect-loop.c (vectorizable_induction): Treat the number
27721 of units as polynomial. Punt on SLP inductions. Use an integer
27722 VEC_SERIES_EXPR for variable-length integer reductions. Use a
27723 cast of such a series for variable-length floating-point
27726 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27727 Alan Hayward <alan.hayward@arm.com>
27728 David Sherwood <david.sherwood@arm.com>
27730 * tree.h (build_index_vector): Declare.
27731 * tree.c (build_index_vector): New function.
27732 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
27733 of units as polynomial, forcibly converting it to a constant if
27734 vectorizable_reduction has already enforced the condition.
27735 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
27736 to create a {1,2,3,...} vector.
27737 (vectorizable_reduction): Treat the number of units as polynomial.
27738 Choose vectype_in based on the largest scalar element size rather
27739 than the smallest number of units. Enforce the restrictions
27742 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27743 Alan Hayward <alan.hayward@arm.com>
27744 David Sherwood <david.sherwood@arm.com>
27746 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
27747 number of units as polynomial.
27749 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27750 Alan Hayward <alan.hayward@arm.com>
27751 David Sherwood <david.sherwood@arm.com>
27753 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
27754 * target.def (autovectorize_vector_sizes): Return the vector sizes
27755 by pointer, using vector_sizes rather than a bitmask.
27756 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
27757 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
27758 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
27760 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
27761 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
27762 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
27763 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
27764 * omp-general.c (omp_max_vf): Likewise.
27765 * omp-low.c (omp_clause_aligned_alignment): Likewise.
27766 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
27767 * tree-vect-loop.c (vect_analyze_loop): Likewise.
27768 * tree-vect-slp.c (vect_slp_bb): Likewise.
27769 * doc/tm.texi: Regenerate.
27770 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
27772 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
27773 the vector size as a poly_uint64 rather than an unsigned int.
27774 (current_vector_size): Change from an unsigned int to a poly_uint64.
27775 (get_vectype_for_scalar_type): Update accordingly.
27776 * tree.h (build_truth_vector_type): Take the size and number of
27777 units as a poly_uint64 rather than an unsigned int.
27778 (build_vector_type): Add a temporary overload that takes
27779 the number of units as a poly_uint64 rather than an unsigned int.
27780 * tree.c (make_vector_type): Likewise.
27781 (build_truth_vector_type): Take the number of units as a poly_uint64
27782 rather than an unsigned int.
27784 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27785 Alan Hayward <alan.hayward@arm.com>
27786 David Sherwood <david.sherwood@arm.com>
27788 * target.def (get_mask_mode): Take the number of units and length
27789 as poly_uint64s rather than unsigned ints.
27790 * targhooks.h (default_get_mask_mode): Update accordingly.
27791 * targhooks.c (default_get_mask_mode): Likewise.
27792 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
27793 * doc/tm.texi: Regenerate.
27795 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27796 Alan Hayward <alan.hayward@arm.com>
27797 David Sherwood <david.sherwood@arm.com>
27799 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
27800 * omp-general.c (omp_max_vf): Likewise.
27801 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
27802 (expand_omp_simd): Handle polynomial safelen.
27803 * omp-low.c (omplow_simd_context): Add a default constructor.
27804 (omplow_simd_context::max_vf): Change from int to poly_uint64.
27805 (lower_rec_simd_input_clauses): Update accordingly.
27806 (lower_rec_input_clauses): Likewise.
27808 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27809 Alan Hayward <alan.hayward@arm.com>
27810 David Sherwood <david.sherwood@arm.com>
27812 * tree-vectorizer.h (vect_nunits_for_cost): New function.
27813 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
27814 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
27815 (vect_analyze_slp_cost): Likewise.
27816 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
27817 (vect_model_load_cost): Likewise.
27819 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27820 Alan Hayward <alan.hayward@arm.com>
27821 David Sherwood <david.sherwood@arm.com>
27823 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
27824 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
27825 from an unsigned int * to a poly_uint64_pod *.
27826 (calculate_unrolling_factor): New function.
27827 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
27829 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27830 Alan Hayward <alan.hayward@arm.com>
27831 David Sherwood <david.sherwood@arm.com>
27833 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
27834 from an unsigned int to a poly_uint64.
27835 (_loop_vec_info::slp_unrolling_factor): Likewise.
27836 (_loop_vec_info::vectorization_factor): Change from an int
27838 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
27839 (vect_get_num_vectors): New function.
27840 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
27841 (vect_get_num_copies): Use vect_get_num_vectors.
27842 (vect_analyze_data_ref_dependences): Change max_vf from an int *
27843 to an unsigned int *.
27844 (vect_analyze_data_refs): Change min_vf from an int * to a
27846 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
27847 than an unsigned HOST_WIDE_INT.
27848 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
27849 (vect_analyze_data_ref_dependence): Change max_vf from an int *
27850 to an unsigned int *.
27851 (vect_analyze_data_ref_dependences): Likewise.
27852 (vect_compute_data_ref_alignment): Handle polynomial vf.
27853 (vect_enhance_data_refs_alignment): Likewise.
27854 (vect_prune_runtime_alias_test_list): Likewise.
27855 (vect_shift_permute_load_chain): Likewise.
27856 (vect_supportable_dr_alignment): Likewise.
27857 (dependence_distance_ge_vf): Take the vectorization factor as a
27858 poly_uint64 rather than an unsigned HOST_WIDE_INT.
27859 (vect_analyze_data_refs): Change min_vf from an int * to a
27861 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
27862 vfm1 as a poly_uint64 rather than an int. Make the same change
27863 for the returned bound_scalar.
27864 (vect_gen_vector_loop_niters): Handle polynomial vf.
27865 (vect_do_peeling): Likewise. Update call to
27866 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
27867 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
27869 * tree-vect-loop.c (vect_determine_vectorization_factor)
27870 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
27871 (vect_get_known_peeling_cost): Likewise.
27872 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
27873 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
27874 (vect_transform_loop): Likewise. Use the lowest possible VF when
27875 updating the upper bounds of the loop.
27876 (vect_min_worthwhile_factor): Make static. Return an unsigned int
27877 rather than an int.
27878 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
27879 polynomial unroll factors.
27880 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
27881 (vect_make_slp_decision): Likewise.
27882 (vect_supported_load_permutation_p): Likewise, and polynomial
27884 (vect_analyze_slp_cost): Handle polynomial vf.
27885 (vect_slp_analyze_node_operations): Likewise.
27886 (vect_slp_analyze_bb_1): Likewise.
27887 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
27888 than an unsigned HOST_WIDE_INT.
27889 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
27890 (vectorizable_load): Handle polynomial vf.
27891 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
27893 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
27895 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27896 Alan Hayward <alan.hayward@arm.com>
27897 David Sherwood <david.sherwood@arm.com>
27899 * match.pd: Handle bit operations involving three constants
27900 and try to fold one pair.
27902 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27904 * tree-vect-loop-manip.c: Include gimple-fold.h.
27905 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
27906 niters_maybe_zero parameters. Handle other cases besides a step of 1.
27907 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
27908 Add a path that uses a step of VF instead of 1, but disable it
27910 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
27911 and niters_no_overflow parameters. Update calls to
27912 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
27913 Create a new SSA name if the latter choses to use a ste other
27914 than zero, and return it via niters_vector_mult_vf_var.
27915 * tree-vect-loop.c (vect_transform_loop): Update calls to
27916 vect_do_peeling, vect_gen_vector_loop_niters and
27917 slpeel_make_loop_iterate_ntimes.
27918 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
27919 (vect_gen_vector_loop_niters): Update declarations after above changes.
27921 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
27923 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
27924 128-bit round to integer instructions.
27925 (ceil<mode>2): Likewise.
27926 (btrunc<mode>2): Likewise.
27927 (round<mode>2): Likewise.
27929 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27931 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
27932 unaligned VSX load/store on P8/P9.
27933 (expand_block_clear): Allow the use of unaligned VSX
27934 load/store on P8/P9.
27936 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27938 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
27940 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
27941 swap associated with both a load and a store.
27943 2018-01-02 Andrew Waterman <andrew@sifive.com>
27945 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
27946 * config/riscv/riscv.md (clear_cache): Use it.
27948 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
27950 * web.c: Remove out-of-date comment.
27952 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
27954 * expr.c (fixup_args_size_notes): Check that any existing
27955 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
27956 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
27957 (emit_single_push_insn): ...here.
27959 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
27961 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
27962 (const_vector_encoded_nelts): New function.
27963 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
27964 (const_vector_int_elt, const_vector_elt): Declare.
27965 * emit-rtl.c (const_vector_int_elt_1): New function.
27966 (const_vector_elt): Likewise.
27967 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
27968 of CONST_VECTOR_ELT.
27970 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
27972 * expr.c: Include rtx-vector-builder.h.
27973 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
27974 directly on the tree encoding.
27975 (const_vector_from_tree): Likewise.
27976 * optabs.c: Include rtx-vector-builder.h.
27977 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
27978 sequence of "u" values.
27979 * vec-perm-indices.c: Include rtx-vector-builder.h.
27980 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
27981 directly on the vec_perm_indices encoding.
27983 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
27985 * doc/rtl.texi (const_vector): Describe new encoding scheme.
27986 * Makefile.in (OBJS): Add rtx-vector-builder.o.
27987 * rtx-vector-builder.h: New file.
27988 * rtx-vector-builder.c: Likewise.
27989 * rtl.h (rtx_def::u2): Add a const_vector field.
27990 (CONST_VECTOR_NPATTERNS): New macro.
27991 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
27992 (CONST_VECTOR_DUPLICATE_P): Likewise.
27993 (CONST_VECTOR_STEPPED_P): Likewise.
27994 (CONST_VECTOR_ENCODED_ELT): Likewise.
27995 (const_vec_duplicate_p): Check for a duplicated vector encoding.
27996 (unwrap_const_vec_duplicate): Likewise.
27997 (const_vec_series_p): Check for a non-duplicated vector encoding.
27998 Say that the function only returns true for integer vectors.
27999 * emit-rtl.c: Include rtx-vector-builder.h.
28000 (gen_const_vec_duplicate_1): Delete.
28001 (gen_const_vector): Call gen_const_vec_duplicate instead of
28002 gen_const_vec_duplicate_1.
28003 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
28004 (gen_const_vec_duplicate): Use rtx_vector_builder.
28005 (gen_const_vec_series): Likewise.
28006 (gen_rtx_CONST_VECTOR): Likewise.
28007 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
28008 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
28009 Build a new vector rather than modifying a CONST_VECTOR in-place.
28010 (handle_special_swappables): Update call accordingly.
28011 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
28012 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
28013 Build a new vector rather than modifying a CONST_VECTOR in-place.
28014 (handle_special_swappables): Update call accordingly.
28016 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
28018 * simplify-rtx.c (simplify_const_binary_operation): Use
28019 CONST_VECTOR_ELT instead of XVECEXP.
28021 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
28023 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
28024 the selector elements to be different from the data elements
28025 if the selector is a VECTOR_CST.
28026 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
28027 ssizetype for the selector.
28029 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
28031 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
28032 before testing each element individually.
28033 * tree-vect-generic.c (lower_vec_perm): Likewise.
28035 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
28037 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
28038 * selftest-run-tests.c (selftest::run_tests): Call it.
28039 * vector-builder.h (vector_builder::operator ==): New function.
28040 (vector_builder::operator !=): Likewise.
28041 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
28042 (vec_perm_indices::all_from_input_p): New function.
28043 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
28044 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
28045 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
28046 instead of reading the VECTOR_CST directly. Detect whether both
28047 vector inputs are the same before constructing the vec_perm_indices,
28048 and update the number of inputs argument accordingly. Use the
28049 utility functions added above. Only construct sel2 if we need to.
28051 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
28053 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
28054 the broadcast of the low byte.
28055 (expand_mult_highpart): Use an explicit encoding for the permutes.
28056 * optabs-query.c (can_mult_highpart_p): Likewise.
28057 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
28058 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
28059 (vectorizable_bswap): Likewise.
28060 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
28061 explicit encoding for the power-of-2 permutes.
28062 (vect_permute_store_chain): Likewise.
28063 (vect_grouped_load_supported): Likewise.
28064 (vect_permute_load_chain): Likewise.
28066 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
28068 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
28069 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
28070 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
28071 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
28072 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
28073 (vect_gen_perm_mask_any): Likewise.
28075 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
28077 * int-vector-builder.h: New file.
28078 * vec-perm-indices.h: Include int-vector-builder.h.
28079 (vec_perm_indices): Redefine as an int_vector_builder.
28080 (auto_vec_perm_indices): Delete.
28081 (vec_perm_builder): Redefine as a stand-alone class.
28082 (vec_perm_indices::vec_perm_indices): New function.
28083 (vec_perm_indices::clamp): Likewise.
28084 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
28085 (vec_perm_indices::new_vector): New function.
28086 (vec_perm_indices::new_expanded_vector): Update for new
28087 vec_perm_indices class.
28088 (vec_perm_indices::rotate_inputs): New function.
28089 (vec_perm_indices::all_in_range_p): Operate directly on the
28090 encoded form, without computing elided elements.
28091 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
28092 encoding. Update for new vec_perm_indices class.
28093 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
28094 the given vec_perm_builder.
28095 (expand_vec_perm_var): Update vec_perm_builder constructor.
28096 (expand_mult_highpart): Use vec_perm_builder instead of
28097 auto_vec_perm_indices.
28098 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
28099 vec_perm_indices instead of auto_vec_perm_indices. Use a single
28100 or double series encoding as appropriate.
28101 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
28102 vec_perm_indices instead of auto_vec_perm_indices.
28103 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
28104 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
28105 (vect_permute_store_chain): Likewise.
28106 (vect_grouped_load_supported): Likewise.
28107 (vect_permute_load_chain): Likewise.
28108 (vect_shift_permute_load_chain): Likewise.
28109 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
28110 (vect_transform_slp_perm_load): Likewise.
28111 (vect_schedule_slp_instance): Likewise.
28112 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
28113 (vectorizable_mask_load_store): Likewise.
28114 (vectorizable_bswap): Likewise.
28115 (vectorizable_store): Likewise.
28116 (vectorizable_load): Likewise.
28117 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
28118 vec_perm_indices instead of auto_vec_perm_indices. Use
28119 tree_to_vec_perm_builder to read the vector from a tree.
28120 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
28121 vec_perm_builder instead of a vec_perm_indices.
28122 (have_whole_vector_shift): Use vec_perm_builder and
28123 vec_perm_indices instead of auto_vec_perm_indices. Leave the
28124 truncation to calc_vec_perm_mask_for_shift.
28125 (vect_create_epilog_for_reduction): Likewise.
28126 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
28127 from auto_vec_perm_indices to vec_perm_indices.
28128 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
28129 instead of changing individual elements.
28130 (aarch64_vectorize_vec_perm_const): Use new_vector to install
28131 the vector in d.perm.
28132 * config/arm/arm.c (expand_vec_perm_d::perm): Change
28133 from auto_vec_perm_indices to vec_perm_indices.
28134 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
28135 instead of changing individual elements.
28136 (arm_vectorize_vec_perm_const): Use new_vector to install
28137 the vector in d.perm.
28138 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
28139 Update vec_perm_builder constructor.
28140 (rs6000_expand_interleave): Likewise.
28141 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
28142 (rs6000_expand_interleave): Likewise.
28144 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
28146 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
28147 to qimode could truncate the indices.
28148 * optabs.c (expand_vec_perm_var): Likewise.
28150 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
28152 * Makefile.in (OBJS): Add vec-perm-indices.o.
28153 * vec-perm-indices.h: New file.
28154 * vec-perm-indices.c: Likewise.
28155 * target.h (vec_perm_indices): Replace with a forward class
28157 (auto_vec_perm_indices): Move to vec-perm-indices.h.
28158 * optabs.h: Include vec-perm-indices.h.
28159 (expand_vec_perm): Delete.
28160 (selector_fits_mode_p, expand_vec_perm_var): Declare.
28161 (expand_vec_perm_const): Declare.
28162 * target.def (vec_perm_const_ok): Replace with...
28163 (vec_perm_const): ...this new hook.
28164 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
28165 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
28166 * doc/tm.texi: Regenerate.
28167 * optabs.def (vec_perm_const): Delete.
28168 * doc/md.texi (vec_perm_const): Likewise.
28169 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
28170 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
28171 expand_vec_perm for constant permutation vectors. Assert that
28172 the mode of variable permutation vectors is the integer equivalent
28173 of the mode that is being permuted.
28174 * optabs-query.h (selector_fits_mode_p): Declare.
28175 * optabs-query.c: Include vec-perm-indices.h.
28176 (selector_fits_mode_p): New function.
28177 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
28178 is defined, instead of checking whether the vec_perm_const_optab
28179 exists. Use targetm.vectorize.vec_perm_const instead of
28180 targetm.vectorize.vec_perm_const_ok. Check whether the indices
28181 fit in the vector mode before using a variable permute.
28182 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
28183 vec_perm_indices instead of an rtx.
28184 (expand_vec_perm): Replace with...
28185 (expand_vec_perm_const): ...this new function. Take the selector
28186 as a vec_perm_indices rather than an rtx. Also take the mode of
28187 the selector. Update call to shift_amt_for_vec_perm_mask.
28188 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
28189 Use vec_perm_indices::new_expanded_vector to expand the original
28190 selector into bytes. Check whether the indices fit in the vector
28191 mode before using a variable permute.
28192 (expand_vec_perm_var): Make global.
28193 (expand_mult_highpart): Use expand_vec_perm_const.
28194 * fold-const.c: Includes vec-perm-indices.h.
28195 * tree-ssa-forwprop.c: Likewise.
28196 * tree-vect-data-refs.c: Likewise.
28197 * tree-vect-generic.c: Likewise.
28198 * tree-vect-loop.c: Likewise.
28199 * tree-vect-slp.c: Likewise.
28200 * tree-vect-stmts.c: Likewise.
28201 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
28203 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
28204 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
28205 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
28206 (aarch64_vectorize_vec_perm_const): ...this new function.
28207 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
28208 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
28209 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
28210 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
28211 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
28212 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
28213 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
28215 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
28216 check for NEON modes.
28217 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
28218 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
28219 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
28220 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
28222 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
28223 the old VEC_PERM_CONST conditions.
28224 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
28225 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
28226 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
28227 (ia64_vectorize_vec_perm_const_ok): Merge into...
28228 (ia64_vectorize_vec_perm_const): ...this new function.
28229 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
28230 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
28231 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
28232 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
28233 * config/mips/mips.c (mips_expand_vec_perm_const)
28234 (mips_vectorize_vec_perm_const_ok): Merge into...
28235 (mips_vectorize_vec_perm_const): ...this new function.
28236 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
28237 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
28238 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
28239 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
28240 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
28241 (rs6000_expand_vec_perm_const): Delete.
28242 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
28244 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
28245 (altivec_expand_vec_perm_const_le): Take each operand individually.
28246 Operate on constant selectors rather than rtxes.
28247 (altivec_expand_vec_perm_const): Likewise. Update call to
28248 altivec_expand_vec_perm_const_le.
28249 (rs6000_expand_vec_perm_const): Delete.
28250 (rs6000_vectorize_vec_perm_const_ok): Delete.
28251 (rs6000_vectorize_vec_perm_const): New function.
28252 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
28253 an element count and rtx array.
28254 (rs6000_expand_extract_even): Update call accordingly.
28255 (rs6000_expand_interleave): Likewise.
28256 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
28257 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
28258 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
28259 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
28260 (rs6000_expand_vec_perm_const): Delete.
28261 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
28262 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
28263 (altivec_expand_vec_perm_const_le): Take each operand individually.
28264 Operate on constant selectors rather than rtxes.
28265 (altivec_expand_vec_perm_const): Likewise. Update call to
28266 altivec_expand_vec_perm_const_le.
28267 (rs6000_expand_vec_perm_const): Delete.
28268 (rs6000_vectorize_vec_perm_const_ok): Delete.
28269 (rs6000_vectorize_vec_perm_const): New function. Remove stray
28270 reference to the SPE evmerge intructions.
28271 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
28272 an element count and rtx array.
28273 (rs6000_expand_extract_even): Update call accordingly.
28274 (rs6000_expand_interleave): Likewise.
28275 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
28276 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
28278 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
28280 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
28282 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
28283 vector mode and that that mode matches the mode of the data
28285 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
28286 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
28287 directly using expand_vec_perm_1 when forcing selectors into
28289 (expand_vec_perm_var): New function, split out from expand_vec_perm.
28291 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
28293 * optabs-query.h (can_vec_perm_p): Delete.
28294 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
28295 * optabs-query.c (can_vec_perm_p): Split into...
28296 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
28297 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
28298 particular selector is valid.
28299 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
28300 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
28301 (vect_grouped_load_supported): Likewise.
28302 (vect_shift_permute_load_chain): Likewise.
28303 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
28304 (vect_transform_slp_perm_load): Likewise.
28305 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
28306 (vectorizable_bswap): Likewise.
28307 (vect_gen_perm_mask_checked): Likewise.
28308 * fold-const.c (fold_ternary_loc): Likewise. Don't take
28309 implementations of variable permutation vectors into account
28310 when deciding which selector to use.
28311 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
28312 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
28313 with a false third argument.
28314 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
28315 to test whether the constant selector is valid and can_vec_perm_var_p
28316 to test whether a variable selector is valid.
28318 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
28320 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
28321 * optabs-query.c (can_vec_perm_p): Likewise.
28322 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
28323 instead of vec_perm_indices.
28324 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
28325 (vect_gen_perm_mask_checked): Likewise,
28326 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
28327 (vect_gen_perm_mask_checked): Likewise,
28329 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
28331 * optabs-query.h (qimode_for_vec_perm): Declare.
28332 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
28333 (qimode_for_vec_perm): ...this new function.
28334 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
28336 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28338 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
28339 does not have a conditional at the top.
28341 2018-01-02 Richard Biener <rguenther@suse.de>
28343 * ipa-inline.c (big_speedup_p): Fix expression.
28345 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
28348 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
28351 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
28355 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
28356 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
28357 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
28358 cond_taken_branch_cost 3->4.
28360 2018-01-01 Jakub Jelinek <jakub@redhat.com>
28362 PR tree-optimization/83581
28363 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
28364 TODO_cleanup_cfg if any changes have been made.
28366 PR middle-end/83608
28367 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
28368 convert_modes if target mode has the right side, but different mode
28371 PR middle-end/83609
28372 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
28373 last argument when extracting from CONCAT. If either from_real or
28374 from_imag is NULL, use expansion through memory. If result is not
28375 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
28376 the parts directly to inner mode, if even that fails, use expansion
28379 PR middle-end/83623
28380 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
28381 check for bswap in mode rather than HImode and use that in expand_unop
28384 Copyright (C) 2018 Free Software Foundation, Inc.
28386 Copying and distribution of this file, with or without modification,
28387 are permitted in any medium without royalty provided the copyright
28388 notice and this notice are preserved.