1 2018-08-28 Martin Sebor <msebor@redhat.com>
4 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
5 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
6 (pass_walloca::gate): Use it.
7 (alloca_call_type): Same.
8 (pass_walloca::execute): Same.
9 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
11 2018-08-28 David Malcolm <dmalcolm@redhat.com>
13 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
14 GCC_VERSION for usage of "__gcc_dump_printf__" format from
17 2018-08-28 Richard Biener <rguenther@suse.de>
19 PR tree-optimization/87124
20 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
21 constants before looking up avail.
23 2018-08-28 Jakub Jelinek <jakub@redhat.com>
26 * calls.c (maybe_warn_nonstring_arg): Punt early if
27 warn_stringop_overflow is zero. Don't call get_range_strlen
28 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
29 Swap comparison operands to have constants on rhs. Only use
30 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
33 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
35 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
36 use of tree_to_shwi. Remove duplicated test for the size being
37 a whole number of bytes.
39 2018-08-28 Richard Biener <rguenther@suse.de>
41 PR tree-optimization/87117
42 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
43 Handle removed stmt without LHS (GIMPLE_NOP).
45 2018-08-28 Richard Biener <rguenther@suse.de>
47 PR tree-optimization/87117
48 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
49 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
51 2018-08-28 Richard Biener <rguenther@suse.de>
53 PR tree-optimization/87117
54 * tree-ssa-pre.c (compute_avail): Do not make expressions
55 with predicated values available.
56 (get_expr_value_id): Assert we do not run into predicated value
59 2018-08-28 Richard Biener <rguenther@suse.de>
61 PR tree-optimization/87117
62 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
64 (get_expr_operands): Handle STRING_CST like other decls.
66 2018-08-28 Martin Liska <mliska@suse.cz>
68 * tree.h: Update documentation of fndecl_built_in_p
71 2018-08-27 Jeff Law <law@redhat.com>
73 PR tree-optimization/87110
74 * tree-ssa-dse.c (compute_trims): Handle non-constant
77 2018-08-27 Martin Sebor <msebor@redhat.com>
79 PR tree-optimization/86914
80 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
82 2018-08-27 Martin Sebor <msebor@redhat.com>
84 PR tree-optimization/87112
85 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
86 the type of the bound argument.
88 2018-08-27 Jeff Law <law@redhat.com>
90 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
91 type does not have a TYPE_SIZE_UNIT.
93 2018-08-27 Steve Ellcey <sellcey@cavium.com>
95 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
96 with include of backend.h.
98 2018-08-27 Richard Biener <rguenther@suse.de>
100 PR tree-optimization/86927
101 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
102 use const cond reduction code.
104 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
106 PR tree-optimization/85758
107 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
109 2018-08-27 David Malcolm <dmalcolm@redhat.com>
112 * diagnostic-show-locus.c (class layout_range): Update for
113 conversion of show_caret_p to a tri-state.
114 (layout_range::layout_range): Likewise.
115 (make_range): Likewise.
116 (layout::maybe_add_location_range): Likewise.
117 (layout::should_print_annotation_line_p): Don't show annotation
118 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
119 (layout::get_state_at_point): Update for conversion of
120 show_caret_p to a tri-state. Bail out early for
121 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
122 underlining or source colorization.
123 (gcc_rich_location::add_location_if_nearby): Update for conversion
124 of show_caret_p to a tri-state.
125 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
126 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
128 (selftest::test_one_liner_labels): Likewise.
129 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
130 conversion of show_caret_p to a tri-state.
131 * pretty-print.c (text_info::set_location): Likewise.
132 * pretty-print.h (text_info::set_location): Likewise.
133 * substring-locations.c (format_warning_n_va): Likewise.
134 * tree-diagnostic.c (default_tree_printer): Likewise.
135 * tree-pretty-print.c (newline_and_indent): Likewise.
137 2018-08-27 David Malcolm <dmalcolm@redhat.com>
140 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
141 line above for line-insertion fix-it hints.
142 (selftest::test_fixit_insert_containing_newline): Update the
143 expected results, and add a test with line-numbering enabled.
145 2018-08-27 Martin Liska <mliska@suse.cz>
148 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
149 params with DECL_HAS_VALUE_EXPR_P.
151 2018-08-27 Martin Liska <mliska@suse.cz>
153 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
154 selected expansion strategy.
156 2018-08-27 Martin Liska <mliska@suse.cz>
158 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
159 * builtins.c (is_builtin_fn): Likewise.
160 * attribs.c (diag_attr_exclusions): Use new function
161 fndecl_built_in_p and remove check for FUNCTION_DECL if
163 (builtin_mathfn_code): Likewise.
164 (fold_builtin_expect): Likewise.
165 (fold_call_expr): Likewise.
166 (fold_builtin_call_array): Likewise.
167 (fold_call_stmt): Likewise.
168 (set_builtin_user_assembler_name): Likewise.
169 (is_simple_builtin): Likewise.
170 * calls.c (gimple_alloca_call_p): Likewise.
171 (maybe_warn_nonstring_arg): Likewise.
172 * cfgexpand.c (expand_call_stmt): Likewise.
173 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
174 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
175 (cgraph_node::verify_node): Likewise.
176 * cgraphclones.c (build_function_decl_skip_args): Likewise.
177 (cgraph_node::create_clone): Likewise.
178 * config/arm/arm.c (arm_insert_attributes): Likewise.
179 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
180 * dse.c (scan_insn): Likewise.
181 * expr.c (expand_expr_real_1): Likewise.
182 * fold-const.c (operand_equal_p): Likewise.
183 (fold_binary_loc): Likewise.
184 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
185 * gimple-low.c (lower_stmt): Likewise.
186 * gimple-pretty-print.c (dump_gimple_call): Likewise.
187 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Likewise.
188 * gimple.c (gimple_build_call_from_tree): Likewise.
189 (gimple_call_builtin_p): Likewise.
190 (gimple_call_combined_fn): Likewise.
191 * gimplify.c (gimplify_call_expr): Likewise.
192 (gimple_boolify): Likewise.
193 (gimplify_modify_expr): Likewise.
194 (gimplify_addr_expr): Likewise.
195 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
196 * ipa-cp.c (determine_versionability): Likewise.
197 * ipa-fnsummary.c (compute_fn_summary): Likewise.
198 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
199 * ipa-split.c (visit_bb): Likewise.
200 (split_function): Likewise.
201 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
202 * lto-cgraph.c (input_node): Likewise.
203 * lto-streamer-out.c (write_symbol): Likewise.
204 * omp-low.c (setjmp_or_longjmp_p): Likewise.
205 (lower_omp_1): Likewise.
206 * predict.c (strip_predict_hints): Likewise.
207 * print-tree.c (print_node): Likewise.
208 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
209 * trans-mem.c (is_tm_irrevocable): Likewise.
210 (is_tm_load): Likewise.
211 (is_tm_simple_load): Likewise.
212 (is_tm_store): Likewise.
213 (is_tm_simple_store): Likewise.
214 (is_tm_abort): Likewise.
215 (tm_region_init_1): Likewise.
216 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
217 * tree-cfg.c (verify_gimple_call): Likewise.
218 (move_stmt_r): Likewise.
219 (stmt_can_terminate_bb_p): Likewise.
220 * tree-eh.c (lower_eh_constructs_2): Likewise.
221 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
222 * tree-inline.c (remap_gimple_stmt): Likewise.
224 (estimate_num_insns): Likewise.
225 (fold_marked_statements): Likewise.
226 * tree-sra.c (scan_function): Likewise.
227 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
228 (optimize_stack_restore): Likewise.
229 (pass_fold_builtins::execute): Likewise.
230 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
231 (mark_all_reaching_defs_necessary_1): Likewise.
232 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
233 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
234 (pass_forwprop::execute): Likewise.
235 * tree-ssa-loop-im.c (stmt_cost): Likewise.
236 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
237 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
238 * tree-ssa-strlen.c (get_string_length): Likewise.
239 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
240 (find_func_aliases_for_call): Likewise.
241 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
242 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
243 * tree-tailcall.c (find_tail_calls): Likewise.
244 * tree.c (need_assembler_name_p): Likewise.
245 (free_lang_data_in_decl): Likewise.
246 (get_call_combined_fn): Likewise.
247 * ubsan.c (is_ubsan_builtin_p): Likewise.
248 * varasm.c (incorporeal_function_p): Likewise.
249 * tree.h (DECL_BUILT_IN): Remove and replace with
251 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
252 (fndecl_built_in_p): New.
254 2018-08-27 Martin Liska <mliska@suse.cz>
256 PR tree-optimization/86847
257 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
258 Dump also subtree probability.
259 (switch_decision_tree::do_jump_if_equal): New function.
260 (switch_decision_tree::emit_case_nodes): Handle special
261 situations in balanced tree that can be emitted much simpler.
262 Fix calculation of probabilities that happen in tree expansion.
263 * tree-switch-conversion.h (struct cluster): Add
265 (struct simple_cluster): Likewise.
266 (struct case_tree_node): Add new function has_child.
267 (do_jump_if_equal): New.
269 2018-08-27 Martin Liska <mliska@suse.cz>
271 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
272 Add new argument to bit_test_cluster constructor.
273 (bit_test_cluster::emit): Set bits really number of values
275 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
276 probability argument.
277 * tree-switch-conversion.h (struct bit_test_cluster):
278 Add m_handles_entire_switch.
280 2018-08-27 Martin Liska <mliska@suse.cz>
282 PR tree-optimization/86702
283 * tree-switch-conversion.c (jump_table_cluster::emit):
284 Make probabilities even for values in jump table
285 according to number of cases handled.
286 (switch_decision_tree::compute_cases_per_edge): Pass
287 argument to reset_out_edges_aux function.
288 (switch_decision_tree::analyze_switch_statement): Likewise.
289 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
292 2018-08-27 Martin Liska <mliska@suse.cz>
294 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
295 cfun argument explicitly.
296 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
297 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
298 function gimple_switch_default_bb.
299 (convert_switch_statements):
301 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
302 * stmt.c (label_to_block_fn): Use label_to_block and pass
303 cfun argument explicitly and use gimple_switch_label_bb.
304 (expand_case): Likewise.
305 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
306 cfun argument explicitly. Likewise.
307 (make_edges_bb): Likewise.
308 (make_cond_expr_edges): Likewise.
309 (get_cases_for_edge): Likewise.
310 (make_gimple_switch_edges): Likewise.
311 (label_to_block_fn): Likewise.
312 (label_to_block): Likewise.
313 (make_goto_expr_edges): Likewise.
314 (make_gimple_asm_edges): Likewise.
315 (main_block_label): Likewise.
316 (group_case_labels_stmt): Likewise.
317 (find_taken_edge_computed_goto): Likewise.
318 (find_taken_edge_switch_expr): Likewise.
319 (gimple_verify_flow_info): Likewise.
320 (gimple_redirect_edge_and_branch): Likewise.
321 (gimple_switch_label_bb): New function.
322 (gimple_switch_default_bb): Likewise.
323 (gimple_switch_edge): Likewise.
324 (gimple_switch_default_edge): Likewise.
325 * tree-cfg.h (label_to_block_fn): Remove and replace ...
326 (label_to_block): ... with this.
327 (gimple_switch_label_bb): New.
328 (gimple_switch_default_bb): Likewise.
329 (gimple_switch_edge): Likewise.
330 (gimple_switch_default_edge): Likewise.
331 * tree-cfgcleanup.c (convert_single_case_switch): Use
332 new gimple functions and pass new argument to label_to_block.
333 (cleanup_control_flow_bb):
334 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
335 cfun argument explicitly.
336 (make_eh_edges): Likewise.
337 (redirect_eh_dispatch_edge): Likewise.
338 (lower_resx): Likewise.
339 (lower_eh_dispatch): Likewise.
340 (maybe_remove_unreachable_handlers): Likewise.
341 (unsplit_eh): Likewise.
342 (cleanup_empty_eh): Likewise.
343 (verify_eh_edges): Likewise.
344 (verify_eh_dispatch_edge): Likewise.
345 * tree-ssa-dom.c (record_edge_info): Likewise.
346 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
347 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
348 (thread_through_normal_block): Likewise.
349 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
350 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
351 * tree-switch-conversion.c (switch_conversion::collect): Use new
353 (switch_conversion::check_final_bb): Likewise.
354 (switch_conversion::gather_default_values): Pass new argument
356 (switch_conversion::build_constructors): Likewise.
357 (switch_decision_tree::compute_cases_per_edge): Use new
358 gimple_switch_edge function.
359 (switch_decision_tree::analyze_switch_statement): Pass new argument
361 (switch_decision_tree::try_switch_expansion): Use
362 gimple_switch_default_edge.
363 * tree-vrp.c (find_switch_asserts): Pass new argument
365 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
366 (vr_values::simplify_switch_using_ranges): Likewise.
368 2018-08-27 Richard Biener <rguenther@suse.de>
370 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
371 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
373 * tree-ssa-sccvn.h (struct vn_pval): New structure.
374 (struct vn_nary_op_s): Add unwind_to member. Add
375 predicated_values flag and put result into a union together
376 with a linked list of vn_pval.
377 (struct vn_ssa_aux): Add name member to make maintaining
378 a map of SSA name to vn_ssa_aux possible. Remove no longer
379 needed info, dfsnum, low, visited, on_sccstack, use_processed
380 and range_info_anti_range_p members.
381 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
382 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
384 (vn_valueize): New global.
385 (vn_context_bb): Likewise.
386 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
387 VN_INFO_PTR_INFO): Remove.
388 * tree-ssa-sccvn.c: ... (rewrite)
389 (pass_fre::execute): For -O2+ initialize loops and run
390 RPO VN in optimistic mode (iterating). For -O1 and -Og
391 run RPO VN in non-optimistic mode.
392 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
393 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
394 * doc/invoke.texi (sccvn-max-scc-size): Remove.
395 (rpo-vn-max-loop-depth): Document.
396 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
397 when valuezing the VUSE signals we walked out of the region.
398 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
399 (phi_translate): Set VN context block to use for availability
401 (compute_avail): Likewise.
402 (pre_valueize): New function.
403 (pass_pre::execute): Adjust to the RPO VN API.
405 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
406 (propagate_constants_for_unrolling): Remove.
407 (tree_unroll_loops_completely): Perform value-numbering
408 on the unrolled bodies loop parent.
410 2018-08-27 Richard Biener <rguenther@suse.de>
412 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
413 for partial antic compute.
415 2018-08-27 Jakub Jelinek <jakub@redhat.com>
417 PR rtl-optimization/87065
418 * combine.c (simplify_if_then_else): Formatting fix.
419 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
421 (known_cond): Don't return const_true_rtx for vector modes. Use
422 CONST0_RTX instead of const0_rtx. Formatting fixes.
424 2018-08-27 Martin Liska <mliska@suse.cz>
426 PR gcov-profile/87069
427 * gcov.c (process_file): Record files already processed
428 and warn about a file being processed multiple times.
430 2018-08-27 Martin Liska <mliska@suse.cz>
433 * config/aarch64/aarch64.c (aarch64_override_options_internal):
434 Set default values for x_aarch64_*_string strings.
435 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
436 prefix. For -mabi do not print '=ABI' in help and use
437 <option_value> format for -msve-vector-bits and -moverride
440 2018-08-26 Jeff Law <law@redhat.com>
442 * config/mips/frame-header-opt.c: Include "backend.h" rather than
445 2018-08-26 Marek Polacek <polacek@redhat.com>
447 PR c++/87029, Implement -Wredundant-move.
448 * doc/invoke.texi: Document -Wredundant-move.
450 2018-08-25 Martin Sebor <msebor@redhat.com>
452 PR tree-optimization/87059
453 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
454 to the same type as the other.
455 * fold-const.c (fold_binary_loc): Assert expectation.
457 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
459 * config/darwin.c (machopic_legitimize_pic_address): Clean up
460 extraneous parentheses, dead code section and formatting.
462 2018-08-24 David Malcolm <dmalcolm@redhat.com>
465 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
466 wide enough for jumps in the line-numbering to be visible.
467 (layout::print_gap_in_line_numbering): New member function.
468 (layout::calculate_line_spans): When using line numbering, merge
469 line spans that are only 1 line apart.
470 (diagnostic_show_locus): When printing line numbers, show gaps in
471 line numbering directly, rather than printing headers.
472 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
473 line-numbering with multiple line spans.
474 (selftest::test_fixit_insert_containing_newline_2): Add test of
475 line-numbering, in which the spans are close enough to be merged.
477 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
479 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
480 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
481 * tree-vrp.c (range_is_nonnull): Remove.
482 (range_includes_zero_p): Accept value_range instead of min/max.
483 (extract_range_from_binary_expr_1): Do not early bail on
485 Use range_includes_zero_p instead of range_is_nonnull.
486 (extract_range_from_unary_expr): Use range_includes_zero_p instead
488 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
489 special case VR_ANTI_RANGE.
490 (vrp_finalize): Same.
491 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
493 (range_is_nonnull): Remove.
494 * vr-values.c (vrp_stmt_computes_nonzero): Use
495 range_includes_zero_p instead of range_is_nonnull.
496 (extract_range_basic): Pass value_range to range_includes_zero_p
497 instead of range_is_nonnull.
499 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
501 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
502 * emit-rtl.h (rtl_data): Remove return_bnd.
503 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
504 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
505 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
506 (POINTER_BOUNDS_MODE): Remove definition.
507 (make_pointer_bounds_mode): Remove.
508 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
509 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
510 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
511 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
512 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
513 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
514 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
516 * config/i386/i386-modes.def (BND32, BND64): Remove.
517 * config/i386/i386.c (dbx_register_map): Remove bound registers.
518 (dbx64_register_map): Ditto.
519 (svr4_dbx_register_map): Ditto.
520 (indirect_thunk_bnd_needed): Remove.
521 (indirect_thunks_bnd_used): Ditto.
522 (indirect_return_bnd_needed): Ditto.
523 (indirect_return_via_cx_bnd): Ditto.
524 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
525 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
526 (output_indirect_thunk): Ditto. Remove need_prefix argument.
527 (output_indirect_thunk_function): Remove handling of
528 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
529 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
530 (ix86_save_reg): Remove handling of crtl->return_bnd.
531 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
532 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
533 and UNSPEC_BNDLX_ADDR.
534 (ix86_output_indirect_branch_via_reg): Remove handling of
535 indirect_thunk_prefix_bnd.
536 (ix86_output_indirect_branch_via_push): Ditto.
537 (ix86_output_function_return): Ditto.
538 (ix86_output_indirect_function_return): Ditto.
539 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
540 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
541 (CALL_USED_REGISTERS): Ditto.
542 (REG_ALLOC_ORDER): Update for removal of bound registers.
543 (HI_REGISTER_NAMES): Ditto.
544 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
545 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
546 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
547 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
548 (FIRST_PSEUDO_REG): Update.
549 (BND): Remove mode iterator.
550 * config/i386/predicates.md (bnd_mem_operator): Remove.
552 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
554 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
557 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
559 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
560 the case in which the permute needs only a single element and
561 repeats for every vector of the result. Extend that case to
562 handle variable-length vectors.
563 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
565 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
568 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
569 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
571 2018-08-24 Richard Biener <rguenther@suse.de>
573 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
574 bb_flags_allocated members.
575 (auto_flag): New RAII class for allocating flags.
576 (auto_edge_flag): New RAII class for allocating edge flags.
577 (auto_bb_flag): New RAII class for allocating bb flags.
578 * cfgloop.c (verify_loop_structure): Allocate temporary edge
580 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
581 in favor of temporarily allocated BB flag.
582 * hsa-brig.c: Re-order includes.
583 * hsa-dump.c: Likewise.
584 * hsa-regalloc.c: Likewise.
585 * print-rtl.c: Likewise.
586 * profile-count.c: Likewise.
588 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
591 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
594 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
597 * wide-int-range.cc (wide_int_range_div): Do not ignore result
598 from wide_int_range_multiplicative_op.
600 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
602 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
605 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
607 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
610 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
612 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
613 full GENERIC expression used for replacement.
615 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
617 * tree-vrp.c (abs_extent_range): Remove.
618 (extract_range_into_wide_ints): Pass wide ints by reference.
619 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
620 Pass wide ints by reference in all calls to
621 extract_range_into_wide_ints.
622 * wide-int-range.cc (wide_int_range_div): New.
623 * wide-int-range.h (wide_int_range_div): New.
624 (wide_int_range_includes_zero_p): New.
625 (wide_int_range_zero_p): New.
627 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
629 * config/aarch64/aarch64.md (arches): New enum.
630 (arch): New enum attr.
631 (arch_enabled): New attr.
632 (enabled): Now uses arch_enabled only.
633 (simd, sve, fp16): Removed attribute.
634 (fp): Attr now defined in terms of 'arch'.
635 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
636 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
637 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
638 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
639 attributes into 'arch'.
640 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
641 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
642 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
643 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
644 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
645 'simd' attribute into 'arch'.
646 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
647 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
648 Convert use of 'fp' attribute to 'arch'.
649 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
650 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
652 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
653 (different modes) Merge 'fp' and 'simd' into 'arch'.
654 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
657 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
659 PR rtl-optimization/87026
660 * expmed.c (canonicalize_comparison): If we can no longer create
661 pseudoregisters, don't.
663 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
666 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
668 * config/arm/arm.c (speculation_barrier_libfunc): New static
670 (arm_init_libfuncs): Initialize it.
671 (arm_emit_speculation_barrier): New function.
672 * config/arm/arm.md (speculation_barrier): Call
673 arm_emit_speculation_barrier for architectures that do not have
675 (speculation_barrier_insn): Only match on Armv7 or later.
677 2018-08-23 Richard Biener <rguenther@suse.de>
680 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
683 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
685 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
686 of single-vector TBLs.
687 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
690 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
693 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
694 aarch64_evpc_tbl guard.
696 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
698 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
701 2018-08-22 Martin Sebor <msebor@redhat.com>
704 * tree-pretty-print.c (pretty_print_string): Add argument.
705 (dump_generic_node): Call to pretty_print_string with string size.
707 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
709 PR rtl-optimization/86771
710 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
711 of two SETs into those two SETs, one to be placed at i2, if that SETs
712 destination is modified between i2 and i3.
714 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
716 PR tree-optimization/86725
717 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
719 (vect_analyze_scalar_cycles_1): Check it.
721 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
723 PR tree-optimization/86725
724 * tree-vect-loop.c (vect_is_simple_reduction): When treating
725 an outer loop phi as a double reduction, make sure that the
726 single user of the phi result is an inner loop phi.
728 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
730 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
731 grouped stores with gaps to a strided group.
733 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
735 * tree-vect-stmts.c (get_group_load_store_type)
736 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
737 first statement in a group.
739 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
741 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
742 the sequence used in gcc/gcc.c.
744 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
747 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
750 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
752 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
753 Darwin10-specific unwinder-shim.
754 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
755 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
756 New to cater for Darwin10 Rosetta.
758 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
760 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
761 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
763 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
768 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
769 (dwarf2out_switch_text_section): Generate a local label for the
770 second function sub-section and apply it as the second FDE start
772 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
773 second sub-section start.
775 2018-08-22 Richard Biener <rguenther@suse.de>
777 PR tree-optimization/86988
778 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
780 2018-08-22 Richard Biener <rguenther@suse.de>
782 PR tree-optimization/86945
783 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
785 2018-08-22 Alexandre Oliva <oliva@adacore.com>
787 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
788 a comment about how uses of r2 for .sdata2 come about.
790 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
792 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
794 2018-08-21 Marek Polacek <polacek@redhat.com>
796 PR c++/86981, Implement -Wpessimizing-move.
797 * doc/invoke.texi: Document -Wpessimizing-move.
799 2018-08-21 Jan Hubicka <jh@suse.cz>
801 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
802 * tree.h (is_redundant_typedef): Remove.
803 * dwarf2out.c (is_redundant_typedef): Turn into static function.
805 2018-08-21 Jan Hubicka <jh@suse.cz>
807 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
810 2018-08-21 Tamar Christina <tamar.christina@arm.com>
812 * expmed.c (extract_low_bits): Reject invalid subregs early.
814 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
817 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
820 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
822 * config/vxworks.h: Guard vxworks_asm_out_constructor and
823 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
824 * config/vxworks.c: Likewise.
826 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
828 * config/vxworks.c: Set targetm.have_ctors_dtors
829 if HAVE_INITFINI_ARRAY_SUPPORT.
830 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
831 if HAVE_INITFINI_ARRAY_SUPPORT.
833 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
835 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
836 default search path for VxWorks < 7.
838 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
840 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
841 (get_format_string): Refer to c_getstr.
843 2018-08-21 Tom de Vries <tdevries@suse.de>
845 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
846 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
847 (debuginfo_early_stop): Declare.
848 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
849 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
850 (debuginfo_early_stop): New function.
851 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
852 and debuginfo_early_stop.
853 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
854 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
855 (general_init): Call debuginfo_early_init.
856 (finalize): Call debuginfo_fini.
857 (do_compile): Call debuginfo_init.
858 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
860 (@item -fdump-debug, @item -fdump-earlydebug): Add.
862 2018-08-21 Tom de Vries <tdevries@suse.de>
864 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
865 flag_dump_noaddr and flag_dump_unnumbered.
867 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
869 * wide-int-range.cc (wide_int_range_abs): New.
870 (wide_int_range_order_set): Rename from wide_int_range_min_max.
871 * wide-int-range.h (wide_int_range_abs): New.
872 (wide_int_range_min_max): New.
873 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
874 case to call wide_int_range_abs.
875 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
876 (extract_range_from_abs_expr): Delete.
878 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
881 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
882 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
885 2018-08-20 Nathan Sidwell <nathan@acm.org>
886 Jeff Law <law@redhat.com>
888 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
889 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
891 2018-08-20 David Malcolm <dmalcolm@redhat.com>
894 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
895 (decl_attributes): Likewise.
896 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
898 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
899 * diagnostic-core.h (class auto_diagnostic_group): New class.
900 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
901 (diagnostic_report_diagnostic): Handle the first diagnostics within
903 (emit_diagnostic): Add auto_diagnostic_group instance.
905 (inform_n): Likewise.
907 (warning_at): Likewise.
908 (warning_n): Likewise.
910 (permerror): Likewise.
913 (error_at): Likewise.
915 (fatal_error): Likewise.
916 (internal_error): Likewise.
917 (internal_error_no_backtrace): Likewise.
918 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
919 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
920 * diagnostic.h (struct diagnostic_context): Add fields
921 "diagnostic_group_nesting_depth",
922 "diagnostic_group_emission_count", "begin_group_cb",
924 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
925 Add auto_diagnostic_group instance(s).
926 (find_explicit_erroneous_behavior): Likewise.
927 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
928 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
929 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
930 (gimplify_va_arg_expr): Likewise.
931 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
932 (HSA_SORRY_AT): Likewise.
933 * ipa-devirt.c (compare_virtual_tables): Likewise.
934 (warn_odr): Likewise.
935 * multiple_target.c (expand_target_clones): Likewise.
936 * opts-common.c (cmdline_handle_error): Likewise.
937 * reginfo.c (globalize_reg): Likewise.
938 * substring-locations.c (format_warning_n_va): Likewise.
939 * tree-inline.c (expand_call_inline): Likewise.
940 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
941 * tree-ssa-loop-niter.c
942 (do_warn_aggressive_loop_optimizations): Likewise.
943 * tree-ssa-uninit.c (warn_uninit): Likewise.
944 * tree.c (warn_deprecated_use): Likewise.
946 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
949 * config/i386/i386.md (eh_return): Always update EH return
950 address in word_mode.
952 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
954 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
955 TARGET_SPLIT_COMPLEX_ARG is defined.
957 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
959 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
961 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
964 * expr.c (expand_assignment): Assert that bitpos is positive.
965 (store_field): Likewise
966 (expand_expr_real_1): Make sure that bitpos is positive.
967 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
970 2018-08-20 Nathan Sidwell <nathan@acm.org>
972 * Makefile.in (CPP_ID_DATA_H): Delete.
973 (CPP_INTERNAL_H): Don't add it.
974 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
975 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
977 2018-08-20 Richard Biener <rguenther@suse.de>
979 PR tree-optimization/78655
980 * tree-vrp.c (extract_range_from_binary_expr_1): Make
981 pointer + offset nonnull if either operand is nonnull work.
983 2018-08-20 Tom de Vries <tdevries@suse.de>
985 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
986 unless the referenced die describes the added property using
987 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
988 Otherwise, add a DW_AT_location to the referenced die.
990 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
993 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
994 register_operand when calling ix86_set_reg_reg_cost.
995 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
996 Set *total to 0 for operands that satisfy x86_64_immediate_operand
997 predicate and to 1 otherwise.
999 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
1001 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
1002 emit a diagnostic that it is not supported and reset the option.
1003 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
1004 supported and consume the option. (ASM_FINAL_SPEC): New.
1006 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
1008 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
1011 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
1013 C-SKY port: Documentation
1015 * doc/extend.texi (C-SKY Function Attributes): New section.
1016 * doc/invoke.texi (Option Summary): Add C-SKY options.
1017 (C-SKY Options): New section.
1018 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
1020 2018-08-17 Jojo <jijie_rong@c-sky.com>
1021 Huibin Wang <huibin_wang@c-sky.com>
1022 Sandra Loosemore <sandra@codesourcery.com>
1023 Chung-Lin Tang <cltang@codesourcery.com>
1025 C-SKY port: Backend implementation
1027 * config/csky/*: New.
1028 * common/config/csky/*: New.
1030 2018-08-17 Jojo <jijie_rong@c-sky.com>
1031 Huibin Wang <huibin_wang@c-sky.com>
1032 Sandra Loosemore <sandra@codesourcery.com>
1033 Chung-Lin Tang <cltang@codesourcery.com>
1034 Andrew Jenner <andrew@codesourcery.com>
1036 C-SKY port: Configury
1038 * config.gcc (csky-*-*): New.
1039 * configure.ac: Add csky to targets for dwarf2 debug_line support.
1040 * configure: Regenerated.
1042 2018-08-17 David Malcolm <dmalcolm@redhat.com>
1044 * dump-context.h: Include "dumpfile.h".
1045 (dump_context::dump_printf_va): Convert final param from va_list
1046 to va_list *. Convert from ATTRIBUTE_PRINTF to
1047 ATTRIBUTE_GCC_DUMP_PRINTF.
1048 (dump_context::dump_printf_loc_va): Likewise.
1049 * dumpfile.c: Include "stringpool.h".
1050 (make_item_for_dump_printf_va): Delete.
1051 (make_item_for_dump_printf): Delete.
1052 (class dump_pretty_printer): New class.
1053 (dump_pretty_printer::dump_pretty_printer): New ctor.
1054 (dump_pretty_printer::emit_items): New member function.
1055 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
1057 (dump_pretty_printer::emit_item): New member function.
1058 (dump_pretty_printer::stash_item): New member function.
1059 (dump_pretty_printer::format_decoder_cb): New member function.
1060 (dump_pretty_printer::decode_format): New member function.
1061 (dump_context::dump_printf_va): Reimplement in terms of
1062 dump_pretty_printer.
1063 (dump_context::dump_printf_loc_va): Convert final param from va_list
1065 (dump_context::begin_scope): Reimplement call to
1066 make_item_for_dump_printf.
1067 (dump_printf): Update for change to dump_printf_va.
1068 (dump_printf_loc): Likewise.
1069 (selftest::test_capture_of_dump_calls): Convert "stmt" from
1070 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
1071 with %T, %E, and %G.
1072 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
1073 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
1074 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
1075 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
1076 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
1077 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
1078 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
1079 within a dump_printf_loc call to "%wu".
1080 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
1081 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
1082 missing space after "=".
1083 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
1084 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
1085 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
1086 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
1087 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
1088 duplicate "vectorized" from message.
1090 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
1092 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
1093 polyNxK_t element's TYPE_STRING_FLAG.
1095 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
1097 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
1098 (they were unnamed before). Fix comments.
1100 2018-08-17 Nathan Sidwell <nathan@acm.org>
1102 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
1104 2018-08-17 Richard Biener <rguenther@suse.de>
1106 PR tree-optimization/86841
1107 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
1109 2018-08-17 Martin Liska <mliska@suse.cz>
1111 * common.opt: Remove Warn, Init and Report for options with
1112 Ignore/Deprecated flag. Warning is done automatically for
1114 * config/i386/i386.opt: Likewise.
1115 * config/ia64/ia64.opt: Likewise.
1116 * config/rs6000/rs6000.opt: Likewise.
1117 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
1118 Remove usage of flag_check_pointer_bounds.
1119 * lto-wrapper.c (merge_and_complain): Do not handle
1120 OPT_fcheck_pointer_bounds.
1121 (append_compiler_options): Likewise.
1122 * opt-functions.awk: Do not handle Deprecated.
1123 * optc-gen.awk: Check that Var, Report and Init are not
1124 used for an option with Ignore/Deprecated flag.
1125 * opts-common.c (decode_cmdline_option): Do not report
1127 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
1129 * opts.h (struct cl_option): Remove cl_deprecated flag.
1130 (CL_ERR_DEPRECATED): Remove error enum value.
1132 2018-08-17 Richard Biener <rguenther@suse.de>
1135 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
1136 across a va-arg-pack using call adjust its return value accordingly.
1138 2018-08-16 Martin Sebor <msebor@redhat.com>
1140 PR tree-optimization/86853
1141 * gimple-ssa-sprintf.c (struct format_result): Rename member.
1142 (struct fmtresult): Add member and initialize it in ctors.
1143 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
1144 (format_string): Handle %S the same as %ls. Set MAYFAIL.
1145 (format_directive): Set POSUNDER4K when MAYFAIL is set.
1146 (parse_directive): Handle %C same as %c.
1147 (sprintf_dom_walker::compute_format_length): Adjust.
1148 (is_call_safe): Adjust.
1150 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
1152 * builtins.c (c_strlen): Add new parameter eltsize. Use it
1153 for determining how to count the elements.
1154 * builtins.h (c_strlen): Adjust prototype.
1155 * expr.c (string_constant): Add new parameter mem_size.
1156 Set *mem_size appropriately.
1157 * expr.h (string_constant): Adjust protoype.
1158 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
1159 * gimple-fold.h (get_range_strlen): Adjust prototype.
1160 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
1161 (format_string): Call get_string_length with eltsize.
1163 2018-08-16 David Malcolm <dmalcolm@redhat.com>
1165 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
1166 to emit the span, rather than setting it as the prefix.
1168 2018-08-16 David Malcolm <dmalcolm@redhat.com>
1170 * diagnostic-show-locus.c (layout::start_annotation_line): Add
1171 "margin_char" parameter, defaulting to space. Use it in place
1172 of pp_space for the initial part of the margin.
1173 (layout::print_leading_fixits): Use '+' when filling the margin
1174 of line-insertion fix-it hints.
1176 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
1178 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
1181 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
1183 * config/rs6000/altivec.md: Don't set length attribute to the default
1185 * config/rs6000/darwin.md: Ditto.
1186 * config/rs6000/dfp.md: Ditto.
1187 * config/rs6000/htm.md: Ditto.
1188 * config/rs6000/rs6000.md: Ditto.
1189 * config/rs6000/sync.md: Ditto.
1190 * config/rs6000/vsx.md: Ditto.
1192 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
1194 * config/rs6000/altivec.md: Don't set length attribute to the default
1195 value, for branch instructions.
1196 * config/rs6000/darwin.md: Ditto.
1197 * config/rs6000/rs6000.md: Ditto.
1199 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
1201 * config/rs6000/rs6000.md (length): Always define as const_int 4.
1202 (unnamed conditional branch define_insn): Set length to 4 or 8
1203 depending on offset.
1204 (<bd>_<mode>): Similar, for alternative 0.
1205 (<bd>tf_<mode>): Ditto.
1207 2018-08-16 Tamar Christina <tamar.christina@arm.com>
1209 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
1211 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
1213 * doc/rtl.texi: Replace old RTX class names with new names.
1216 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
1218 * expmed.h (canonicalize_comparison): New declaration.
1219 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
1220 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
1221 * optabs.c (prepare_cmp_insn): Likewise.
1222 * rtl.h (unsigned_condition_p): New function which checks if a
1223 comparison operator is unsigned.
1225 2018-08-16 Nathan Sidwell <nathan@acm.org>
1227 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
1228 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
1230 2018-08-16 Tamar Christina <tamar.christina@arm.com>
1233 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
1234 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
1235 (mov<mov>): ..this and enable unconditionally.
1237 2018-08-16 Tamar Christina <tamar.christina@arm.com>
1239 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
1241 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
1243 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
1244 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
1245 "Common" with "Target".
1247 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
1249 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
1250 * doc/invoke.texi (mmitigate-rop): Remove.
1251 * config/i386/i386.c: Do not include "regrename.h".
1252 (ix86_rop_should_change_byte_p, reg_encoded_number)
1253 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
1255 (ix86_reorg): Remove call to ix86_mitigate_rop.
1256 * config/i386/i386.md (attr "modrm_class"): Remove.
1257 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
1258 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
1259 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
1261 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
1263 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
1264 allow folding of mergeh() and mergel() for the float and double types.
1265 (fold_mergehl_helper): Rework to handle building a permute tree
1268 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
1270 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
1273 2018-08-15 David Malcolm <dmalcolm@redhat.com>
1275 * common.opt (fdiagnostics-show-labels): New option.
1276 * diagnostic-show-locus.c (class layout_range): Add field
1278 (class layout): Add field "m_show_labels_p".
1279 (layout_range::layout_range): Add param "label" and use it to
1281 (make_range): Pass in NULL for new "label" param of layout_range's
1283 (layout::layout): Initialize m_show_labels_p.
1284 (layout::maybe_add_location_range): Pass in loc_range->m_label
1285 when constructing layout_range instances.
1286 (struct line_label): New struct.
1287 (layout::print_any_labels): New member function.
1288 (layout::print_line): Call it if label-printing is enabled.
1289 (selftest::test_one_liner_labels): New test.
1290 (selftest::test_diagnostic_show_locus_one_liner): Call it.
1291 * diagnostic.c (diagnostic_initialize): Initialize
1292 context->show_labels_p.
1293 * diagnostic.h (struct diagnostic_context): Add field
1295 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
1296 -fno-diagnostics-show-labels.
1297 * dwarf2out.c (gen_producer_string): Add
1298 OPT_fdiagnostics_show_labels to the ignored options.
1299 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
1301 (gcc_rich_location::maybe_add_expr): Likewise.
1302 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
1303 label" param, defaulting to NULL.
1304 (gcc_rich_location::add_expr): Add "label" param.
1305 (gcc_rich_location::maybe_add_expr): Likewise.
1306 (class text_range_label): New class.
1307 (class range_label_for_type_mismatch): New class.
1308 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
1309 of format_warning_va.
1310 (fmtwarn_n): Likewise for new params of format_warning_n_va.
1311 * lto-wrapper.c (merge_and_complain): Add
1312 OPT_fdiagnostics_show_labels to the "pick one setting" options.
1313 (append_compiler_options): Likewise to the dropped options.
1314 (append_diag_options): Likewise to the passed-on options.
1315 * opts.c (common_handle_option): Handle the new option.
1316 * selftest-diagnostic.c
1317 (test_diagnostic_context::test_diagnostic_context): Enable
1319 * substring-locations.c: Include "gcc-rich-location.h".
1320 (format_warning_n_va): Add "fmt_label" and "param_label" params
1321 and use them as appropriate.
1322 (format_warning_va): Add "fmt_label" and "param_label" params,
1323 passing them on to format_warning_n_va.
1324 (format_warning_at_substring): Likewise.
1325 (format_warning_at_substring_n): Likewise.
1326 * substring-locations.h (format_warning_va): Add "fmt_label" and
1327 "param_label" params.
1328 (format_warning_n_va): Likewise.
1329 (format_warning_at_substring): Likewise.
1330 (format_warning_at_substring_n): Likewise.
1331 * toplev.c (general_init): Initialize global_dc->show_labels_p.
1333 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
1336 * builtins.c (expand_builtin_memcmp): Do not expand the call
1337 when overflow is detected.
1339 2018-08-15 Martin Sebor <msebor@redhat.com>
1341 PR tree-optimization/71625
1342 * config/aarch64/aarch64-builtins.c
1343 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
1345 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
1347 * config/s390/s390.c (s390_reorg): Remove loop.
1349 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
1352 (darwin_function_switched_text_sections): Delete.
1353 * gcc/config/darwin.h
1354 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
1356 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
1359 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
1360 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
1361 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
1363 2018-08-15 Martin Liska <mliska@suse.cz>
1365 PR tree-optimization/86925
1366 * predict.c (expr_expected_value_1): When taking
1367 later predictor, assign also probability.
1368 Use fold_build2_initializer_loc in order to fold
1369 the expression in -frounding-math.
1371 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
1373 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
1375 (expand_vec_perm_1): Try the new method.
1377 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
1380 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
1381 Check whether lra_live_max_point is 0 before dividing.
1383 2018-08-14 Martin Sebor <msebor@redhat.com>
1385 PR tree-optimization/86650
1386 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
1387 (vrp_prop::check_mem_ref): Same.
1389 2018-08-13 Liu Hao <lh_mouse@126.com>
1391 * pretty-print.c (eat_esc_sequence): Swap the foreground and
1392 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
1393 and clear it thereafter, as it only works for DBCS.
1395 2018-08-13 Liu Hao <lh_mouse@126.com>
1397 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
1398 handle returned by _get_osf_handle().
1400 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
1402 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
1403 for folding vec_perm.
1405 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
1407 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
1408 Add support for gimple-folding of vec_pack() and vec_unpack()
1411 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
1413 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
1414 vec_xst variants to the list.
1415 (rs6000_gimple_fold_builtin): Add support for folding unaligned
1416 vector loads and stores.
1418 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
1420 * config.gcc (rs6000-ibm-aix4.x): Delete.
1421 (rs6000-ibm-aix5.1): Delete.
1422 (rs6000-ibm-aix5.2): Delete.
1423 (rs6000-ibm-aix5.3): Delete.
1424 * config/rs6000/aix43.h: Delete.
1425 * config/rs6000/aix51.h: Delete.
1426 * config/rs6000/aix52.h: Delete.
1427 * config/rs6000/t-aix43: Delete.
1429 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
1431 * config/s390/s390.c (s390_decompose_constant_pool_ref):
1433 (s390_decompose_address): Factor out constant pool ref
1436 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
1438 * config/nds32/nds32-predicates.c
1439 (nds32_can_use_bclr_p): Change return type as bool.
1440 (nds32_can_use_bset_p): Ditto.
1441 (nds32_can_use_btgl_p): Ditto.
1442 (nds32_can_use_bitci_p): Ditto.
1443 * config/nds32/nds32-protos.h
1444 (nds32_can_use_bclr_p): Change declaration.
1445 (nds32_can_use_bset_p): Ditto.
1446 (nds32_can_use_btgl_p): Ditto.
1447 (nds32_can_use_bitci_p): Ditto.
1449 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
1451 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
1452 Support -msched-prolog-epilog option.
1453 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
1455 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
1457 * common/config/nds32/nds32-common.c
1458 (nds32_option_optimization_table): Enalbe -malways-align.
1460 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
1462 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
1464 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
1465 OPT_misr_secure_ case.
1466 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
1467 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
1468 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
1470 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
1471 (nds32_isr_info): New field security_level.
1472 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
1473 * config/nds32/nds32.md (return_internal): Consider critical attribute.
1474 * config/nds32/nds32.opt (misr-secure): New option.
1475 * config/nds32/nds32_init.inc: New file.
1476 * config/nds32/nds32_isr.h: New file.
1478 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
1480 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
1481 Update comment for atomic instructions.
1482 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
1483 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
1485 (atomic_loaddi): Revise fence expansion to only emit fence prior to
1486 load for __ATOMIC_SEQ_CST model.
1487 (atomic_loaddi_1): Remove float register target.
1488 (atomic_storedi): Handle CONST_INT values.
1489 (atomic_storedi_1): Remove float register source. Add special case
1491 (memory_barrier): New expander and insn.
1493 2018-08-11 Jakub Jelinek <jakub@redhat.com>
1495 PR tree-optimization/86835
1496 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
1497 new_stmt after def_gsi, make sure to insert new_square_stmt after
1498 that stmt, not 2 stmts before it.
1500 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
1503 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
1506 2018-08-10 Martin Liska <mliska@suse.cz>
1509 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
1511 * builtins.c (expand_builtin_expect_with_probability):
1513 (expand_builtin_expect_with_probability): New function.
1514 (build_builtin_expect_predicate): Add new argumnet probability
1515 for BUILT_IN_EXPECT_WITH_PROBABILITY.
1516 (fold_builtin_expect):
1519 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
1520 * builtins.h (fold_builtin_expect): Set new argument.
1521 * doc/extend.texi: Document __builtin_expect_with_probability.
1522 * doc/invoke.texi: Likewise.
1523 * gimple-fold.c (gimple_fold_call): Pass new argument.
1524 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
1525 also BUILT_IN_EXPECT_WITH_PROBABILITY.
1526 * predict.c (get_predictor_value): New function.
1527 (expr_expected_value): Add new argument probability. Assume
1528 that predictor and probability are always non-null.
1529 (expr_expected_value_1): Likewise. For __builtin_expect and
1530 __builtin_expect_with_probability set probability. Handle
1531 combination in binary expressions.
1532 (tree_predict_by_opcode): Simplify code by simply calling
1533 get_predictor_value.
1534 (pass_strip_predict_hints::execute): Add handling of
1535 BUILT_IN_EXPECT_WITH_PROBABILITY.
1536 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
1538 * tree.h (DECL_BUILT_IN_P): New function.
1540 2018-08-10 Martin Liska <mliska@suse.cz>
1542 PR tree-optimization/85799
1543 * passes.def: Add argument for pass_strip_predict_hints.
1544 * predict.c (class pass_strip_predict_hints): Add new argument
1546 (strip_predictor_early): New function.
1547 (pass_strip_predict_hints::execute): Call the function to
1549 (strip_predict_hints): New function.
1550 * predict.def: Fix comment.
1552 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
1554 * Makefile.in: Clarify which tm.texi to copy over to assert the
1555 right to grant a GFDL license for all.
1557 2018-08-09 Jeff Law <law@redhat.com>
1559 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
1562 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
1564 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
1567 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
1569 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
1570 reductions for variable-length vectors.
1572 2018-08-09 David Malcolm <dmalcolm@redhat.com>
1575 * common.opt (fdiagnostics-show-line-numbers): New option.
1576 * diagnostic-show-locus.c (class layout): Add fields
1577 "m_show_line_numbers_p" and "m_linenum_width";
1578 (num_digits): New function.
1579 (test_num_digits): New function.
1580 (layout::layout): Initialize new fields. Update m_x_offset
1581 logic to handle any left margin.
1582 (layout::print_source_line): Print line number when requested.
1583 (layout::start_annotation_line): New member function.
1584 (layout::print_annotation_line): Call it.
1585 (layout::print_leading_fixits): Likewise.
1586 (layout::print_trailing_fixits): Likewise. Update calls to
1587 move_to_column for new parameter.
1588 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
1589 it to potentially call start_annotation_line.
1590 (layout::show_ruler): Call start_annotation_line.
1591 (selftest::test_line_numbers_multiline_range): New selftest.
1592 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
1593 and selftest::test_line_numbers_multiline_range.
1594 * diagnostic.c (diagnostic_initialize): Initialize
1595 show_line_numbers_p.
1596 * diagnostic.h (struct diagnostic_context): Add field
1597 "show_line_numbers_p".
1598 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
1599 -fno-diagnostics-show-line-numbers.
1600 * dwarf2out.c (gen_producer_string): Add
1601 OPT_fdiagnostics_show_line_numbers to the ignored options.
1602 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
1603 one setting" options.
1604 (append_compiler_options): Likewise to the dropped options.
1605 (append_diag_options): Likewise to the passed-on options.
1606 * opts.c (common_handle_option): Handle the new option.
1607 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
1609 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
1611 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
1612 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
1613 third argument of type "const signed char" to descriptions of
1614 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
1615 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
1616 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
1617 __builtin_bcdsub_ov functions.
1619 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
1621 PR tree-optimization/86858
1622 * tree-vect-loop.c (vect_is_simple_reduction): Restore
1623 flow_bb_inside_loop_p calls.
1625 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
1627 PR tree-optimization/86871
1628 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
1629 instead of gimple_assign_lhs.
1631 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
1634 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
1635 register constraint to operand 0.
1636 (add<mode>3_carryinC): Likewise.
1637 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
1639 2018-08-09 Martin Liska <mliska@suse.cz>
1642 * common.opt: Remove extra line.
1644 2018-08-09 Martin Liska <mliska@suse.cz>
1646 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
1647 at the end of a line, make first letter capital and end up
1648 a sentence with a dot.
1649 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
1650 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
1651 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
1652 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
1653 (PARAM_MAX_ISL_OPERATIONS): Likewise.
1654 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
1655 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
1656 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
1657 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
1658 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
1659 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
1660 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
1661 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
1662 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
1663 (PARAM_TREE_REASSOC_WIDTH): Likewise.
1664 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
1665 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
1666 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
1668 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
1671 * config/s390/s390.c (s390_option_override_internal): Reduce the
1672 stack-clash-protection-probe-interval param if it would be too big
1675 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
1678 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
1679 * config/m68k/m68k.c (handle_move_double): Don't call
1680 m68k_final_prescan_insn.
1681 (m68k_adjust_decorated_operand): Renamed from
1682 m68k_final_prescan_insn, remove first and third operand and
1684 (print_operand): Call it.
1685 (print_operand_address): Call it.
1687 2018-08-08 Nathan Sidwell <nathan@acm.org>
1689 * diagnostic.c (diagnostic_report_current_module): Use
1690 linemap_included_from & linemap_included_from_linemap.
1692 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
1694 * config/aarch64/aarch64-cores.def: Add phecda core.
1695 * config/aarch64/aarch64-tune.md: Regenerate.
1696 * doc/invoke.texi: Add phecda core.
1698 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
1701 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
1703 * config/s390/s390.md ("movti"): Add more alternatives for
1704 constant to GPR copies.
1706 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
1708 * config/s390/s390.c: Fix whitespace damage throughout the file.
1709 * config/s390/s390.h: Likewise.
1710 * config/s390/tpf.h: Likewise.
1712 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
1714 * config/s390/s390.c (s390_loadrelative_operand_p):
1715 Remove TARGET_CPU_ZARCH usages.
1716 (s390_rtx_costs): Likewise.
1717 (s390_legitimate_constant_p): Likewise.
1718 (s390_cannot_force_const_mem): Likewise.
1719 (legitimate_reload_constant_p): Likewise.
1720 (s390_preferred_reload_class): Likewise.
1721 (legitimize_pic_address): Likewise.
1722 (legitimize_tls_address): Likewise.
1723 (s390_split_branches): Removed.
1724 (s390_add_execute): Removed.
1725 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
1726 (s390_mainpool_start): Likewise.
1727 (s390_mainpool_finish): Likewise.
1728 (s390_mainpool_cancel): Removed.
1729 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
1730 (s390_chunkify_cancel): Likewise.
1731 (s390_return_addr_rtx): Likewise.
1732 (s390_register_info): Remove split_branches_pending_p uages.
1733 (s390_optimize_register_info): Likewise.
1734 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
1735 split_branches_pending_p usages.
1736 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
1737 (s390_load_got): Likewise.
1738 (s390_expand_split_stack_prologue): Likewise.
1739 (output_asm_nops): Likewise.
1740 (s390_function_profiler): Likewise.
1741 (s390_emit_call): Likewise.
1742 (s390_conditional_register_usage): Likewise.
1743 (s390_optimize_prologue): Likewise.
1744 (s390_reorg): Remove TARGET_CPU_ZARCH and
1745 split_branches_pending_p usages.
1746 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
1748 (s390_output_indirect_thunk_function): Likewise.
1749 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
1750 (TARGET_CPU_ZARCH_P): Removed.
1751 (struct machine_function): Remove split_branches_pending_p.
1752 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
1754 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
1756 * common/config/s390/s390-common.c (processor_flags_table):
1758 * config.gcc: Remove with_arch/with_tune support.
1759 * config/s390/2064.md: Remove cpu attribute comparisons.
1760 * config/s390/driver-native.c (s390_host_detect_local_cpu):
1762 * config/s390/linux.h (ASM_SPEC):
1763 Remove -march support.
1764 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
1765 Use a table to get an arch level.
1766 * config/s390/s390-opts.h (enum processor_type):
1768 * config/s390/s390.c
1769 (processor_table): Remove entries, add arch_level values.
1770 (s390_issue_rate): Remove cases.
1771 (s390_option_override): Adjust
1772 s390_option_override_internal() call.
1773 (s390_option_override_internal): Remove deprecation warning.
1774 (s390_valid_target_attribute_tree): Adjust
1775 s390_option_override_internal() call.
1776 * config/s390/s390.h (struct s390_processor):
1777 Share with s390-c.c, add arch_level field.
1778 * config/s390/s390.md:
1779 Remove occurrences in cpu attribute.
1780 * config/s390/s390.opt: Remove -march/-mtune support.
1781 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
1782 * doc/invoke.texi: Remove deprecation warning.
1784 2018-08-08 Luis Machado <luis.machado@linaro.org>
1786 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
1788 (qdf24xx_tunings): Set vector cost structure to
1789 qdf24xx_vector_cost.
1791 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
1792 <register_sextend>: Set to 3.
1794 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
1797 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
1798 * config/aarch64/aarch64-simd.md
1799 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
1800 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
1801 (aarch64_frecpx<mode>): ...this new pattern.
1802 * config/aarch64/aarch64-simd-builtins.def: Remove comment
1803 about aarch64_frecp<FRECP:frecp_suffix><mode>.
1805 2018-08-07 Martin Liska <mliska@suse.cz>
1808 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
1809 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
1810 * predict.def (PRED_MALLOC_NONNULL): New predictor.
1811 * doc/extend.texi: Document that malloc attribute adds
1814 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
1817 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
1818 Define to speculation_safe_value_not_needed.
1820 2018-08-06 Jeff Law <law@redhat.com>
1822 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
1823 the vr_values instance to cprop_into_stmt.
1824 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
1825 (cprop_operand): Also query EVRP to determine if OP is a constant.
1827 2018-08-06 Nathan Sidwell <nathan@acm.org>
1829 * diagnostic.c (diagnostic_report_current_module): Reroll
1830 included-at loop. Translate text.
1832 2018-08-06 David Malcolm <dmalcolm@redhat.com>
1834 * function-tests.c (selftest::test_expansion_to_rtl): Call
1835 free_after_compilation.
1837 2018-08-06 Alan Hayward <alan.hayward@arm.com>
1839 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
1841 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
1843 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
1844 loops with memory block operations from getting unrolled.
1846 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
1849 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
1850 Define to speculation_safe_value_not_needed.
1852 2018-08-06 Jeff Law <law@redhat.com>
1854 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
1857 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
1860 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
1861 with all enabled __intN types.
1863 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
1865 2018-08-06 Alan Hayward <alan.hayward@arm.com>
1867 * alias.c (record_set): Check for clobber high.
1868 * cfgexpand.c (expand_gimple_stmt): Likewise.
1869 * combine-stack-adj.c (single_set_for_csa): Likewise.
1870 * combine.c (find_single_use_1): Likewise.
1871 (set_nonzero_bits_and_sign_copies): Likewise.
1872 (get_combine_src_dest): Likewise.
1873 (is_parallel_of_n_reg_sets): Likewise.
1874 (try_combine): Likewise.
1875 (record_dead_and_set_regs_1): Likewise.
1876 (reg_dead_at_p_1): Likewise.
1877 (reg_dead_at_p): Likewise.
1878 * dce.c (deletable_insn_p): Likewise.
1879 (mark_nonreg_stores_1): Likewise.
1880 (mark_nonreg_stores_2): Likewise.
1881 * df-scan.c (df_find_hard_reg_defs): Likewise.
1882 (df_uses_record): Likewise.
1883 (df_get_call_refs): Likewise.
1884 * dwarf2out.c (mem_loc_descriptor): Likewise.
1885 * haifa-sched.c (haifa_classify_rtx): Likewise.
1886 * ira-build.c (create_insn_allocnos): Likewise.
1887 * ira-costs.c (scan_one_insn): Likewise.
1888 * ira.c (equiv_init_movable_p): Likewise.
1889 (rtx_moveable_p): Likewise.
1890 (interesting_dest_for_shprep): Likewise.
1891 * jump.c (mark_jump_label_1): Likewise.
1892 * postreload-gcse.c (record_opr_changes): Likewise.
1893 * postreload.c (reload_cse_simplify): Likewise.
1894 (struct reg_use): Add source expr.
1895 (reload_combine): Check for clobber high.
1896 (reload_combine_note_use): Likewise.
1897 (reload_cse_move2add): Likewise.
1898 (move2add_note_store): Likewise.
1899 * print-rtl.c (print_pattern): Likewise.
1900 * recog.c (decode_asm_operands): Likewise.
1901 (store_data_bypass_p): Likewise.
1902 (if_test_bypass_p): Likewise.
1903 * regcprop.c (kill_clobbered_value): Likewise.
1904 (kill_set_value): Likewise.
1905 * reginfo.c (reg_scan_mark_refs): Likewise.
1906 * reload1.c (maybe_fix_stack_asms): Likewise.
1907 (eliminate_regs_1): Likewise.
1908 (elimination_effects): Likewise.
1909 (mark_not_eliminable): Likewise.
1910 (scan_paradoxical_subregs): Likewise.
1911 (forget_old_reloads_1): Likewise.
1912 * reorg.c (find_end_label): Likewise.
1913 (try_merge_delay_insns): Likewise.
1914 (redundant_insn): Likewise.
1915 (own_thread_p): Likewise.
1916 (fill_simple_delay_slots): Likewise.
1917 (fill_slots_from_thread): Likewise.
1918 (dbr_schedule): Likewise.
1919 * resource.c (update_live_status): Likewise.
1920 (mark_referenced_resources): Likewise.
1921 (mark_set_resources): Likewise.
1922 * rtl.c (copy_rtx): Likewise.
1923 * rtlanal.c (reg_referenced_p): Likewise.
1924 (single_set_2): Likewise.
1925 (noop_move_p): Likewise.
1926 (note_stores): Likewise.
1927 * sched-deps.c (sched_analyze_reg): Likewise.
1928 (sched_analyze_insn): Likewise.
1930 2018-08-06 Alan Hayward <alan.hayward@arm.com>
1932 * cse.c (invalidate_reg): New function extracted from...
1933 (invalidate): ...here.
1934 (canonicalize_insn): Check for clobber high.
1935 (invalidate_from_clobbers): invalidate clobber highs.
1936 (invalidate_from_sets_and_clobbers): Likewise.
1937 (count_reg_usage): Check for clobber high.
1938 (insn_live_p): Likewise.
1939 * cselib.c (cselib_expand_value_rtx_1):Likewise.
1940 (cselib_invalidate_regno): Check for clobber in setter.
1941 (cselib_invalidate_rtx): Pass through setter.
1942 (cselib_invalidate_rtx_note_stores):
1943 (cselib_process_insn): Check for clobber high.
1944 * cselib.h (cselib_invalidate_rtx): Add operand.
1946 2018-08-06 Alan Hayward <alan.hayward@arm.com>
1948 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
1949 (mark_not_eliminable): Likewise.
1950 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
1951 * lra-lives.c (process_bb_lives): Check for clobber high.
1952 * lra.c (new_insn_reg): Remember clobber highs.
1953 (collect_non_operand_hard_regs): Check for clobber high.
1954 (lra_set_insn_recog_data): Likewise.
1955 (add_regs_to_insn_regno_info): Likewise.
1956 (lra_update_insn_regno_info): Likewise.
1958 2018-08-06 Alan Hayward <alan.hayward@arm.com>
1960 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
1961 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
1963 2018-08-06 Alan Hayward <alan.hayward@arm.com>
1965 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
1966 (copy_insn_1): Likewise.
1967 (gen_hard_reg_clobber_high): New gen function.
1968 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
1969 * genemit.c (gen_exp): Likewise.
1970 (gen_emit_seq): Pass through info.
1971 (gen_insn): Check for CLOBBER_HIGH.
1972 (gen_expand): Pass through info.
1973 (gen_split): Likewise.
1974 (output_add_clobbers): Likewise.
1975 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
1976 (remove_clobbers): Likewise.
1977 * rtl.h (gen_hard_reg_clobber_high): New declaration.
1979 2018-08-06 Alan Hayward <alan.hayward@arm.com>
1981 * doc/rtl.texi (clobber_high): Add.
1982 (parallel): Add in clobber high
1983 * rtl.c (rtl_check_failed_code3): Add function.
1984 * rtl.def (CLOBBER_HIGH): Add expression.
1985 * rtl.h (RTL_CHECKC3): Add macro.
1986 (rtl_check_failed_code3): Add declaration.
1987 (XC3EXP): Add macro.
1989 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
1992 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
1993 cfun->machine->max_used_stack_alignment if needed.
1995 2018-08-04 Martin Sebor <msebor@redhat.com>
1997 PR tree-optimization/86571
1998 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
2001 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
2004 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
2007 2018-08-03 Jeff Law <law@redhat.com>
2010 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
2011 Define to speculation_safe_value_not_needed.
2013 2018-08-03 David Malcolm <dmalcolm@redhat.com>
2015 * doc/gcov.texi (-x): Remove duplicate "to".
2016 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
2017 (-Wif-not-aligned): Remove duplicate "is".
2018 (-flto): Remove duplicate "the".
2019 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
2020 duplicate "v5.00.b".
2021 (MSP430 Options): Remove duplicate "and" from the description
2022 of "-mgprel-sec=regexp".
2023 (x86 Options): Remove duplicate copies of "vmldLog102" and
2024 vmlsLog104 from description of "-mveclibabi=type".
2026 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
2028 * internal-fn.h (first_commutative_argument): Declare.
2029 * internal-fn.c (first_commutative_argument): New function.
2030 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
2031 restrictions for pattern statements. Use first_commutative_argument
2032 to look for commutative operands in calls to internal functions.
2034 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
2036 * Makefile.in (wide-int-range.o): New.
2037 * tree-vrp.c: Move all the wide_int_* functions to...
2038 * wide-int-range.cc: ...here.
2039 * tree-vrp.h: Move all the wide_int_* prototypes to...
2040 * wide-int-range.h: ...here.
2042 2018-08-03 Tom de Vries <tdevries@suse.de>
2044 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
2046 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
2047 * except.c (output_function_exception_table): Do early exit if
2048 targetm_common.except_unwind_info (&global_options) == UI_NONE.
2050 2018-08-03 Martin Liska <mliska@suse.cz>
2052 * predict.c (dump_prediction): Change to 2 digits
2055 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
2057 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
2059 (falkor_am_1_gtov_gtov): ... a new insn reservation.
2061 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
2063 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
2064 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
2065 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
2066 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
2067 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
2069 2018-08-02 David Malcolm <dmalcolm@redhat.com>
2071 * diagnostic-show-locus.c (diagnostic_show_locus): Use
2072 pp_take_prefix when saving the existing prefix.
2073 * diagnostic.c (diagnostic_append_note): Likewise.
2074 * langhooks.c (lhd_print_error_function): Likewise.
2075 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
2076 param's type. Free the existing prefix.
2077 (pp_take_prefix): New function.
2078 (pretty_printer::pretty_printer): Drop the prefix parameter.
2079 Rename the length parameter to match the comment.
2080 (pretty_printer::~pretty_printer): Free the prefix.
2081 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
2083 (struct pretty_printer): Drop the "const" from "prefix" field's
2084 type and clarify memory management.
2085 (pp_set_prefix): Drop the "const" from the 2nd param.
2086 (pp_take_prefix): New decl.
2088 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
2090 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
2091 for word_mode_ok here instead of passing as argument.
2092 (expand_block_compare): Change select_block_compare_mode() call.
2093 (expand_strncmp_gpr_sequence): New function.
2094 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
2096 2018-08-02 Jeff Law <law@redhat.com>
2099 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
2100 Define to speculation_safe_value_not_needed.
2103 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
2104 Define to speculation_safe_value_not_needed.
2106 2018-08-02 Tom de Vries <tdevries@suse.de>
2109 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
2110 function. Return UI_TARGET unconditionally.
2111 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
2112 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
2114 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
2116 * genemit.c (print_overload_test): Fix typo.
2118 2018-08-02 Richard Biener <rguenther@suse.de>
2120 PR tree-optimization/86816
2121 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
2122 which checks for value availability before querying it.
2123 (gvn_uses_equal): Use it.
2124 (same_succ_hash): Likewise.
2125 (gimple_equal_p): Likewise.
2127 2018-08-02 Nick Clifton <nickc@redhat.com>
2130 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
2131 Define to speculation_safe_value_not_needed.
2134 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
2135 Define to speculation_safe_value_not_needed.
2138 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
2139 Define to speculation_safe_value_not_needed.
2142 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
2143 Define to speculation_safe_value_not_needed.
2146 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
2147 Define to speculation_safe_value_not_needed.
2150 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
2151 Define to speculation_safe_value_not_needed.
2154 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
2155 Define to speculation_safe_value_not_needed.
2158 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
2159 Define to speculation_safe_value_not_needed.
2162 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
2163 speculation_safe_value_not_needed.
2166 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
2167 to speculation_safe_value_not_needed.
2169 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
2171 * doc/md.texi: Expand the documentation of instruction names
2172 to mention port-local uses. Document '@' in pattern names.
2173 * read-md.h (overloaded_instance, overloaded_name): New structs.
2175 (md_reader::handle_overloaded_name): New member function.
2176 (md_reader::get_overloads): Likewise.
2177 (md_reader::m_first_overload): New member variable.
2178 (md_reader::m_next_overload_ptr): Likewise.
2179 (md_reader::m_overloads_htab): Likewise.
2180 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
2181 m_next_overload_ptr and m_overloads_htab.
2182 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
2183 (get_mode_token, get_code_token, get_int_token): New functions.
2184 (map_attr_string): Add an optional argument that passes back
2185 the associated iterator.
2186 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
2187 (md_reader::handle_overloaded_name, add_overload_instance): New
2189 (apply_iterators): Handle '@' names. Report an error if '@'
2190 is used without iterators.
2191 (initialize_iterators): Initialize the new iterator_group fields.
2192 * genopinit.c (handle_overloaded_code_for)
2193 (handle_overloaded_gen): New functions.
2194 (main): Use them to print declarations of maybe_code_for_* and
2195 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
2196 * genemit.c (print_overload_arguments, print_overload_test)
2197 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
2198 (main): Use it to print definitions of maybe_code_for_* and
2199 maybe_gen_* functions.
2200 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
2201 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
2202 instead of explicit mode checks.
2203 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
2204 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
2205 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
2206 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
2207 (aarch64_expand_compare_and_swap): Likewise
2208 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
2209 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
2210 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
2211 (aarch64_constant_pool_reload_icode): Delete.
2212 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
2213 instead of aarch64_constant_pool_reload_icode. Use
2214 code_for_aarch64_reload_mov instead of explicit mode checks.
2215 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
2216 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
2217 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
2218 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
2219 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
2220 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
2221 (aarch64_atomic_load_op_code): Delete.
2222 (aarch64_emit_atomic_load_op): Likewise.
2223 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
2224 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
2225 instead of aarch64_emit_atomic_load_op.
2226 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
2227 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
2228 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
2229 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
2230 character before the pattern name.
2231 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
2232 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
2233 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
2234 (aarch64_frecps<mode>): Likewise.
2235 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
2236 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
2237 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
2238 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
2239 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
2241 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
2243 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
2244 Allow HFmode constants if TARGET_FP_F16INST.
2246 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
2249 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
2250 No longer check last store for clobber of address register.
2252 2018-08-02 Martin Liska <mliska@suse.cz>
2254 PR gcov-profile/86817
2255 * gcov.c (process_all_functions): New function.
2257 (process_file): Move functions processing to
2258 process_all_functions.
2260 2018-08-02 David Malcolm <dmalcolm@redhat.com>
2262 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
2263 "const" to the "gimple *" and "rtx_insn *" parameters.
2264 * dumpfile.h (dump_user_location_t::dump_user_location_t):
2266 (dump_location_t::dump_location_t): Likewise.
2268 2018-08-01 Martin Sebor <msebor@redhat.com>
2270 PR tree-optimization/86650
2271 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
2272 rather than a "gcall *". Directly pass the data of interest
2273 to percent_K_format, rather than building a temporary CALL_EXPR
2275 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
2276 (gimple_fold_builtin_strncat): Adjust.
2277 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
2278 gcall* argument with gimple*.
2279 * gimple-ssa-warn-restrict.c (check_call): Same.
2280 (wrestrict_dom_walker::before_dom_children): Same.
2281 (builtin_access::builtin_access): Same.
2282 (check_bounds_or_overlap): Same
2283 (maybe_diag_overlap): Same.
2284 (maybe_diag_offset_bounds): Same.
2285 * tree-diagnostic.c (default_tree_printer): Move usage of
2286 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
2288 * tree-pretty-print.c (percent_K_format): Add argument.
2289 * tree-pretty-print.h: Add argument.
2290 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
2291 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
2292 (maybe_diag_stxncpy_trunc): Same.
2293 (handle_builtin_stxncpy): Same.
2294 (handle_builtin_strcat): Same.
2296 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
2298 * match.pd: Optimise pointer range checks.
2300 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
2302 PR tree-optimization/86758
2303 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
2304 to remove pattern statements.
2306 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
2308 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
2309 result of dfs_enumerate_from when constructing stmt_vec_infos,
2310 instead of additionally calling get_loop_body.
2312 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
2314 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
2316 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
2317 When creating an iv, assert that the step is not known to be zero.
2318 (vect_setup_realignment): Update call accordingly.
2319 * tree-vect-stmts.c (vectorizable_store): Likewise.
2320 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
2322 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
2324 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
2325 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
2326 (vectorizable_reduction): Likewise.
2327 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
2328 (vect_detect_hybrid_slp_stmts): Likewise.
2329 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
2331 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
2333 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
2334 (wide_int_set_zero_nonzero_bits): ...this.
2335 (zero_nonzero_bits_from_vr): Rename to...
2336 (vrp_set_zero_nonzero_bits): ...this.
2337 (extract_range_from_multiplicative_op_1): Abstract wide int
2339 (wide_int_range_multiplicative_op): ...here.
2340 (extract_range_from_binary_expr_1): Extract wide int binary
2341 operations into their own functions.
2342 (wide_int_range_lshift): New.
2343 (wide_int_range_can_optimize_bit_op): New.
2344 (wide_int_range_shift_undefined_p): New.
2345 (wide_int_range_bit_xor): New.
2346 (wide_int_range_bit_ior): New.
2347 (wide_int_range_bit_and): New.
2348 (wide_int_range_trunc_mod): New.
2349 (extract_range_into_wide_ints): New.
2350 (vrp_shift_undefined_p): New.
2351 (extract_range_from_multiplicative_op): New.
2352 (vrp_can_optimize_bit_op): New.
2353 * tree-vrp.h (value_range::dump): New.
2354 (wide_int_range_multiplicative_op): New.
2355 (wide_int_range_lshift):New.
2356 (wide_int_range_shift_undefined_p): New.
2357 (wide_int_range_bit_xor): New.
2358 (wide_int_range_bit_ior): New.
2359 (wide_int_range_bit_and): New.
2360 (wide_int_range_trunc_mod): New.
2361 (zero_nonzero_bits_from_bounds): Rename to...
2362 (wide_int_set_zero_nonzero_bits): ...this.
2363 (zero_nonzero_bits_from_vr): Rename to...
2364 (vrp_set_zero_nonzero_bits): ...this.
2365 (range_easy_mask_min_max): Rename to...
2366 (wide_int_range_can_optimize_bit_op): this.
2367 * vr-values.c (simplify_bit_ops_using_ranges): Rename
2368 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
2370 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
2372 * tree-vectorizer.h (vect_orig_stmt): New function.
2373 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
2374 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
2375 (vect_create_epilog_for_reduction): Likewise.
2376 (vectorizable_live_operation): Likewise.
2377 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
2378 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
2379 * tree-vect-stmts.c (vectorizable_call): Likewise.
2380 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
2382 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
2384 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
2386 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
2387 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
2388 (vect_transform_loop): Likewise.
2389 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
2391 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
2393 * tree-vectorizer.h (vect_schedule_slp): Return void.
2394 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
2395 (vect_schedule_slp): Likewise.
2397 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
2399 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
2401 (vect_transform_loop): Update calls accordingly. Schedule SLP
2402 instances before the main loop, if any exist.
2404 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
2406 PR tree-optimization/86749
2407 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
2408 If the lhs is used in a COND_EXPR, check that it is being used
2409 as the "then" or "else" value.
2411 2018-08-01 Tom de Vries <tdevries@suse.de>
2414 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
2415 speculation_safe_value_not_needed.
2417 2018-08-01 Richard Biener <rguenther@suse.de>
2419 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
2422 2018-08-01 Martin Liska <mliska@suse.cz>
2424 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
2425 format how successful transformation is dumped.
2426 (gimple_mod_pow2_value_transform): Likewise.
2427 (gimple_mod_subtract_transform): Likewise.
2428 (gimple_stringops_transform): Likewise.
2430 2018-08-01 Martin Liska <mliska@suse.cz>
2433 * value-prof.c (interesting_stringop_to_profile_p):
2434 Simplify the code and add BUILT_IN_MEMMOVE.
2435 (gimple_stringops_transform): Likewise.
2437 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
2439 * config/aarch64/aarch64-simd.md
2440 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
2441 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
2442 use GPI iterator instead of SI mode.
2444 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
2446 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
2447 rs6000_speculation_barrier.
2448 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
2449 new barrier pattern name.
2451 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
2453 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
2454 (speculation_barrier): New insn.
2456 2018-08-01 Richard Biener <rguenther@suse.de>
2459 * graphite.h: Include isl/id.h and isl/space.h to allow build
2462 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
2465 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
2466 mode for COFF targets.
2467 * defaults.h (TARGET_COFF): Define.
2468 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
2469 TARGET_COFF): Define.
2470 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
2471 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
2472 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
2474 2018-07-31 Alexandre Oliva <oliva@adacore.com>
2476 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
2477 * gimple-streamer-out.c (output_bb): Save it.
2478 * lto-streamer-in.c (input_struct_function_base): Restore
2479 instance discriminator if available. Create map on demand.
2480 * lto-streamer-out.c (output_struct_function_base): Save it if
2482 * final.c (decl_to_instance_map): Document LTO strategy.
2484 2018-07-31 Alexandre Oliva <oliva@adacore.com>
2485 Olivier Hainque <hainque@adacore.com>
2487 * debug.h (decl_to_instance_map_t): New type.
2488 (decl_to_instance_map): Declare.
2489 (maybe_create_decl_to_instance_map): New inline function.
2490 * final.c (bb_discriminator, last_bb_discriminator): New statics,
2491 to track basic block discriminators.
2492 (final_start_function_1): Initialize them.
2493 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
2495 (decl_to_instance_map): New variable.
2496 (map_decl_to_instance, maybe_set_discriminator): New functions.
2497 (notice_source_line): Set discriminator.
2499 2018-07-31 Ian Lance Taylor <iant@golang.org>
2501 * targhooks.c (default_have_speculation_safe_value): Add
2504 2018-07-31 David Malcolm <dmalcolm@redhat.com>
2506 * dump-context.h: Include "pretty-print.h".
2507 (dump_context::refresh_dumps_are_enabled): New decl.
2508 (dump_context::emit_item): New decl.
2509 (class dump_context): Add fields "m_test_pp" and
2511 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
2512 (temp_dump_context::get_dumped_text): New decl.
2513 (class temp_dump_context): Add field "m_pp".
2514 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
2515 (dump_context::refresh_dumps_are_enabled): ...and add a test for
2517 (set_dump_file): Update for above change.
2518 (set_alt_dump_file): Likewise.
2519 (dump_loc): New overload, taking a pretty_printer *.
2520 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
2521 to any test pretty-printer.
2522 (make_item_for_dump_gimple_stmt): New function, adapted from
2523 optinfo::add_gimple_stmt.
2524 (dump_context::dump_gimple_stmt): Call it, and use the result,
2525 eliminating the direct usage of dump_file and alt_dump_file in
2526 favor of indirectly using them via emit_item.
2527 (make_item_for_dump_gimple_expr): New function, adapted from
2528 optinfo::add_gimple_expr.
2529 (dump_context::dump_gimple_expr): Call it, and use the result,
2530 eliminating the direct usage of dump_file and alt_dump_file in
2531 favor of indirectly using them via emit_item.
2532 (make_item_for_dump_generic_expr): New function, adapted from
2534 (dump_context::dump_generic_expr): Call it, and use the result,
2535 eliminating the direct usage of dump_file and alt_dump_file in
2536 favor of indirectly using them via emit_item.
2537 (make_item_for_dump_printf_va): New function, adapted from
2538 optinfo::add_printf_va.
2539 (make_item_for_dump_printf): New function.
2540 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
2541 and use the result, eliminating the direct usage of dump_file and
2542 alt_dump_file in favor of indirectly using them via emit_item.
2543 (make_item_for_dump_dec): New function.
2544 (dump_context::dump_dec): Call it, and use the result,
2545 eliminating the direct usage of dump_file and alt_dump_file in
2546 favor of indirectly using them via emit_item.
2547 (make_item_for_dump_symtab_node): New function, adapted from
2548 optinfo::add_symtab_node.
2549 (dump_context::dump_symtab_node): Call it, and use the result,
2550 eliminating the direct usage of dump_file and alt_dump_file in
2551 favor of indirectly using them via emit_item.
2552 (dump_context::begin_scope): Reimplement, avoiding direct usage
2553 of dump_file and alt_dump_file in favor of indirectly using them
2555 (dump_context::emit_item): New member function.
2556 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
2557 Set up test pretty-printer on the underlying context. Call
2558 refresh_dumps_are_enabled.
2559 (temp_dump_context::~temp_dump_context): Call
2560 refresh_dumps_are_enabled.
2561 (temp_dump_context::get_dumped_text): New member function.
2562 (selftest::verify_dumped_text): New function.
2563 (ASSERT_DUMPED_TEXT_EQ): New macro.
2564 (selftest::test_capture_of_dump_calls): Run all tests twice, with
2565 and then without optinfo enabled. Add uses of
2566 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
2567 * dumpfile.h: Update comment for the dump_* API.
2568 * optinfo-emit-json.cc
2569 (selftest::test_building_json_from_dump_calls): Update for new
2570 param for temp_dump_context ctor.
2571 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
2572 and "m_owned" field.
2573 (optinfo_item::~optinfo_item): Likewise.
2574 (optinfo::add_item): New member function.
2575 (optinfo::emit): Update comment.
2576 (optinfo::add_string): Delete.
2577 (optinfo::add_printf): Delete.
2578 (optinfo::add_printf_va): Delete.
2579 (optinfo::add_gimple_stmt): Delete.
2580 (optinfo::add_gimple_expr): Delete.
2581 (optinfo::add_tree): Delete.
2582 (optinfo::add_symtab_node): Delete.
2583 (optinfo::add_dec): Delete.
2584 * optinfo.h (class dump_context): New forward decl.
2585 (optinfo::add_item): New decl.
2586 (optinfo::add_string): Delete.
2587 (optinfo::add_printf): Delete.
2588 (optinfo::add_printf_va): Delete.
2589 (optinfo::add_gimple_stmt): Delete.
2590 (optinfo::add_gimple_expr): Delete.
2591 (optinfo::add_tree): Delete.
2592 (optinfo::add_symtab_node): Delete.
2593 (optinfo::add_dec): Delete.
2594 (optinfo::add_poly_int): Delete.
2595 (optinfo_item::optinfo_item): Remove "owned" param.
2596 (class optinfo_item): Remove field "m_owned".
2598 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
2601 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
2602 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
2603 requested variable alignment.
2604 (expand_one_ssa_partition): Likewise.
2605 (expand_one_var): Likewise.
2607 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
2609 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
2610 to speculation_safe_value_not_needed.
2612 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
2614 * targhooks.h (speculation_safe_value_not_needed): New prototype.
2615 * targhooks.c (speculation_safe_value_not_needed): New function.
2616 * target.def (have_speculation_safe_value): Update documentation.
2617 * doc/tm.texi: Regenerated.
2619 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
2621 * config/aarch64/iterators.md (ALLI_TI): New iterator.
2622 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
2624 (despeculate_copy<ALLI:mode>_insn): New insn.
2625 (despeculate_copyti_insn): New insn.
2626 (despeculate_simple<ALLI:mode>): New insn
2627 (despeculate_simpleti): New insn.
2628 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
2630 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
2631 aarch64_speculation_safe_value.
2632 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
2634 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
2636 * config/aarch64/aarch64-speculation.cc: New file.
2637 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
2638 before pass_reorder_blocks.
2639 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
2641 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
2642 X14 and X15 when tracking speculation.
2643 * config/aarch64/aarch64.md (register name constants): Add
2644 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
2645 (unspec): Add UNSPEC_SPECULATION_TRACKER.
2646 (speculation_barrier): New insn attribute.
2647 (cmp<mode>): Allow SP in comparisons.
2648 (speculation_tracker): New insn.
2649 (speculation_barrier): Add speculation_barrier attribute.
2650 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
2651 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
2652 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
2654 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
2656 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
2657 aarch64_track_speculation is true.
2658 (tb<optab><mode>1): Likewise.
2659 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
2660 generate CB[N]Z when tracking speculation.
2661 (aarch64_split_compare_and_swap): Likewise.
2662 (aarch64_split_atomic_op): Likewise.
2664 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
2666 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
2668 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
2670 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
2671 (speculation_barrier): New insn.
2673 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
2675 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
2676 * config/arm/arm.md (speculation_barrier): New expand.
2677 (speculation_barrier_insn): New pattern.
2679 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
2681 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
2682 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
2683 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
2684 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
2686 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
2687 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
2688 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
2689 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
2690 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
2691 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
2692 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
2693 * builtins.c (expand_speculation_safe_value): New function.
2694 (expand_builtin): Call it.
2695 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
2696 * doc/extend.texi: Document __builtin_speculation_safe_value.
2697 * doc/md.texi: Document "speculation_barrier" pattern.
2698 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
2699 TARGET_HAVE_SPECULATION_SAFE_VALUE.
2700 * doc/tm.texi: Regenerated.
2701 * target.def (have_speculation_safe_value, speculation_safe_value): New
2703 * targhooks.c (default_have_speculation_safe_value): New function.
2704 (default_speculation_safe_value): New function.
2705 * targhooks.h (default_have_speculation_safe_value): Add prototype.
2706 (default_speculation_safe_value): Add prototype.
2708 2018-07-31 David Malcolm <dmalcolm@redhat.com>
2710 * dump-context.h (dump_context::dump_loc): New decl.
2711 * dumpfile.c (dump_context::dump_loc): New member function.
2712 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
2713 and dump_gimple_stmt.
2714 (dump_context::dump_gimple_expr_loc): Likewise, using
2716 (dump_context::dump_generic_expr_loc): Likewise, using
2718 (dump_context::dump_printf_loc_va): Likewise, using
2720 (dump_context::begin_scope): Explicitly using the global function
2721 "dump_loc", rather than the member function.
2723 2018-07-31 Martin Sebor <msebor@redhat.com>
2725 PR tree-optimization/86741
2726 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
2728 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
2730 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
2731 depend on whether prefetch instructions will be emitted or not.
2732 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
2733 will be emitted or not.
2734 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
2735 (TARGET_SETMEM_PFD): New macros.
2737 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2739 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
2740 (NULL_STMT_VEC_INFO): Delete.
2741 (stmt_vec_info::operator*): Likewise.
2742 (stmt_vec_info::operator gimple *): Likewise.
2743 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
2744 of NULL_STMT_VEC_INFO.
2745 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
2746 (vect_reassociating_reduction_p): Likewise.
2747 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
2748 (vectorizable_store): Likewise.
2749 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
2750 (vec_info::free_stmt_vec_infos): Likewise.
2752 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2754 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
2755 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
2757 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2759 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
2760 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
2761 (vec_info::free_stmt_vec_info): New private member functions.
2762 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
2763 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
2764 * tree-parloops.c (gather_scalar_reductions): Remove calls to
2765 set_stmt_vec_info_vec and free_stmt_vec_infos.
2766 * tree-vect-loop.c (_loop_vec_info): Remove call to
2767 set_stmt_vec_info_vec.
2768 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
2769 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
2770 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
2771 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
2772 (vec_info::free_stmt_vec_info): ...these new functions. Remove
2773 assignments in {vec_info::,}new_stmt_vec_info that are redundant
2774 with the clearing in the xcalloc.
2775 (stmt_vec_info_vec): Delete.
2776 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
2777 (vectorize_loops): Likewise.
2778 (vec_info::~vec_info): Remove argument from call to
2779 free_stmt_vec_infos.
2780 (vec_info::add_stmt): Remove vinfo argument from call to
2783 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2785 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
2786 rather than a gimple stmt.
2787 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
2788 information for pattern statements when passed the original
2789 statement; instead wait to be passed the pattern statement itself.
2790 Don't call set_vinfo_for_stmt here.
2791 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
2792 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
2793 stmt_vec_infos here.
2794 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
2795 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
2796 stmt_vec_infos entry.
2798 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2800 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
2801 * tree-vectorizer.c (vec_info::replace_stmt): New function.
2802 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
2803 * tree-vect-stmts.c (vectorizable_call): Likewise.
2804 (vectorizable_simd_clone_call): Likewise.
2806 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2808 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
2809 * tree-vectorizer.c (vec_info::remove_stmt): New function.
2810 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
2811 * tree-vect-loop.c (vect_transform_loop): Likewise.
2812 * tree-vect-slp.c (vect_schedule_slp): Likewise.
2813 * tree-vect-stmts.c (vect_remove_stores): Likewise.
2815 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2817 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
2818 (vect_dr_stmt): Delete.
2819 * tree-vectorizer.c (vec_info::lookup_dr): New function.
2820 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
2822 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
2823 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
2824 (vect_verify_datarefs_alignment, vect_peeling_supportable)
2825 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
2826 (vect_analyze_data_refs): Likewise.
2827 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
2829 (vect_find_same_alignment_drs): Likewise.
2830 (vect_slp_analyze_node_dependences): Update calls accordingly.
2831 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
2832 instead of DR_VECT_AUX.
2833 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
2834 of a vector data references. Use vec_info::lookup_dr instead of
2836 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
2837 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
2838 instead of DR_VECT_AUX.
2840 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2842 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
2844 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
2846 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
2847 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
2848 (vect_gen_prolog_loop_niters): Likewise.
2850 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2852 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
2853 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
2854 (vect_known_alignment_in_bytes, vect_dr_behavior)
2855 (vect_get_scalar_dr_size): Take references as dr_vec_infos
2856 instead of data_references. Update calls to other routines for
2857 which the same change has been made.
2858 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
2859 dr_vec_infos instead of stmt_vec_infos.
2860 (vect_analyze_data_ref_dependence): Update call accordingly.
2861 (vect_slp_analyze_data_ref_dependence)
2862 (vect_record_base_alignments): Use DR_VECT_AUX.
2863 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
2864 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
2865 (vector_alignment_reachable_p, vect_get_data_access_cost)
2866 (vect_peeling_supportable, vect_analyze_group_access_1)
2867 (vect_analyze_group_access, vect_analyze_data_ref_access)
2868 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
2869 (vect_compile_time_alias, vect_small_gap_p)
2870 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
2871 (vect_supportable_dr_alignment): Take references as dr_vec_infos
2872 instead of data_references. Update calls to other routines for
2873 which the same change has been made.
2874 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
2875 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
2876 (vect_slp_analyze_and_verify_node_alignment)
2877 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
2878 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
2879 (vect_setup_realignment): Use dr_vec_infos. Update calls after
2881 (_vect_peel_info::dr): Replace with...
2882 (_vect_peel_info::dr_info): ...this new field.
2883 (vect_peeling_hash_get_most_frequent)
2884 (vect_peeling_hash_choose_best_peeling): Update accordingly.
2885 (vect_peeling_hash_get_lowest_cost):
2886 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
2887 routines for which the same change has been made.
2888 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
2890 * tree-vect-loop-manip.c (get_misalign_in_elems)
2891 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
2893 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
2894 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
2895 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
2896 (get_group_load_store_type, get_negative_load_store_type)
2897 (vect_get_data_ptr_increment, vectorizable_store)
2898 (vectorizable_load): Likewise.
2899 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
2900 Update calls to other routines for which the same change has been made.
2902 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2904 * tree-vectorizer.h (vec_info::move_dr): New member function.
2905 (dataref_aux): Rename to...
2906 (dr_vec_info): ...this and add "dr" and "stmt" fields.
2907 (_stmt_vec_info::dr_aux): Update accordingly.
2908 (_stmt_vec_info::data_ref_info): Delete.
2909 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
2910 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
2911 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
2912 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
2914 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
2915 (STMT_VINFO_DR_INFO): New macro.
2916 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
2917 (set_dr_misalignment): Update after rename of dataref_aux.
2918 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
2919 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
2920 initialization of STMT_VINFO_DATA_REF.
2921 * tree-vectorizer.c (vec_info::move_dr): New function.
2922 * tree-vect-patterns.c (vect_recog_bool_pattern)
2923 (vect_recog_mask_conversion_pattern)
2924 (vect_recog_gather_scatter_pattern): Use it.
2925 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
2926 the "dr" and "stmt" fields of dr_vec_info instead of
2927 STMT_VINFO_DATA_REF.
2929 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2931 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
2932 (is_pattern_stmt_p): Use it.
2933 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
2934 on pattern statements.
2936 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2938 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
2939 original stmt as a stmt_vec_info rather than a gimple stmt.
2940 (vect_pattern_recog_1): Take the statement directly as a
2941 stmt_vec_info, rather than via a gimple_stmt_iterator.
2942 Update call to vect_mark_pattern_stmts.
2943 (vect_pattern_recog): Update calls accordingly.
2945 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2947 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
2948 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
2949 a vect_def_type for the first argument.
2950 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
2951 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
2952 operand if it isn't defined by a vectorized statement.
2953 (vect_build_gather_load_calls): Remove the mask_dt argument and
2954 update calls to vect_get_vec_def_for_stmt_copy.
2955 (vectorizable_bswap): Likewise the dt argument.
2956 (vectorizable_call): Update calls to vectorizable_bswap and
2957 vect_get_vec_def_for_stmt_copy.
2958 (vectorizable_simd_clone_call, vectorizable_assignment)
2959 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
2960 (vectorizable_comparison): Update calls to
2961 vect_get_vec_def_for_stmt_copy.
2962 (vectorizable_store): Likewise. Remove now-unnecessary calls to
2964 (vect_get_loop_based_defs): Remove dt argument and update call
2965 to vect_get_vec_def_for_stmt_copy.
2966 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
2967 and vect_get_vec_def_for_stmt_copy.
2968 (vectorizable_load): Update calls to vect_build_gather_load_calls
2969 and vect_get_vec_def_for_stmt_copy.
2970 * tree-vect-loop.c (vect_create_epilog_for_reduction)
2971 (vectorizable_reduction, vectorizable_live_operation): Update calls
2972 to vect_get_vec_def_for_stmt_copy.
2974 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2976 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
2977 and gimple stmt arguments with a stmt_vec_info.
2978 (vect_record_base_alignments): Update calls accordingly.
2979 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
2980 and gimple stmt arguments with a stmt_vec_info.
2981 (vect_build_slp_tree_1): Remove vinfo argument and update call
2982 to vect_record_max_nunits.
2983 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
2984 and vect_record_max_nunits.
2986 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
2988 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
2989 file and take a stmt_vec_info instead of a gimple stmt.
2990 (supportable_widening_operation, vect_finish_replace_stmt)
2991 (vect_finish_stmt_generation, vect_get_store_rhs)
2992 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
2993 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
2994 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
2995 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
2996 (vect_create_data_ref_ptr, bump_vector_ptr)
2997 (vect_permute_store_chain, vect_setup_realignment)
2998 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
2999 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
3000 (vectorizable_reduction, vectorizable_induction)
3001 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
3002 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
3003 than gimple stmts as arguments.
3004 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
3005 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
3006 (can_group_stmts_p, vect_check_gather_scatter)
3007 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
3008 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
3009 (vect_permute_load_chain, vect_shift_permute_load_chain)
3010 (vect_transform_grouped_load)
3011 (vect_record_grouped_load_vectors): Likewise.
3012 * tree-vect-loop.c (vect_fixup_reduc_chain)
3013 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
3014 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
3015 (vectorizable_reduction, vectorizable_induction)
3016 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
3017 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
3018 (vect_get_load_store_mask): Likewise.
3019 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
3020 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
3021 * tree-vect-stmts.c (vect_mark_relevant)
3022 (is_simple_and_all_uses_invariant)
3023 (exist_non_indexing_operands_for_use_p, process_use)
3024 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
3025 (vect_get_vec_def_for_operand, vect_get_vec_defs)
3026 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
3027 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
3028 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
3029 (get_negative_load_store_type, get_load_store_type)
3030 (vect_check_load_store_mask, vect_check_store_rhs)
3031 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
3032 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
3033 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
3034 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
3035 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
3036 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
3037 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
3038 (supportable_widening_operation): Likewise.
3040 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3042 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
3043 a stmt_vec_info instead of a gcall.
3044 (vect_check_gather_scatter): Update call accordingly.
3045 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
3047 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
3048 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
3049 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
3050 instead of a gimple stmt.
3051 (vect_transform_loop): Update calls accordingly.
3052 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
3053 stmt_vec_infos instead of gimple stmts.
3054 (vect_analyze_slp_instance): Update use accordingly.
3055 * tree-vect-stmts.c (read_vector_array, write_vector_array)
3056 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
3057 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
3058 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
3059 (vect_gen_widened_results_half, vect_get_loop_based_defs)
3060 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
3061 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
3062 down to subroutines.
3064 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3066 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
3067 of the worklist from a vector of gimple stmts to a vector of
3069 * tree-vect-stmts.c (vect_mark_relevant, process_use)
3070 (vect_mark_stmts_to_be_vectorized): Likewise
3072 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3074 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
3075 statement before passing it to vect_analyze_stmt.
3076 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
3077 the chain of phi vector definitions. Track the exit phi via its
3079 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
3080 STMT_VINFO_REDUC_DEF.
3081 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
3082 stmt_vec_infos to handle the statement chains.
3083 (vect_get_slp_defs): Record the first statement in the node
3084 using a stmt_vec_info.
3085 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
3086 statements here and pass their stmt_vec_info down to subroutines.
3087 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
3088 down to vect_finish_stmt_generation.
3089 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
3090 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
3091 stmt_vec_infos to subroutines.
3092 (vect_remove_stores): Use stmt_vec_infos to handle the statement
3095 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3097 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
3098 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
3099 (vect_permute_store_chain, vect_setup_realignment)
3100 (vect_permute_load_chain, vect_shift_permute_load_chain)
3101 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
3102 stmts internally, and when passing values to other vectorizer routines.
3103 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
3104 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
3105 (vect_analyze_loop_operations, get_initial_def_for_reduction)
3106 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
3107 (vectorizable_reduction, vectorizable_induction)
3108 (vectorizable_live_operation, vect_transform_loop_stmt)
3109 (vect_transform_loop): Likewise.
3110 * tree-vect-patterns.c (vect_reassociating_reduction_p)
3111 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
3112 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
3113 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
3114 (vect_slp_analyze_node_operations_1): Likewise.
3115 * tree-vect-stmts.c (vect_mark_relevant, process_use)
3116 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
3117 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
3118 (vect_finish_stmt_generation_1, get_group_load_store_type)
3119 (get_load_store_type, vect_build_gather_load_calls)
3120 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
3121 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
3122 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
3123 (vectorizable_store, vectorizable_load, vectorizable_condition)
3124 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
3125 (supportable_widening_operation): Likewise.
3126 (vect_get_vector_types_for_stmt): Likewise.
3127 * tree-vectorizer.h (vect_dr_behavior): Likewise.
3129 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3131 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
3132 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
3133 (vect_permute_store_chain, vect_permute_load_chain)
3134 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
3135 repeated stmt_vec_info lookups.
3136 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
3137 (vect_update_ivs_after_vectorizer): Likewise.
3138 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
3139 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
3140 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
3141 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
3142 (vect_bb_slp_scalar_cost): Likewise.
3143 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
3145 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3147 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
3148 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
3149 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
3150 to get gassigns and gcalls, rather than operating on generc gimple
3152 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
3153 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
3154 (vectorizable_load, vect_analyze_stmt): Likewise.
3155 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
3157 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3159 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
3160 return stmt_vec_infos rather than gimple stmts. Do not accept
3162 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
3164 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
3165 Update use of get_later_stmt.
3166 (vect_get_constant_vectors): Update call accordingly.
3167 (vect_schedule_slp_instance): Likewise
3168 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
3169 (vect_slp_analyze_instance_dependence): Likewise.
3170 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
3172 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3174 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
3175 (stmt_info_for_cost::stmt_info): ...this new field.
3176 (add_stmt_costs): Update accordingly.
3177 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
3178 (vect_get_known_peeling_cost): Likewise.
3179 (vect_estimate_min_profitable_iters): Likewise.
3180 * tree-vect-stmts.c (record_stmt_cost): Likewise.
3182 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3184 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
3185 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
3186 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
3188 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
3190 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3192 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
3193 a gimple stmt to a stmt_vec_info.
3194 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
3196 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3198 * tree-vectorizer.h (vec_info::grouped_stores): Change from
3199 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
3200 (_loop_vec_info::reduction_chains): Likewise.
3201 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
3203 * tree-vect-slp.c (vect_analyze_slp): Likewise.
3205 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3207 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
3208 a gimple stmt to a stmt_vec_info.
3209 (_stmt_vec_info::next_element): Likewise.
3210 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
3211 (vect_slp_analyze_and_verify_node_alignment)
3212 (vect_analyze_group_access_1, vect_analyze_group_access)
3213 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
3214 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
3215 (vect_supportable_dr_alignment): Update accordingly.
3216 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
3217 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
3218 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
3219 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
3220 * tree-vect-slp.c (vect_build_slp_tree_1)
3221 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
3222 (vect_split_slp_store_group, vect_analyze_slp_instance)
3223 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
3224 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
3225 (get_group_load_store_type, get_load_store_type)
3226 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
3227 (vect_transform_stmt, vect_remove_stores): Likewise.
3229 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3231 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
3233 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
3234 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
3235 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
3236 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
3237 (vector_alignment_reachable_p, vect_get_data_access_cost)
3238 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
3239 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
3240 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
3241 (vect_analyze_group_access_1, vect_analyze_group_access)
3242 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
3243 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
3244 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
3245 result of vect_dr_stmt and use the stmt_vec_info instead of
3246 the associated gimple stmt.
3247 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
3248 (vect_gen_prolog_loop_niters): Likewise.
3249 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
3251 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3253 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
3254 to a vec<stmt_vec_info>.
3255 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
3256 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
3258 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
3259 to a vec<stmt_vec_info>.
3260 (bst_traits::value_type, bst_traits::value_type): Likewise.
3261 (bst_traits::hash): Update accordingly.
3262 (vect_get_and_check_slp_defs): Change the stmts parameter from
3263 a vec<gimple *> to a vec<stmt_vec_info>.
3264 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
3265 (vect_build_slp_tree): Likewise.
3266 (vect_build_slp_tree_2): Likewise. Update uses of
3267 SLP_TREE_SCALAR_STMTS.
3268 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
3269 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
3270 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
3271 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
3272 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
3273 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
3274 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
3275 (vect_get_constant_vectors, vect_get_slp_defs)
3276 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
3277 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
3278 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
3279 instead of gimple stmts.
3280 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
3281 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
3282 (vect_slp_analyze_instance_dependence): Update uses of
3283 SLP_TREE_SCALAR_STMTS.
3284 (vect_slp_analyze_and_verify_node_alignment): Likewise.
3285 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
3286 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
3287 (get_initial_defs_for_reduction): Likewise.
3288 (vect_create_epilog_for_reduction): Likewise.
3289 (vectorize_fold_left_reduction): Likewise.
3290 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
3291 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
3292 (can_vectorize_live_stmts): Likewise.
3294 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3296 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
3297 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
3298 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
3300 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
3302 (gather_scalar_reductions): Update after above interface changes.
3303 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
3304 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
3306 (vect_force_simple_reduction): Likewise.
3307 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
3308 LOOP_VINFO_REDUCTIONS.
3309 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
3311 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3313 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
3314 a gimple stmt to a stmt_vec_info.
3315 * tree-vect-loop.c (vect_active_double_reduction_p)
3316 (vect_force_simple_reduction, vectorizable_reduction): Update
3319 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3321 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
3322 vec<gimple *> to a vec<stmt_vec_info>.
3323 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
3324 the reduction_phis argument from a vec<gimple *> to a
3326 (vectorizable_reduction): Likewise the phis local variable that
3327 is passed to vect_create_epilog_for_reduction. Update for new type
3328 of SLP_TREE_VEC_STMTS.
3329 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
3330 (vectorizable_live_operation): Likewise.
3331 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
3332 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
3334 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3336 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
3337 a gimple stmt to a stmt_vec_info.
3338 (vectorizable_condition, vectorizable_live_operation)
3339 (vectorizable_reduction, vectorizable_induction): Pass back the
3340 vectorized statement as a stmt_vec_info.
3341 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
3342 use of STMT_VINFO_VEC_STMT.
3343 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
3344 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
3345 as stmt_vec_infos rather than gimple stmts.
3346 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
3348 (vectorizable_live_operation): Likewise.
3349 (vectorizable_reduction, vectorizable_induction): Likewise,
3350 updating use of STMT_VINFO_VEC_STMT.
3351 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
3352 of STMT_VINFO_VEC_STMT.
3353 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
3354 (vectorizable_simd_clone_call, vectorizable_conversion)
3355 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
3356 (vectorizable_store, vectorizable_load, vectorizable_condition)
3357 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
3358 from a gimple stmt to a stmt_vec_info.
3359 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
3360 pointer to a stmt_vec_info to the vectorizable_* routines.
3362 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3364 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
3365 a gimple stmt to a stmt_vec_info.
3366 (is_pattern_stmt_p): Update accordingly.
3367 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
3368 (vect_record_grouped_load_vectors): Likewise.
3369 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
3370 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
3371 (vect_model_reduction_cost): Likewise.
3372 (vect_create_epilog_for_reduction): Likewise.
3373 (vectorizable_reduction, vectorizable_induction): Likewise.
3374 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
3375 Return the stmt_vec_info for the pattern statement.
3376 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
3377 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
3378 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
3379 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
3380 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
3381 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
3382 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
3383 (free_stmt_vec_info, vect_is_simple_use): Likewise.
3385 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3387 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
3388 (vect_finish_stmt_generation): Likewise.
3389 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
3390 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
3391 (vect_build_gather_load_calls): Use the return value of the above
3392 functions instead of a separate call to vinfo_for_stmt. Use narrow
3393 scopes for the input gimple stmt and wider scopes for the associated
3394 stmt_vec_info. Use vec_info::lookup_def when setting these
3395 stmt_vec_infos from an SSA_NAME definition.
3396 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
3397 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
3398 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
3399 (vectorizable_store, vectorizable_load, vectorizable_condition)
3400 (vectorizable_comparison): Likewise.
3401 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
3402 (vectorizable_reduction): Likewise.
3404 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3406 * tree-vectorizer.h (vect_is_simple_use): Add an optional
3407 stmt_vec_info * parameter before the optional gimple **.
3408 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
3409 (process_use, vect_get_vec_def_for_operand_1): Update callers.
3410 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
3411 * tree-vect-loop.c (vectorizable_reduction): Likewise.
3412 (vectorizable_live_operation): Likewise.
3413 * tree-vect-patterns.c (type_conversion_p): Likewise.
3414 (vect_look_through_possible_promotion): Likewise.
3415 (vect_recog_rotate_pattern): Likewise.
3416 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
3418 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3420 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
3421 a typedef to a wrapper class.
3422 (NULL_STMT_VEC_INFO): New macro.
3423 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
3424 (stmt_vec_info::operator*): New function.
3425 (stmt_vec_info::operator gimple *): Likewise.
3426 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
3427 (add_stmt_costs): Likewise.
3428 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
3429 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
3430 (vect_get_known_peeling_cost): Likewise.
3431 (vect_estimate_min_profitable_iters): Likewise.
3432 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
3433 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
3434 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
3435 (vectorizable_store, free_stmt_vec_infos): Likewise.
3436 (new_stmt_vec_info): Change return type of xcalloc to
3439 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3441 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
3442 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
3443 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
3444 a single_imm_use-based sequence.
3445 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
3447 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3449 * tree-vectorizer.h (vec_info::lookup_def): Declare.
3450 * tree-vectorizer.c (vec_info::lookup_def): New function.
3451 * tree-vect-patterns.c (vect_get_internal_def): Use it.
3452 (vect_widened_op_tree): Likewise.
3453 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
3454 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
3455 (vectorizable_reduction): Likewise.
3456 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
3458 (vect_is_slp_reduction): Update calls accordingly. Use
3459 vec_info::lookup_def.
3460 (vect_is_simple_reduction): Likewise
3461 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
3463 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3465 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
3466 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
3467 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
3469 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
3470 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
3471 (vect_update_vf_for_slp, vect_analyze_loop_operations)
3472 (vect_is_slp_reduction, vectorizable_induction)
3473 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
3474 * tree-vect-patterns.c (vect_init_pattern_stmt):
3475 (vect_determine_min_output_precision_1, vect_determine_precisions)
3476 (vect_pattern_recog): Likewise.
3477 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
3478 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
3479 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
3480 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
3481 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
3482 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
3483 info field from a loop to a loop_vec_info.
3485 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3487 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
3488 (vec_info::add_stmt): Declare.
3489 * tree-vectorizer.c (vec_info::add_stmt): New function.
3490 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
3491 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
3492 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
3493 (vectorizable_induction): Likewise.
3494 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
3495 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
3496 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
3497 (vectorizable_load): Likewise.
3498 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
3499 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
3500 (vect_recog_gather_scatter_pattern): Likewise.
3501 (append_pattern_def_seq): Likewise. Remove a check that is
3502 performed by add_stmt itself.
3504 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3506 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
3507 which make_ssa_name was called with new_stmt before new_stmt
3510 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3512 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
3514 (vect_is_slp_reduction): ...here...
3515 (vect_is_simple_reduction): ...and here. Remove repetition of tests
3516 that are already known to be false.
3518 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3520 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
3521 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
3522 STMT_VINFO_NUM_SLP_USES when it's true.
3523 (vect_free_slp_instance): Add a final_p parameter and pass it to
3525 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
3526 (vect_analyze_slp_instance): Likewise.
3527 (vect_slp_analyze_operations): Likewise.
3528 (vect_slp_analyze_bb_1): Likewise.
3529 * tree-vectorizer.c (vec_info): Likewise.
3530 * tree-vect-loop.c (vect_transform_loop): Likewise.
3532 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3534 * tree-vect-loop.c (vectorizable_reduction): Assert that the
3535 function is not called for second and subsequent members of
3538 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3540 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
3541 cases for nested loops from here to ...
3542 (vect_create_epilog_for_reduction): ...here. Only call
3543 vect_is_simple_use for inner-loop reductions.
3545 2018-07-31 Martin Liska <mliska@suse.cz>
3547 PR gcov-profile/85338
3548 PR gcov-profile/85350
3549 PR gcov-profile/85372
3550 * profile.c (struct location_triplet): New.
3551 (struct location_triplet_hash): Likewise.
3552 (output_location): Do not output a BB that
3553 is already recorded for a line.
3554 (branch_prob): Use streamed_locations.
3556 2018-07-31 Martin Liska <mliska@suse.cz>
3558 PR gcov-profile/85370
3559 * coverage.c (coverage_begin_function): Do not mark target
3560 clones as artificial functions.
3562 2018-07-31 Martin Liska <mliska@suse.cz>
3564 PR gcov-profile/83813
3565 PR gcov-profile/84758
3566 PR gcov-profile/85217
3567 PR gcov-profile/85332
3568 * profile.c (branch_prob): Do not record GOTO expressions
3569 for GIMPLE statements which locations are already streamed.
3571 2018-07-31 Olivier Hainque <hainque@adacore.com>
3573 * gcc.c (handle_spec_function): Accept a soft_matched_part
3574 argument, as do_spec_1. Pass it down to ...
3575 (eval_spec_function): Accept a soft_matched_part argument,
3576 and pass it down to ...
3577 (do_spec_2): Accept a soft_matched_part argument, and pass
3578 it down to do_spec_1.
3579 (do_spec_1): Pass soft_matched_part to handle_spec_function.
3580 (handle_braces): Update call to handle_spec_function.
3581 (driver::set_up_specs): Update calls to do_spec_2.
3582 (compare_debug_dump_opt_spec_function): Likewise.
3583 (compare_debug_self_opt_spec_function): Likewise.
3585 2018-07-31 Olivier Hainque <hainque@adacore.com>
3587 * common.opt (nolibc): New option.
3588 * doc/invoke.texi (Link Options): Document it.
3589 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
3590 * config/alpha/linux.h: Likewise.
3591 * config/arc/elf.h: Likewise.
3592 * config/arm/uclinux-elf.h: Likewise.
3593 * config/arm/unknown-elf.h: Likewise.
3594 * config/avr/avrlibc.h: Likewise.
3595 * config/bfin/bfin.h: Likewise.
3596 * config/bfin/linux.h: Likewise.
3597 * config/bfin/uclinux.h: Likewise.
3598 * config/darwin.h: Likewise.
3599 * config/darwin10.h: Likewise.
3600 * config/darwin12.h: Likewise.
3601 * config/gnu-user.h: Likewise.
3602 * config/lm32/uclinux-elf.h: Likewise.
3603 * config/pa/pa-hpux11.h: Likewise.
3604 * config/pa/pa64-hpux.h: Likewise.
3605 * config/sparc/sparc.h: Likewise.
3607 2018-07-31 Olivier Hainque <hainque@adacore.com>
3609 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
3610 undefined variables.
3612 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
3615 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
3618 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
3620 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
3621 terminated string literal.
3623 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
3625 PR rtl-optimization/85160
3626 * combine.c (is_just_move): New function.
3627 (try_combine): Allow combining two instructions into two if neither of
3628 the original instructions was a move.
3630 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
3633 * doc/extend.texi (Global Register Variables): Discourage use of type
3635 (Local Register Variables): Likewise.
3637 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
3639 PR tree-optimization/86506
3640 * hwint.h (ceil_log2): Resync with hwint.c implementation.
3642 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
3645 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
3646 hard_regno, make sure no insn between `from` and `to` clobbers it.
3648 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
3649 Tom de Vries <tdevries@suse.de>
3651 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
3652 (PTX_DEFAULT_RUNTIME_DIM): ... this.
3653 (nvptx_goacc_validate_dims): Set default worker and gang dims to
3654 PTX_DEFAULT_RUNTIME_DIM.
3655 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
3657 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
3659 * config/pa/pa.c (pa_output_addr_vec): Align address table.
3660 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
3661 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
3663 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
3665 * config/rs6000/constraints.md (wG constraint): Delete, no longer
3667 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
3668 predicate to reflect toc fusion has been deleted.
3669 (toc_fusion_mem_raw): Delete, no longer used.
3670 (toc_fusion_mem_wrapped): Likewise.
3671 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
3673 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
3674 Delete, no longer used.
3675 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
3676 meant to be used for toc fusion.
3677 (rs6000_debug_print_mode): Delete toc fusion debugging.
3678 (rs6000_debug_reg_global): Likewise.
3679 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
3680 fusion and secondary reload support that were never used.
3681 (rs6000_option_override_internal): Delete TOC fusion, that was only
3682 partially defined, and it did not work unless you also used the
3684 (rs6000_legitimate_address_p): Delete TOC fusion support.
3685 (rs6000_opt_masks): Likewise.
3686 (fusion_wrap_memory_address): Delete function, no longer used.
3687 (fusion_split_address); Delete TOC fusion support.
3688 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
3689 longer used with toc fusion being deleted.
3690 (TARGET_TOC_FUSION_FP): Likewise.
3691 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
3693 (toc fusion spliter): Delete TOC fusion support.
3694 (toc_fusionload_<mode>): Likewise.
3695 (toc_fusionload_di): Likewise.
3696 (fusion_gpr_load_<mode>): Delete generator function, this insn no
3697 longer needs to be named. Rename predicate to delete TOC fusion.
3698 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
3699 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
3700 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
3701 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
3702 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
3704 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
3706 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
3707 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
3708 __int128 in built-in function prototypes.
3709 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
3710 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
3712 2018-07-27 Martin Sebor <msebor@redhat.com>
3714 PR tree-optimization/86696
3715 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
3716 types, including enums.
3717 (handle_char_store): Be prepared for the above function to fail.
3719 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
3721 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
3722 when optimization level is lower than 2 or optimize for size.
3724 2018-07-26 Martin Sebor <msebor@redhat.com>
3726 PR tree-optimization/86043
3727 PR tree-optimization/86042
3728 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
3729 (get_string_cst_length): Rename...
3730 (get_min_string_length): ...to this. Add argument.
3731 (handle_char_store): Extend to handle multi-character stores by
3733 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
3734 * tree.h (initializer_zerop): Add argument.
3736 2018-07-26 Jakub Jelinek <jakub@redhat.com>
3739 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
3740 declare target to variables if they have always,{to,from,tofrom} map
3743 2018-07-26 Martin Liska <mliska@suse.cz>
3746 * lto-wrapper.c: Add linker_output as prefix
3747 for ltrans_output_file.
3749 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
3751 PR rtl-optimization/85805
3752 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
3753 value for hard registers if that was written in the same mode.
3755 2018-07-26 Martin Liska <mliska@suse.cz>
3757 PR gcov-profile/86536
3758 * gcov.c (format_gcov): Use printf format %.*f directly
3759 and do not handle special values.
3761 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
3763 * common/config/arc/arc-common.c (arc_option_optimization_table):
3764 Update default optimizations for size.
3766 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
3768 * config/arc/arc.md (movsf_insn): Add short instruction selection.
3769 * config/arc/constraints.md (CfZ): New constraint.
3770 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
3771 (subsf3_fpu): Likewise.
3772 (cmpsf_fpu): Likewise.
3773 (cmpsf_fpu_uneq): Likewise.
3775 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
3777 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
3779 (arc_is_uncached_mem_p): uncached applies to both the variable and
3782 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
3784 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
3787 2018-07-25 David Malcolm <dmalcolm@redhat.com>
3789 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
3790 field "m_scopes" from vec to auto_vec.
3792 2018-07-25 Martin Liska <mliska@suse.cz>
3794 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
3797 2018-07-25 Richard Biener <rguenther@suse.de>
3800 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
3801 special wrt context_die late.
3802 (gen_subprogram_die): Re-use DIEs in local scope.
3804 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
3806 PR tree-optimization/86644
3807 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
3809 2018-07-25 Martin Liska <mliska@suse.cz>
3812 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
3813 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
3815 2018-07-25 Martin Liska <mliska@suse.cz>
3818 * params.def: Explain ASan abbreviation and provide
3819 a documentation link.
3821 2018-07-24 Martin Sebor <msebor@redhat.com>
3823 PR tree-optimization/86622
3824 PR tree-optimization/86532
3825 * builtins.h (string_length): Declare.
3826 * builtins.c (c_strlen): Correct handling of non-constant offsets.
3827 (check_access): Be prepared for non-constant length ranges.
3828 (string_length): Make extern.
3829 * expr.c (string_constant): Only handle the minor non-constant
3830 array index. Use string_constant to compute the length of
3831 a generic string constant.
3833 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
3835 PR tree-optimization/86618
3836 * tree-vect-stmts.c (vectorizable_call): Don't take the address
3837 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
3839 2018-07-24 David Malcolm <dmalcolm@redhat.com>
3841 PR tree-optimization/86636
3842 * json.cc (json::object::set): Fix comment. Add assertions.
3843 (json::array::append): Move here from json.h. Add comment and an
3845 (json::string::string): Likewise.
3846 * json.h (json::array::append): Move to json.cc.
3847 (json::string::string): Likewise.
3848 * optinfo-emit-json.cc
3849 (optrecord_json_writer::impl_location_to_json): Assert that we
3850 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
3851 wrapper around it. Expand the location once, rather than three
3853 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
3854 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
3856 (optrecord_json_writer::optinfo_to_json): Likewise, in four
3857 places. Fix some overlong lines.
3859 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
3861 * config/aarch64/aarch64-simd.md
3862 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
3863 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
3864 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
3865 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
3866 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
3867 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
3868 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
3869 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
3870 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
3872 2018-07-24 Jakub Jelinek <jakub@redhat.com>
3875 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
3876 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
3877 and abs_d == d, do the power of two handling if profitable.
3879 2018-07-24 Richard Biener <rguenther@suse.de>
3881 * match.pd: Add BIT_FIELD_REF canonicalizations.
3883 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
3886 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
3888 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
3890 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
3891 terminated STRING_CST object.
3893 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
3895 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
3897 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
3899 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
3900 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
3901 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
3902 the elements into a register.
3903 (rs6000_split_v4si_init_di_reg): Delete.
3904 (rs6000_split_v4si_init): Delete.
3905 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
3906 (vsx_init_v4si): Rewrite as a define_expand.
3908 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
3910 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
3911 zero_extend argument from memory): New.
3913 2018-07-22 Martin Sebor <msebor@redhat.com>
3916 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
3917 diagnosing calls with unknown arguments unless -Walloca-larger-than
3918 is restricted to less than PTRDIFF_MAX bytes.
3920 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
3922 * doc/gcov.texi (Invoking Gcov): Editorial changes.
3924 2018-07-20 David Malcolm <dmalcolm@redhat.com>
3926 * pretty-print.c (text_info::set_location): Remove redundant
3927 "line_table" parameter from call to rich_location::set_range.
3929 2018-07-20 Martin Sebor <msebor@redhat.com>
3932 * builtins.c (expand_builtin_alloca): Adjust.
3933 * calls.c (alloc_max_size): Simplify.
3934 * cgraphunit.c (cgraph_node::expand): Adjust.
3935 * common.opt (larger_than_size, warn_frame_larger_than): Remove
3937 (frame_larger_than_size): Same.
3938 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
3939 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
3941 * doc/invoke.texi (GCC Command Options): Document option arguments.
3942 Explain byte-size arguments and suffixes.
3943 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
3944 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
3945 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
3946 * doc/options.texi (UInteger): Expand.
3947 (Host_Wide_Int, ByteSize): Document new properties.
3948 * final.c (final_start_function_1): Include sizes in an error message.
3949 * function.c (frame_offset_overflow): Same.
3950 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
3951 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
3952 Diagnose unbounded alloca calls only for limits of less than
3954 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
3955 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
3957 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
3958 only for limits of less than PTRDIFF_MAX.
3959 * langhooks-def.h (lhd_handle_option): Change function argument
3961 * langhooks.c (lhd_handle_option): Same.
3962 * langhooks.h (handle_option): Same.
3963 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
3965 (var_type, var_type_struct): Same.
3966 (var_set): Handle ByteSize flag.
3967 * optc-gen.awk: Add comments to output to ease debugging. Make
3968 use of HOST_WIDE_INT where appropriate.
3969 * opts-gen-save.awk: Use %lx to format unsigned long.
3970 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
3971 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
3972 arguments. Parse bytes-size suffixes.
3973 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
3974 (enum_value_to_arg): Same.
3975 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
3976 (handle_option): Adjust.
3977 (generate_option): Change function argument to HOST_WIDE_INT.
3978 (cmdline_handle_error): Adjust.
3979 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
3980 (set_option): Change function argument to HOST_WIDE_INT.
3981 (option_enabled): Handle cl_host_wide_int.
3982 (get_option_state): Handle CLVC_SIZE.
3983 (control_warning_option): Same.
3984 * opts.c (common_handle_option): Change function argument to
3985 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
3986 OPT_Wvla_larger_than_.
3987 * opts.h (enum cl_var_type): Add an enumerator.
3988 * stor-layout.c (layout_decl): Print a more meaningful warning.
3989 * toplev.c (output_stack_usage): Adjust.
3991 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
3993 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
3994 call to inline_expand_builtin_string_cmp.
3995 (expand_builtin_strcmp): Likewise.
3996 (expand_builtin_strncmp): Likewise.
3997 (inline_string_cmp): Delete the last parameter, change char_type_node
3998 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
4000 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
4001 the inlining expansion on target where the type of the call has same or
4002 narrower precision than unsigned char.
4004 2018-07-20 David Malcolm <dmalcolm@redhat.com>
4006 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
4007 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
4008 * common.opt (fsave-optimization-record): New option.
4009 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
4010 * doc/invoke.texi (-fsave-optimization-record): New option.
4011 * dumpfile.c: Include "optinfo-emit-json.h".
4012 (struct kv_pair): Move to coretypes.h.
4013 (optgroup_options): Make non-static.
4014 (dump_context::end_scope): Call
4015 optimization_records_maybe_pop_dump_scope.
4016 * dumpfile.h (optgroup_options): New decl.
4017 * json.cc: New file.
4019 * optinfo-emit-json.cc: New file.
4020 * optinfo-emit-json.h: New file.
4021 * optinfo.cc: Include "optinfo-emit-json.h".
4022 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
4023 (optinfo_enabled_p): Check optimization_records_enabled_p.
4024 (optinfo_wants_inlining_info_p): Likewise.
4025 * optinfo.h: Update comment.
4026 * profile-count.c (profile_quality_as_string): New function.
4027 * profile-count.h (profile_quality_as_string): New decl.
4028 (profile_count::quality): New accessor.
4029 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
4030 and optinfo_emit_json_cc_tests.
4031 * selftest.h (selftest::json_cc_tests): New decl.
4032 (selftest::optinfo_emit_json_cc_tests): New decl.
4033 * toplev.c: Include "optinfo-emit-json.h".
4034 (compile_file): Call optimization_records_finish.
4035 (do_compile): Call optimization_records_start.
4036 * tree-ssa-live.c: Include optinfo.h.
4037 (remove_unused_scope_block_p): Retain inlining information if
4038 optinfo_wants_inlining_info_p returns true.
4040 2018-07-20 Richard Biener <rguenther@suse.de>
4043 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
4044 to cover -flto-partition=none.
4046 2018-07-20 Martin Liska <mliska@suse.cz>
4048 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
4049 (get_decl_source_range): Remove unused function.
4051 2018-07-20 Richard Biener <rguenther@suse.de>
4053 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
4054 (struct vn_phi_s): Likewise.
4055 (struct vn_reference_s): Likewise.
4056 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
4057 for searching the slot of an entry known to be in the hash itself.
4058 (vn_phi_hasher::equal): Likewise.
4059 (vn_reference_hasher::equal): Likewise.
4060 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
4062 (optimistic_info, current_info): Remove, keeping only valid_info.
4063 (vn_reference_lookup_1): Remove fallback lookup.
4064 (vn_reference_lookup_2): Likewise.
4065 (vn_nary_op_lookup_1): Likewise.
4066 (vn_phi_lookup): Likewise.
4067 (vn_nary_build_or_lookup_1): Make sure to not chain the built
4069 (vn_reference_insert): Adjust, chain the inserted hash element
4070 at last_inserted_ref.
4071 (vn_reference_insert_pieces): Likewise.
4072 (visit_reference_op_call): Likewise.
4073 (vn_nary_op_insert_into): Chain the inserted hash element at
4075 (vn_nary_op_insert_pieces): Adjust.
4076 (vn_nary_op_insert): Likewise.
4077 (vn_nary_op_insert_stmt): Likewise.
4078 (vn_phi_insert): Adjust, chain the inserted hash element at
4080 (process_scc): Remove clearing and copying the optimistic
4081 table. Instead remove elements inserted during an optimistic
4082 iteration from the single table we maintain.
4083 (init_scc_vn): Adjust.
4084 (free_scc_vn): Likewise.
4085 (sccvn_dom_walker::record_cond): Likewise.
4086 (sccvn_dom_walker::after_dom_children): Likewise.
4088 2018-07-19 Martin Sebor <msebor@redhat.com>
4090 PR tree-optimization/84047
4091 PR tree-optimization/83776
4092 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
4093 (check_array_bounds): Call it.
4095 2018-07-19 Martin Sebor <msebor@redhat.com>
4097 * align.h (align_flags): Use member initialization.
4099 2018-07-19 David Malcolm <dmalcolm@redhat.com>
4101 * Makefile.in (OBJS): Add optinfo.o.
4102 * coretypes.h (class symtab_node): New forward decl.
4103 (struct cgraph_node): New forward decl.
4104 (class varpool_node): New forward decl.
4105 * dump-context.h: New file.
4106 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
4108 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
4109 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
4110 (set_alt_dump_file): Likewise.
4111 (dump_context::~dump_context): New dtor.
4112 (dump_gimple_stmt): Move implementation to...
4113 (dump_context::dump_gimple_stmt): ...this new member function.
4114 Add the stmt to any pending optinfo, creating one if need be.
4115 (dump_gimple_stmt_loc): Move implementation to...
4116 (dump_context::dump_gimple_stmt_loc): ...this new member function.
4117 Start a new optinfo and add the stmt to it.
4118 (dump_gimple_expr): Move implementation to...
4119 (dump_context::dump_gimple_expr): ...this new member function.
4120 Add the stmt to any pending optinfo, creating one if need be.
4121 (dump_gimple_expr_loc): Move implementation to...
4122 (dump_context::dump_gimple_expr_loc): ...this new member function.
4123 Start a new optinfo and add the stmt to it.
4124 (dump_generic_expr): Move implementation to...
4125 (dump_context::dump_generic_expr): ...this new member function.
4126 Add the tree to any pending optinfo, creating one if need be.
4127 (dump_generic_expr_loc): Move implementation to...
4128 (dump_context::dump_generic_expr_loc): ...this new member
4129 function. Add the tree to any pending optinfo, creating one if
4131 (dump_printf): Move implementation to...
4132 (dump_context::dump_printf_va): ...this new member function. Add
4133 the text to any pending optinfo, creating one if need be.
4134 (dump_printf_loc): Move implementation to...
4135 (dump_context::dump_printf_loc_va): ...this new member function.
4136 Start a new optinfo and add the stmt to it.
4137 (dump_dec): Move implementation to...
4138 (dump_context::dump_dec): ...this new member function. Add the
4139 value to any pending optinfo, creating one if need be.
4140 (dump_context::dump_symtab_node): New member function.
4141 (dump_context::get_scope_depth): New member function.
4142 (dump_context::begin_scope): New member function.
4143 (dump_context::end_scope): New member function.
4144 (dump_context::ensure_pending_optinfo): New member function.
4145 (dump_context::begin_next_optinfo): New member function.
4146 (dump_context::end_any_optinfo): New member function.
4147 (dump_context::s_current): New global.
4148 (dump_context::s_default): New global.
4149 (dump_scope_depth): Delete global.
4150 (dumpfile_ensure_any_optinfo_are_flushed): New function.
4151 (dump_symtab_node): New function.
4152 (get_dump_scope_depth): Reimplement in terms of dump_context.
4153 (dump_begin_scope): Likewise.
4154 (dump_end_scope): Likewise.
4155 (selftest::temp_dump_context::temp_dump_context): New ctor.
4156 (selftest::temp_dump_context::~temp_dump_context): New dtor.
4157 (selftest::verify_item): New function.
4158 (ASSERT_IS_TEXT): New macro.
4159 (ASSERT_IS_TREE): New macro.
4160 (ASSERT_IS_GIMPLE): New macro.
4161 (selftest::test_capture_of_dump_calls): New test.
4162 (selftest::dumpfile_c_tests): Call it.
4163 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
4164 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
4165 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
4166 descriptive comment.
4167 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
4168 (dump_node, dump_bb): Move these unrelated decls.
4169 (class dump_manager): Add leading comment.
4170 * optinfo.cc: New file.
4171 * optinfo.h: New file.
4173 2018-07-19 Michael Collison <michael.collison@arm.com>
4174 Richard Henderson <rth@redhat.com>
4176 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
4177 (subti): Handle op1 zero.
4178 (subvti4, usub4ti4): New.
4179 (*sub<GPI>3_compare1_imm): New.
4180 (sub<GPI>3_carryinCV): New.
4181 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
4182 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
4184 2018-07-19 Michael Collison <michael.collison@arm.com>
4185 Richard Henderson <rth@redhat.com>
4187 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
4188 (addti3): Create simpler code if low part is already known to be 0.
4189 (addvti4, uaddvti4): New.
4190 (*add<GPI>3_compareC_cconly_imm): New.
4191 (*add<GPI>3_compareC_cconly): New.
4192 (*add<GPI>3_compareC_imm): New.
4193 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
4194 handle constants within this pattern..
4195 (*add<GPI>3_compareV_cconly_imm): New.
4196 (*add<GPI>3_compareV_cconly): New.
4197 (*add<GPI>3_compareV_imm): New.
4198 (add<GPI>3_compareV): New.
4199 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
4200 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
4201 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
4202 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
4203 with 'comparison' operator.
4204 (*add<GPI>3_compareV_cconly_imm): Ditto.
4205 (*add<GPI>3_compareV_cconly): Ditto.
4206 (*add<GPI>3_compareV_imm): Ditto.
4207 (add<GPI>3_compareV): Ditto.
4208 (add<mode>3_carryinC): Ditto.
4209 (*add<mode>3_carryinC_zero): Ditto.
4210 (*add<mode>3_carryinC): Ditto.
4211 (add<mode>3_carryinV): Ditto.
4212 (*add<mode>3_carryinV_zero): Ditto.
4213 (*add<mode>3_carryinV): Ditto.
4215 2018-07-19 Michael Collison <michael.collison@arm.com>
4216 Richard Henderson <rth@redhat.com>
4218 * config/aarch64/aarch64-modes.def (CC_V): New.
4219 * config/aarch64/aarch64-protos.h
4220 (aarch64_addti_scratch_regs): Declare
4221 (aarch64_subvti_scratch_regs): Declare.
4222 (aarch64_expand_subvti): Declare.
4223 (aarch64_gen_unlikely_cbranch): Declare
4224 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
4225 for signed overflow using CC_Vmode.
4226 (aarch64_get_condition_code_1): Handle CC_Vmode.
4227 (aarch64_gen_unlikely_cbranch): New function.
4228 (aarch64_addti_scratch_regs): New function.
4229 (aarch64_subvti_scratch_regs): New function.
4230 (aarch64_expand_subvti): New function.
4232 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
4234 * config/aarch64/aarch64-option-extensions.def: New entry for profile
4236 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
4237 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
4240 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
4243 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
4244 address check not strict.
4246 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
4248 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
4250 (store_pair_lanes<mode>): Likewise.
4251 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
4252 enum value 'ADDR_QUERY_LDP_STP_N'.
4253 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
4254 (aarch64_print_address_internal): Add declaration.
4255 (aarch64_print_ldpstp_address): Remove.
4256 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
4257 (aarch64_print_operand): Change printing of 'y'.
4258 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
4259 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
4260 'true' rather than '1'.
4261 * config/aarch64/constraints.md (Uml): Likewise.
4262 (Uml): Rename to Umn.
4265 2018-07-19 Richard Biener <rguenther@suse.de>
4267 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
4269 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
4270 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
4271 (vn_reference_hasher): Likewise.
4272 (struct vn_tables_s): Remove obstack and alloc-pool members.
4273 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
4274 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
4275 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
4276 (vn_reference_insert_pieces): Likewise.
4277 (alloc_vn_nary_op_noinit): Adjust.
4278 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
4279 (vn_phi_eq): Adjust.
4280 (shared_lookup_phiargs): Remove.
4281 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
4282 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
4283 (visit_reference_op_call): Likewise.
4284 (copy_nary, copy_phi, copy_reference): Remove.
4285 (process_scc): Rewind the obstack when iterating. Do not
4286 copy the elements to valid_info but just move them from one
4287 hashtable to the other.
4288 (allocate_vn_table): Adjust.
4289 (free_vn_table): Likewise.
4290 (init_scc_vn): Likewise.
4291 (free_scc_vn): Likewise.
4293 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
4296 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
4297 indirect_return as function type attribute.
4298 (ix86_attribute_table): Change indirect_return to function
4300 * doc/extend.texi: Update indirect_return attribute.
4302 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
4304 * wide-int.h (widest2_int): New.
4305 * gimple-fold.c (arith_overflowed_p): Use it.
4306 * tree.h (widest2_int_cst): New.
4307 * tree-vrp.c (wide_int_binop_overflow): Rename from
4308 vrp_int_const_binop.
4309 Rewrite to work on trees.
4310 (extract_range_from_multiplicative_op_1): Abstract code to...
4311 (wide_int_range_min_max): ...here.
4312 (wide_int_range_cross_product): ...and here.
4313 (extract_range_from_binary_expr_1): Abstract overflow code to...
4314 (wide_int_range_mult_wrapping): ...here.
4315 * tree-vrp.h (wide_int_range_cross_product): New.
4316 (wide_int_range_mult_wrapping): New.
4318 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
4319 Julia Koval <julia.koval@intel.com>
4321 * config/i386/x86-tune-costs.h (skylake_memcpy,
4322 skylake_memset): Replace rep_prefix with unrolling for size 512.
4324 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
4327 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
4328 comparision with EQ_EXPR in last stmt.
4330 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
4332 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
4333 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
4334 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
4335 previously known as "PowerPC AltiVec Built-in Functions". Move
4336 some material to new subsubsections "PowerPC AltiVec Built-in
4337 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
4339 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
4340 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
4341 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
4342 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
4344 2018-07-18 Richard Biener <rguenther@suse.de>
4346 PR tree-optimization/86557
4347 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
4350 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
4352 * config/s390/s390.c (s390_function_profiler): Generate CFI.
4354 2018-07-17 Jeff Law <law@redhat.com>
4356 * config/arm/arm.c (get_label_padding): Update for recent
4357 changes to label_to_alignment.
4359 PR tree-optimization/86010
4360 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
4362 * config/mips/mips.c (vr4130_align_insns): Update for recent
4363 changes to label_to_alignment.
4365 * config/frv/frv.c (frv_label_align): Update for recent changes
4366 to label_to_alignment.
4368 * config/nios2/nios2.c (nios2_label_align): Update for recent
4369 changes which dropped ALIGN_LABELS_LOG.
4371 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
4373 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
4374 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
4376 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
4378 * config/arc/arc.c (arc_label_align): Use align_labels instead of
4379 deprecated align_labels_log.
4381 2018-07-17 Richard Biener <rguenther@suse.de>
4384 * dwarf2out.c (init_sections_and_labels): Always generate
4385 a debug_line_str_section for early LTO debug.
4386 (dwarf2out_finish): Reset debug_line_str_hash output early.
4387 Bump counter for extra dwarf5 .debug_loc labels to not conflict
4389 (dwarf2out_early_finish): Output debug_line_str.
4391 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
4393 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
4394 index register on z196 or later.
4396 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
4398 * config/s390/s390.c (s390_default_align): Set default function
4400 (s390_override_options_after_change): Call s390_default align.
4401 (s390_option_override_internal): Call s390_default_align.
4402 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
4404 2018-07-17 Jakub Jelinek <jakub@redhat.com>
4407 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
4408 to _looptemp_ clauses, other than the first two.
4410 2018-07-17 Martin Liska <mliska@suse.cz>
4412 * opts.c: Do not enable OPT_falign_* for -Os.
4414 2018-07-17 Martin Liska <mliska@suse.cz>
4416 * align.h (MAX_CODE_ALIGN): New.
4417 (MAX_CODE_ALIGN_VALUE): New.
4418 * common/config/i386/i386-common.c (ix86_handle_option):
4419 (MAX_CODE_ALIGN): Moved to align.h.
4420 * final.c (MAX_CODE_ALIGN): Likewise.
4421 * opts.c (parse_and_check_align_values):
4422 (MAX_CODE_ALIGN): Likewise.
4423 (MAX_CODE_ALIGN_VALUE): Likewise.
4425 2018-07-17 Martin Liska <mliska@suse.cz>
4427 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
4428 in order to fulfil coding style.
4429 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
4430 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
4431 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
4432 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
4433 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
4434 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
4435 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
4436 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
4437 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
4439 2018-07-17 Martin Liska <mliska@suse.cz>
4441 * align.h: New file.
4442 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
4444 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
4445 align_flags of label_to_alignment.
4446 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
4448 * config/m68k/m68k.c: Do not use removed align_labels_value and
4450 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
4451 (LOOP_ALIGN): Likewise.
4452 (LABEL_ALIGN): Likewise.
4453 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
4454 Remove not used macro.
4455 (rs6000_loop_align): Change return type to align_flags.
4456 (rs6000_loop_align_max_skip): Remove.
4457 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
4458 Change return type to align_flags.
4459 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
4460 Remove not used macro.
4461 (rs6000_loop_align): Change return type to align_flags.
4462 (rs6000_loop_align_max_skip): Remove.
4463 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
4464 * config/rx/rx-protos.h (rx_align_for_label): Make it
4466 * config/rx/rx.c (rx_align_for_label): Change return type
4468 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
4470 into align_flags class.
4471 (LABEL_ALIGN): Likewise.
4472 (LOOP_ALIGN): Likewise.
4473 * config/s390/s390.c (s390_label_align): Use align_flags
4475 (s390_asm_output_function_label): Likewise.
4476 * config/sh/sh.c (sh_override_options_after_change):
4477 Use align_flags class directly without macros.
4478 (find_barrier): Likewise.
4479 (barrier_align): Likewise.
4480 (sh_loop_align): Likewise.
4481 * config/spu/spu.c (spu_option_override):
4482 Use align_flags_tuple::get_value instead of removed macros.
4483 (spu_sched_init): Likewise.
4484 * config/spu/spu.h (GTY): Likewise.
4485 * config/visium/visium.c (visium_option_override):
4486 Set "8" as default secondary alignment.
4487 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
4488 in order to guarantee secondary alignment of 8.
4489 * coretypes.h: Include align.h header file.
4490 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
4491 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
4492 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
4493 * doc/tm.texi.in: Likewise.
4494 * final.c (struct label_alignment): Remove not used structure.
4495 (LABEL_ALIGN): Change type to align_flags.
4496 (LOOP_ALIGN): Likewise.
4497 (JUMP_ALIGN): Likewise.
4498 (default_loop_align_max_skip): Remove.
4499 (default_label_align_max_skip): Likewise.
4500 (default_jump_align_max_skip): Likewise.
4501 (default_label_align_after_barrier_max_skip):
4502 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
4503 (LABEL_TO_MAX_SKIP): Remove.
4504 (label_to_alignment): Return align_flags type instead of integer.
4505 (label_to_max_skip): Remove.
4506 (align_fuzz): Use align_flags type.
4507 (compute_alignments): Use align_flags type and use align_flags::max
4508 to combine multiple alignments.
4509 (grow_label_align): Grow vec instead of C array.
4510 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
4511 (shorten_branches): Use align_flags type and use align_flags::max
4512 to combine multiple alignments.
4513 (final_scan_insn_1): Remove usage of secondary alignment that comes
4514 from label alignment, but instead use proper secondary alignment
4515 which is computed in grow_label_align.
4516 * flags.h (struct align_flags_tuple): Move to align.h.
4517 (struct align_flags): Likewise.
4518 (state_align_loops): Rename to align_loops.
4519 (state_align_jumps): Rename to align_jumps.
4520 (state_align_labels): Rename to align_labels.
4521 (state_align_functions): Rename to align_functions.
4522 (align_loops_log): Remove.
4523 (align_jumps_log): Remove.
4524 (align_labels_log): Remove.
4525 (align_functions_log): Remove.
4526 (align_loops_max_skip): Remove.
4527 (align_jumps_max_skip): Remove.
4528 (align_labels_max_skip): Remove.
4529 (align_functions_max_skip): Remove.
4530 (align_loops_value): Remove.
4531 (align_jumps_value): Remove.
4532 (align_labels_value): Remove.
4533 (align_functions_value): Remove.
4534 * output.h (label_to_alignment): Change return type to align_flags.
4535 (label_to_max_skip): Remove.
4536 * target.def: Remove loop_align_max_skip, label_align_max_skip,
4537 jump_align_max_skip macros.
4538 * targhooks.h (default_loop_align_max_skip): Remove.
4539 (default_label_align_max_skip): Likewise.
4540 (default_jump_align_max_skip): Likewise.
4541 (default_label_align_after_barrier_max_skip): Remove.
4542 * toplev.c (read_log_maxskip): Use ::normalize function.
4543 (parse_N_M): Remove not used argument and also call ::normalize.
4544 (parse_alignment_opts): Do not pass unused arguments.
4545 * varasm.c (assemble_start_function): Use directly align_functions
4546 instead of removed macros.
4547 * system.h: Do not poison removed macros.
4549 2018-07-17 Jakub Jelinek <jakub@redhat.com>
4552 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
4553 and cond temporaries don't have reference type if iterator has
4554 pointer type. For init use &for_pre_body instead of pre_p if
4555 for_pre_body is non-empty.
4557 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
4559 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
4560 double-double modes to SFmode directly directly.
4561 (trunc<mode>sf2_fprs): Delete.
4563 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
4565 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
4566 for conversions between IFmode and the decimal floating point modes.
4567 (init_float128_ieee): Use the correct names for conversions between
4568 KFmode and the decimal floating point modes.
4570 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
4572 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
4573 for the conversions between TDmode and IFmode.
4574 (init_float128_ieee): Use more correct names for the conversions
4575 between TDmode and KFmode.
4577 2018-07-16 Jakub Jelinek <jakub@redhat.com>
4579 PR tree-optimization/86526
4580 * builtins.c (expand_builtin_memcmp): Formatting fixes.
4581 (inline_expand_builtin_string_cmp): Likewise.
4582 (inline_string_cmp): Likewise. Use c_readstr instead of
4583 builtin_memcpy_read_str. Add unit_mode temporary.
4585 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
4588 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
4589 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
4591 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
4593 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
4594 Alphabetize prototypes of built-in functions, separating out
4595 built-in functions that are listed in this section but should be
4596 described elsewhere.
4598 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
4601 * expmed.c (emit_store_flag): Do not emit setcc followed by a
4602 conditional move when trapping comparison was split to a
4603 non-trapping one (and vice versa).
4605 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
4607 * config/s390/s390.c (s390_function_profiler): Generate nops
4608 instead of profiler call sequences.
4609 * config/s390/s390.opt: Add the new option.
4611 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
4613 * config/s390/s390.c (s390_function_profiler): Generate
4614 __mcount_loc section.
4615 * config/s390/s390.opt: Add the new option.
4617 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
4619 * common.opt: Add the new warning.
4620 * config/s390/s390.c (s390_function_profiler): Emit "brasl
4621 %r0,__fentry__" when -mfentry is specified.
4622 (s390_option_override_internal): Disallow -mfentry for 31-bit
4624 * config/s390/s390.opt: Add the new option.
4626 2018-07-16 Richard Biener <rguenther@suse.de>
4629 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
4630 for function-local FUNCTION_DECL and RESULT_DECL immediately.
4632 2018-07-16 Martin Liska <mliska@suse.cz>
4635 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
4638 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
4640 * config/arc/arcHS.md: Update ARCHS scheduling rules.
4642 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
4644 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
4646 * config/arc/arc-cpus.def (hs4x): New cpu.
4648 * config/arc/arc-tables.opt: Regenerate.
4649 * config/arc/arc.c (arc_sched_issue_rate): New function.
4650 (TARGET_SCHED_ISSUE_RATE): Define.
4651 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
4652 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
4654 (attr tune): Add ARCHS4x tune values.
4655 (attr tune_dspmpy): Define.
4656 (*tst): Correct instruction type.
4657 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
4658 * config/arc/arcHS4x.md: New file.
4659 * config/arc/fpu.md: Update instruction type attributes.
4660 * config/arc/t-multilib: Regenerate.
4662 2018-07-16 Tom de Vries <tdevries@suse.de>
4665 * var-tracking.c (vt_initialize): Fix pre_dec handling.
4667 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4669 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
4672 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
4674 PR tree-optimization/86514
4675 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
4676 conversion to a boolean type from a type with greater precision.
4678 2018-07-16 Tom de Vries <tdevries@suse.de>
4680 * var-tracking.c (vt_initialize): Print adjusted insn slim if
4681 dump_flags request TDF_SLIM.
4683 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
4685 * fold-const.c (int_const_binop_1): Abstract...
4686 (wide_int_binop): ...wide int code here.
4687 (poly_int_binop): ...poly int code here.
4688 Abstract the rest of int_const_binop_1 into int_const_binop.
4689 * fold-const.h (wide_int_binop): New.
4690 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
4691 Remove useless PLUS/MINUS_EXPR case.
4692 (zero_nonzero_bits_from_vr): Move wide int code...
4693 (zero_nonzero_bits_from_bounds): ...here.
4694 (extract_range_from_binary_expr_1): Move mask optimization code...
4695 (range_easy_mask_min_max): ...here.
4696 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
4697 (range_easy_mask_min_max): New.
4699 2018-07-15 Jeff Law <law@redhat.com>
4702 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
4705 2018-07-14 Jim Wilson <jimw@sifive.com>
4707 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
4709 2018-07-14 Paul Koning <ni1d@arrl.net>
4711 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
4713 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
4715 * lto-streamer-out.c (copy_function_or_variable): Dump info about
4718 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
4719 Steve Munroe <munroesj52@gmail.com>
4721 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
4722 (_mm_andnot_si128): Likewise.
4723 (_mm_or_si128): Likewise.
4724 (_mm_xor_si128): Likewise.
4726 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
4729 * builtins.c (expand_builtin_memcmp): Inline the calls first
4730 when result_eq is false.
4731 (expand_builtin_strcmp): Inline the calls first.
4732 (expand_builtin_strncmp): Likewise.
4733 (inline_string_cmp): New routine. Expand a string compare
4734 call by using a sequence of char comparison.
4735 (inline_expand_builtin_string_cmp): New routine. Inline expansion
4736 a call to str(n)cmp/memcmp.
4737 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
4739 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
4741 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
4743 * config/arm/driver-arm.c: Include arm-native.h.
4744 (host_detect_local_cpu): Use auto-generated data tables.
4745 (vendors, arm_cpu_table): Delete. Move part information to ...
4746 * config/arm/arm-cpus.in: ... here.
4747 * config/arm/parsecpu.awk (gen_native): New function.
4748 (vendor, part): New CPU fields.
4749 (END): Add support for building the native CPU detection tables.
4750 * config/arm/t-arm (arm-native.h): Add build rule.
4751 (driver-arm.o): Add dependency on arm-native.h.
4753 2018-07-13 Richard Biener <rguenther@suse.de>
4756 * match.pd (addr1 - addr2): Allow either of the operand to
4759 2018-07-13 Tom de Vries <tdevries@suse.de>
4761 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
4764 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
4766 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
4767 arrays instead of numbered variables.
4769 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
4771 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
4772 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
4774 2018-07-13 Richard Biener <rguenther@suse.de>
4777 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
4778 instead of get_context_die.
4780 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
4781 Richard Biener <rguenther@suse.de>
4784 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
4785 that the loop latch destination where phi is defined.
4787 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
4789 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
4790 (riscv_expand_epilogue): Add assertion to check interrupt mode.
4791 (riscv_set_current_function): Extract getting interrupt type to new
4793 (riscv_get_interrupt_type): New function.
4794 (riscv_merge_decl_attributes): New function, checking interrupt type is
4796 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
4798 2018-07-12 Paul Koning <ni1d@arrl.net>
4800 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
4803 2018-07-12 Paul Koning <ni1d@arrl.net>
4805 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
4806 zero reference, add doloop_end instead.
4807 * doc/md.texi (decrement_and_branch_until_zero): Remove.
4808 (Looping patterns): Remove decrement_and_branch_until_zero. Add
4809 detail for doloop_end.
4811 2018-07-12 Martin Sebor <msebor@redhat.com>
4814 * attribs.c (decl_attributes): Reject conflicting attributes before
4815 calling attribute handlers.
4817 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
4819 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
4821 (gcc::dump_manager::get_dump_file_name): likewise.
4822 (dump_begin): Likewise.
4823 * dumpfile.h (dump_begin): Update prototype.
4824 (gcc::dump_manager::get_dump_file_name,
4825 gcc::dump_manager::get_dump_file_name): Update prototype.
4827 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
4829 * internal-fn.h (vectorizable_internal_fn_p): New function.
4830 * tree-vect-slp.c (compatible_calls_p): Likewise.
4831 (vect_build_slp_tree_1): Remove nops argument. Handle calls
4832 to internal functions.
4833 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
4835 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
4837 * fold-const.h (inverse_conditions_p): Declare.
4838 * fold-const.c (inverse_conditions_p): New function.
4839 * match.pd: Use inverse_conditions_p. Add folds of view_converts
4840 that test the inverse condition of a conditional internal function.
4841 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
4842 * internal-fn.c (internal_fn_mask_index): Handle conditional
4844 (vectorized_internal_fn_supported_p): New function.
4845 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
4846 (any_pred_load_store): Replace with...
4847 (need_to_predicate): ...this new variable.
4848 (redundant_ssa_names): New variable.
4849 (ifcvt_can_use_mask_load_store): Move initial checks to...
4850 (ifcvt_can_predicate): ...this new function. Handle tree codes
4851 for which a conditional internal function exists.
4852 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
4853 instead of ifcvt_can_use_mask_load_store. Update after variable
4855 (predicate_load_or_store): New function, split out from
4856 predicate_mem_writes.
4857 (check_redundant_cond_expr): New function.
4858 (value_available_p): Likewise.
4859 (predicate_rhs_code): Likewise.
4860 (predicate_mem_writes): Rename to...
4861 (predicate_statements): ...this. Use predicate_load_or_store
4862 and predicate_rhs_code.
4863 (combine_blocks, tree_if_conversion): Update after above name changes.
4864 (ifcvt_local_dce): Handle redundant_ssa_names.
4865 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
4866 general conditional functions.
4867 * tree-vect-stmts.c (vectorizable_call): Likewise.
4869 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
4870 Alan Hayward <alan.hayward@arm.com>
4871 David Sherwood <david.sherwood@arm.com>
4873 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
4874 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
4875 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
4876 plus and minus and convert them into IFN_COND_FMA-based sequences.
4877 (convert_mult_to_fma): Handle conditional plus and minus.
4879 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
4881 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
4882 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
4883 (cond_fnms_optab): New optabs.
4884 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
4886 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
4887 * internal-fn.h (get_conditional_internal_fn): Declare.
4888 (get_unconditional_internal_fn): Likewise.
4889 * internal-fn.c (cond_ternary_direct): New macro.
4890 (expand_cond_ternary_optab_fn): Likewise.
4891 (direct_cond_ternary_optab_supported_p): Likewise.
4892 (FOR_EACH_COND_FN_PAIR): Likewise.
4893 (get_conditional_internal_fn): New function.
4894 (get_unconditional_internal_fn): Likewise.
4895 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
4896 (gimple_match_op::gimple_match_op): Add a new overload for 5
4898 (gimple_match_op::set_op): Likewise.
4899 (gimple_resimplify5): Declare.
4900 * genmatch.c (decision_tree::gen): Generate simplifications for
4902 * gimple-match-head.c (gimple_simplify): Define an overload for
4903 5 operands. Handle calls with 5 arguments in the top-level overload.
4904 (convert_conditional_op): Handle conversions from unconditional
4905 internal functions to conditional ones.
4906 (gimple_resimplify5): New function.
4907 (build_call_internal): Pass a fifth operand.
4908 (maybe_push_res_to_seq): Likewise.
4909 (try_conditional_simplification): Try converting conditional
4910 internal functions to unconditional internal functions.
4911 Handle 3-operand unconditional forms.
4912 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
4913 Define ternary equivalents of the current rules for binary conditional
4915 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
4917 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
4918 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
4919 (optab): Handle them.
4920 (SVE_COND_FP_TERNARY): New int iterator.
4921 (sve_fmla_op, sve_fmad_op): New int attributes.
4922 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
4923 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
4924 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
4926 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
4928 * target.def (preferred_else_value): New target hook.
4929 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
4930 * doc/tm.texi: Regenerate.
4931 * targhooks.h (default_preferred_else_value): Declare.
4932 * targhooks.c (default_preferred_else_value): New function.
4933 * internal-fn.h (conditional_internal_fn_code): Declare.
4934 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
4935 (get_conditional_internal_fn): Use it.
4936 (conditional_internal_fn_code): New function.
4937 * gimple-match.h (gimple_match_cond): New struct.
4938 (gimple_match_op): Add a cond member function.
4939 (gimple_match_op::gimple_match_op): Update all forms to take a
4941 * genmatch.c (expr::gen_transform): Use the same condition as res_op
4942 for the suboperation, but don't specify a particular else_value.
4943 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
4944 (visit_nary_op, visit_reference_op_load): Pass
4945 gimple_match_cond::UNCOND to the gimple_match_op constructor.
4946 * gimple-match-head.c: Include tree-eh.h
4947 (convert_conditional_op): New function.
4948 (maybe_resimplify_conditional_op): Likewise.
4949 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
4950 (gimple_resimplify2): Likewise.
4951 (gimple_resimplify3): Likewise.
4952 (gimple_resimplify4): Likewise.
4953 (maybe_push_res_to_seq): Return null for conditional operations.
4954 (try_conditional_simplification): New function.
4955 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
4957 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
4959 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
4961 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
4963 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
4965 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
4967 (hash_tree): Do not hash DECL_FCONTEXT
4968 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
4969 Do not stream DECL_FCONTEXT.
4970 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
4971 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
4973 2018-07-12 Richard Biener <rguenther@suse.de>
4976 * dwarf2out.c (gen_block_die): Only output blocks when they have
4977 at least one !DECL_IGNORED_P variable.
4979 2018-07-12 Richard Biener <rguenther@suse.de>
4982 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
4983 Remove -mieee-fp handling.
4985 2018-07-12 Richard Biener <rguenther@suse.de>
4987 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
4988 left-over from last patch.
4990 2018-07-12 Jakub Jelinek <jakub@redhat.com>
4992 PR tree-optimization/86492
4993 * gimple-ssa-store-merging.c
4994 (imm_store_chain_info::coalesce_immediate_stores): Call
4995 check_no_overlap even for the merge_overlapping case. Formatting fix.
4997 2018-07-12 Richard Biener <rguenther@suse.de>
5000 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
5001 move possibly trapping operations into the conditional.
5003 2018-07-12 Richard Biener <rguenther@suse.de>
5005 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
5006 (vn_lookup_simplify_result): Remove recursion limit applied
5008 (vn_nary_build_or_lookup_1): Adjust.
5009 (try_to_simplify): Likewise.
5010 * gimple-match-head.c (gimple_resimplify1): Instead apply one
5012 (gimple_resimplify2): Likewise.
5013 (gimple_resimplify3): Likewise.
5014 (gimple_resimplify4): Likewise.
5016 2018-07-11 Jakub Jelinek <jakub@redhat.com>
5018 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
5019 Use __mmask64 type instead of __mmask8 for __M argument.
5020 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
5021 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
5023 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
5024 __mmask16 for __M argument.
5025 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
5026 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
5027 to __mmask16 instead of __mmask8.
5028 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
5029 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
5030 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
5031 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
5032 instead of __mmask16 for __U argument.
5033 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
5034 __mmask16 instead of __mmask8 for __U argument.
5035 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
5037 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
5039 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
5041 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
5043 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
5045 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
5047 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
5049 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
5051 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
5053 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
5055 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
5056 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
5057 return type as well as __M argument type and all casts from __mmask8
5059 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
5060 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
5061 return type as well as __M argument type and all casts from __mmask8
5063 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
5064 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
5065 return type as well as __M argument type and all casts from __mmask8
5067 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
5068 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
5069 return type as well as __M argument type and all casts from __mmask8
5071 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
5072 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
5075 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
5077 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
5078 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
5081 2018-07-11 Paul Koning <ni1d@arrl.net>
5083 * doc/md.texi (define_subst): Document how multiple occurrences of
5084 the same argument in the replacement pattern are handled.
5086 2018-07-11 Paul Koning <ni1d@arrl.net>
5088 * doc/extend.texi (Common Variable Attributes): Move "mode" into
5090 (Common Type Attributes): Add "mode" attribute.
5092 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
5094 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
5095 stream DECL_ORIGINAL_TYPE.
5096 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
5097 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
5098 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
5099 Do not walk original type.
5100 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
5102 (write_ts_decl_non_common_tree_pointers): Do not stream
5104 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
5105 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
5107 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
5109 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
5110 thread twice from the same starting edge.
5112 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
5114 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
5115 * gimple.c (gimple_call_nonnull_result_p): ...here...
5116 (gimple_call_nonnull_arg): ...and here.
5117 * gimple.h (gimple_call_nonnull_result_p): New.
5118 (gimple_call_nonnull_arg): New.
5120 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
5122 * config/arm/arm-cpus.in: Move information from fpu field of each
5123 cpu definition to the isa field.
5124 * config/arm/parsecpu.awk (fpu): Delete match rule.
5125 (gen_comm_data): Don't add bits from the CPU's FPU entry.
5127 2018-07-11 Richard Biener <rguenther@suse.de>
5130 * dwarf2out.c (init_sections_and_labels): Use
5131 output_asm_line_debug_info consistently.
5132 (dwarf2out_early_finish): Likewise.
5133 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
5136 2018-07-11 Richard Biener <rguenther@suse.de>
5138 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
5139 Rework father_bb setting in a way to avoid propagating constants
5140 multiple times on a loop body.
5142 2018-07-10 Mark Wielaard <mark@klomp.org>
5145 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
5147 2018-07-10 Richard Biener <rguenther@suse.de>
5149 * hash-map.h (hash_map::iterator::operator*): Return
5150 references to key and value.
5152 2018-07-10 Jakub Jelinek <jakub@redhat.com>
5155 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
5156 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
5157 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
5158 (gimplify_omp_for): For composite loops, move outer
5159 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
5160 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
5161 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
5162 TREE_LIST for both the original class iterator and the "last" helper
5163 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
5164 loop, remember has_decl_expr from outer composite loops for the
5165 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
5167 2018-07-09 Martin Sebor <msebor@redhat.com>
5171 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
5172 accessing implicitly initialized array elements.
5173 * expr.c (string_constant): Handle string initializers of
5174 character arrays within aggregates.
5175 * gimple-fold.c (fold_array_ctor_reference): Add argument.
5176 Store element offset. As a special case, handle zero size.
5177 (fold_nonarray_ctor_reference): Same.
5178 (fold_ctor_reference): Add argument. Store subobject offset.
5179 * gimple-fold.h (fold_ctor_reference): Add argument.
5181 2018-07-09 Paul Koning <ni1d@arrl.net>
5183 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
5184 (pdp11_insn_cost): New function.
5185 (pdp11_md_asm_adjust): New function.
5186 (TARGET_INVALID_WITHIN_DOLOOP): Define.
5187 (pdp11_rtx_costs): Update to match machine better.
5188 (output_addr_const_pdp11): Correct format mismatch warnings.
5189 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
5190 * config/pdp11/pdp11.md: General change to add base_cost and/or
5191 length attributes for use by new pdp11_insn_cost function.
5192 (MIN_BRANCH): Correct definition.
5194 (doloop_end): Use standard pattern name for looping pattern.
5195 (doloop_end_nocc): New.
5196 (movsf): Add another constraint alternative.
5197 (zero_extendqihi2): Add constraint alternatives for not in place
5199 (zero_extendhisi2): Remove.
5200 (shift patterns): Add CC handling variants.
5204 (define_peephole2): New peephole to recognize mov that sets CC for
5207 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5209 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
5210 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
5213 2018-07-09 Jakub Jelinek <jakub@redhat.com>
5216 * real.c (real_nextafter): Return true if result is denormal.
5218 2018-07-09 Martin Liska <mliska@suse.cz>
5220 * common.opt: Add back wrongly removed attribute.
5222 2018-07-09 Richard Biener <rguenther@suse.de>
5225 * dwarf2out.c (gen_block_die): For an early generated DIE
5226 always output high/low PC attributes.
5228 2018-07-09 Tom de Vries <tdevries@suse.de>
5230 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
5231 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
5232 onto VAR_DECL with abstract origin.
5234 2018-07-07 Jim Wilson <jimw@sifive.com>
5236 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
5238 2018-07-07 Tom de Vries <tdevries@suse.de>
5240 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
5242 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
5244 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
5246 (combine_bound): Use wide-int overflow calculation instead of
5248 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
5250 * fold-const.c (int_const_binop_2): Same.
5251 (extract_muldiv_1): Same.
5252 (fold_div_compare): Same.
5253 (fold_abs_const): Same.
5255 * poly-int.h (add): Same.
5259 * predict.c (predict_iv_comparison): Same.
5260 * profile-count.c (slow_safe_scale_64bit): Same.
5261 * simplify-rtx.c (simplify_const_binary_operation): Same.
5262 * tree-chrec.c (tree_fold_binomial): Same.
5263 * tree-data-ref.c (split_constant_offset_1): Same.
5264 * tree-if-conv.c (idx_within_array_bound): Same.
5265 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
5266 * tree-ssa-phiopt.c (minmax_replacement): Same.
5267 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
5268 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
5269 * vr-values.c (vr_values::adjust_range_with_scev): Same.
5270 * wide-int.cc (wi::add_large): Same.
5271 (wi::mul_internal): Same.
5272 (wi::sub_large): Same.
5273 (wi::divmod_internal): Same.
5274 * wide-int.h: Change overflow type to overflow_type for neg, add,
5275 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
5276 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
5277 mul_internal, divmod_internal.
5278 (overflow_type): New enum.
5279 (accumulate_overflow): New.
5281 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
5283 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
5284 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
5286 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
5288 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
5289 argument is checked for zero before entering loop, avoid checking again.
5291 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
5293 * gimplify.h (generic_expr_could_trap_p): Set as global function.
5294 * gimplify.h (generic_expr_could_trap_p): Likwise.
5295 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
5297 2018-07-06 Jakub Jelinek <jakub@redhat.com>
5299 PR tree-optimization/86401
5300 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
5301 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
5302 (fold_bit_and_mask): ... here. New helper function for match.pd.
5303 * fold-const.h (fold_bit_and_mask): Declare.
5304 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
5306 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
5309 * target.def (translate_mode_attribute): New hook.
5310 * targhooks.h (default_translate_mode_attribute): Declare.
5311 * targhooks.c (default_translate_mode_attribute): New function.
5312 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
5313 * doc/tm.texi: Regenerate.
5314 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
5315 (rs6000_translate_mode_attribute): New function.
5317 2018-07-06 Paul Koning <ni1d@arrl.net>
5319 * doc/md.texi (define_split): Document DONE and FAIL.
5320 (define_peephole2): Ditto.
5322 2018-07-05 Jeff Law <law@redhat.com>
5324 PR tree-optimization/86010
5325 * tree-ssa-dse.c (compute_trims): More aggressively trim at
5326 both the head and tail of mem* and str* calls.
5328 2018-07-05 Jim Wilson <jimw@sifive.com>
5330 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
5332 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
5334 * config/aarch64/aarch64-simd.md: correct flags text for
5335 MIN_EXPR replacement.
5337 2018-07-05 James Clarke <jrtc27@jrtc27.com>
5339 * configure: Regenerated.
5341 2018-07-05 Carl Love <cel@us.ibm.com>
5343 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
5344 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
5345 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
5346 VSX_BUILTIN_DOUBLEL_V4SF.
5348 2018-07-05 Martin Sebor <msebor@redhat.com>
5351 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
5352 than its domain to compute its the upper bound of a char array.
5354 2018-07-05 Nathan Sidwell <nathan@acm.org>
5356 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
5357 * doc/cpp.texi: Update comment.
5358 * doc/tm.texi: Rebuilt.
5359 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
5360 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
5361 * doc/extend.texi (Backwards Compatibility): Clarify it is system
5362 headers affected by extern "C".
5363 * system.h: Poison NO_IMPLICIT_EXTERN_C.
5364 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
5365 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
5366 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
5367 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
5368 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
5369 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
5370 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
5371 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
5372 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
5373 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
5374 config/sparc/sp64-elf.h, config/spu/spu.h,
5375 config/stormy16/stormy16.h, config/v850/v850.h,
5376 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
5377 define NO_IMPLICIT_EXTERN_C.
5378 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
5380 2018-07-05 Tamar Christina <tamar.christina@arm.com>
5383 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
5384 instead of GET_MODE_SIZE when comparing Units.
5386 2018-07-05 Tamar Christina <tamar.christina@arm.com>
5389 * rtlanal.c (set_noop_p): Constrain on mode change,
5390 include hard-reg-set.h
5392 2018-07-05 Tamar Christina <tamar.christina@arm.com>
5394 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
5396 2018-07-05 Jakub Jelinek <jakub@redhat.com>
5399 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
5402 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
5404 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
5406 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
5409 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
5411 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
5413 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
5416 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
5418 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
5420 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
5421 neg_*_op* variables.
5423 2018-07-04 Martin Liska <mliska@suse.cz>
5425 * tree-switch-conversion.c: Define
5426 max_ratio_for_speed and max_ratio_for_size constants.
5428 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
5429 Martin Liska <mliska@suse.cz>
5434 * common.opt: Rename align options with 'str_' prefix.
5435 * common/config/i386/i386-common.c (set_malign_value): New
5437 (ix86_handle_option): Use it to set -falign-* options/
5438 * config/aarch64/aarch64-protos.h (struct tune_params): Change
5439 type from int to string.
5440 * config/aarch64/aarch64.c: Update default values from int
5442 * config/alpha/alpha.c (alpha_override_options_after_change):
5444 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
5445 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
5446 max skip conditionally.
5447 * config/i386/freebsd.h (SUBALIGN_LOG): New.
5448 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
5449 max skip conditionally.
5450 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
5451 max skip conditionally.
5452 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
5453 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
5454 max skip conditionally.
5455 * config/i386/i386.c (struct ptt): Change type from int to
5457 (ix86_default_align): Set default values.
5458 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
5459 max skip conditionally.
5460 * config/i386/iamcu.h (SUBALIGN_LOG): New.
5461 (ASM_OUTPUT_MAX_SKIP_ALIGN):
5462 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
5463 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
5464 max skip conditionally.
5465 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
5466 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
5467 * config/i386/x86-64.h (SUBALIGN_LOG): New.
5468 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
5469 max skip conditionally.
5470 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
5471 * config/ia64/ia64.c (ia64_option_override): Set default values
5472 for alignment options.
5473 * config/m68k/m68k.c: Handle new str_align_* options.
5474 * config/mips/mips.c (mips_set_compression_mode): Change
5476 (mips_option_override): Set default values for options.
5477 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
5479 * config/rs6000/rs6000.c (rs6000_option_override_internal):
5481 * config/rx/rx.c (rx_option_override): Likewise.
5482 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
5483 (LABEL_ALIGN): Use align_labels_log.
5484 (LOOP_ALIGN): Use align_loops_align.
5485 * config/s390/s390.c (s390_asm_output_function_label): Use new
5487 * config/sh/sh.c (sh_override_options_after_change):
5488 Change type of constants.
5489 * config/spu/spu.c (spu_sched_init): Likewise.
5490 * config/sparc/sparc.c (sparc_option_override): Set default
5492 * config/visium/visium.c (visium_option_override): Likewise.
5493 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
5494 emit p2align format with last argument if it's not needed.
5495 * doc/invoke.texi: Document extended format of -falign-*.
5496 * final.c: Use align_labels alignment.
5497 * flags.h (struct target_flag_state): Change type to use
5499 (struct align_flags_tuple): New.
5500 (struct align_flags): Likewise.
5501 (align_loops_log): Redefine macro to use new types.
5502 (align_loops_max_skip): Redefine macro to use new types.
5503 (align_jumps_log): Redefine macro to use new types.
5504 (align_jumps_max_skip): Redefine macro to use new types.
5505 (align_labels_log): Redefine macro to use new types.
5506 (align_labels_max_skip): Redefine macro to use new types.
5507 (align_functions_log): Redefine macro to use new types.
5508 (align_loops): Redefine macro to use new types.
5509 (align_jumps): Redefine macro to use new types.
5510 (align_labels): Redefine macro to use new types.
5511 (align_functions): Redefine macro to use new types.
5512 (align_functions_max_skip): Redefine macro to use new types.
5513 (align_loops_value): New macro.
5514 (align_jumps_value): New macro.
5515 (align_labels_value): New macro.
5516 (align_functions_value): New macro.
5517 * function.c (invoke_set_current_function_hook): Propagate
5518 alignment values from flags to global variables default in
5520 * ipa-icf.c (sem_function::equals_wpa): Use
5521 cl_optimization_option_eq instead of memcmp.
5522 * lto-streamer.h (cl_optimization_stream_out): Support streaming
5524 (cl_optimization_stream_in): Likewise.
5525 * optc-save-gen.awk: Support strings in cl_optimization.
5526 * opth-gen.awk: Likewise.
5527 * opts.c (finish_options): Remove error checking of invalid
5529 (MAX_CODE_ALIGN): Remove.
5530 (MAX_CODE_ALIGN_VALUE): Likewise.
5531 (parse_and_check_align_values): New function.
5532 (check_alignment_argument): Likewise.
5533 (common_handle_option): Use check_alignment_argument.
5534 * opts.h (parse_and_check_align_values): Declare.
5535 * toplev.c (init_alignments): Remove.
5536 (read_log_maxskip): New.
5537 (parse_N_M): Likewise.
5538 (parse_alignment_opts): Likewise.
5539 (backend_init_target): Remove usage of init_alignments.
5540 * toplev.h (parse_alignment_opts): Declare.
5541 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
5543 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
5544 * tree.c (cl_option_hasher::equal): New.
5545 * varasm.c: Use new global macros.
5547 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
5549 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
5550 Use a simpler align directive also if MAXSKIP = ALIGN-1.
5551 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
5552 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
5553 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
5554 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
5555 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
5556 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
5557 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
5558 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
5559 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
5560 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
5561 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
5563 2018-07-04 Martin Liska <mliska@suse.cz>
5564 Jonathan Wakely <jwakely@redhat.com>
5566 * coverage.c: Use correct type.
5567 * doc/invoke.texi: Language correction.
5569 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
5572 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
5573 ENDBRANCH for non-tail call which may return via indirect branch.
5574 * doc/extend.texi: Document indirect_return attribute.
5576 2018-07-03 Martin Sebor <msebor@redhat.com>
5578 PR tree-optimization/86274
5579 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
5581 (format_floating): Correct handling of infinities and NaNs.
5583 2018-07-03 Martin Sebor <msebor@redhat.com>
5585 * print-tree.c (print_real_cst): New function.
5586 (print_node_brief): Call it.
5587 (print_node): Ditto.
5589 2018-07-03 Jeff Law <law@redhat.com>
5591 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
5592 into a single pattern.
5594 * config/h8300/h8300.md (ors code_iterator): New.
5595 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
5596 a single pattern and single splitter.
5597 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
5598 (iorqi3_1, xorqi3_1): Likewise.
5599 (iorqi3, xorqi3 expanders): Similarly.
5601 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
5602 (movmd_internal) into a single pattern using the P mode iterator.
5603 (movmd splitters): Similarly.
5604 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
5605 (movsd splitters): Similarly.
5607 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
5608 ADDB, ADDW and ADDL into a single ADD attribute which selects the
5609 right table based on the size of the operand.
5610 * config/h8300/h8300.md (length_table): Corresponding changes. All
5611 references to "addb", "addw" and "addl" changed to "add".
5612 (btst patterns): Merge two variants into a single pattern.
5613 (tstqi, tsthi): Likewise.
5614 (addhi3_incdec, addsi3_incdec): Likewise.
5615 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
5616 (mulhi3, mulsi3): Likewise.
5617 (udivhi3, udivsi3): Likewise.
5618 (divhi3, divsi3): Likewise.
5619 (andorqi3, andorhi3, andorsi3): Likewise.
5621 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
5624 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
5625 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
5627 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
5629 PR tree-optimization/85694
5630 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
5631 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
5633 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
5634 (<u>avg<mode>3_ceil): New patterns.
5636 2018-07-03 David Malcolm <dmalcolm@redhat.com>
5638 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
5639 scan-tree-dump directive.
5640 * gcc.dg/vect/slp-perm-2.c: Likewise.
5641 * gcc.dg/vect/slp-perm-3.c: Likewise.
5642 * gcc.dg/vect/slp-perm-5.c: Likewise.
5643 * gcc.dg/vect/slp-perm-6.c: Likewise.
5644 * gcc.dg/vect/slp-perm-7.c: Likewise.
5645 * gcc.dg/vect/slp-perm-8.c: Likewise.
5647 2018-07-03 Marek Polacek <polacek@redhat.com>
5650 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
5652 2018-07-03 Richard Biener <rguenther@suse.de>
5655 * tree-ssa-structalias.c (find_func_clobbers): Properly
5656 handle indirect calls.
5658 2018-07-03 Jeff Law <law@redhat.com>
5660 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
5661 (shifts): New code iterator.
5662 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
5663 expander. Fix HImode handling on H8/SX.
5664 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
5665 (subqi3, subhi3, subsi3 expanders): Likewise.
5666 (andqi3, andhi3, andsi3 expanders): Likewise.
5667 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
5668 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
5669 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
5670 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
5671 (zero_extendqihi2, zero_extendqisi2): Likewise.
5672 (extendqihi2, extendqisi2): Likewise.
5673 (rotlqi3, rotlhi3, rotlsi3): Likewise.
5674 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
5675 (rotlqi3_1, rotlhi3_1): Likewise.
5676 (logicalhi3_sn, logicalsi3_sn): Likewise.
5677 (logicalhi3, logicalsi3): Likewise.
5679 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
5681 * tree-vect-patterns.c (vect_recog_rotate_pattern)
5682 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
5683 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
5684 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
5685 type to append_pattern_def_seq instead of creating a stmt_vec_info
5687 (build_mask_conversion): Likewise. Remove vinfo argument.
5688 (vect_add_conversion_to_patterm): Likewise, renaming to...
5689 (vect_add_conversion_to_pattern): ...this.
5690 (vect_recog_mask_conversion_pattern): Update call to
5691 build_mask_conversion. Pass the vector type to
5692 append_pattern_def_seq here too.
5693 (vect_recog_gather_scatter_pattern): Update call to
5694 vect_add_conversion_to_pattern.
5696 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
5698 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
5699 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
5700 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
5701 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
5702 STMT_VINFO_PATTERN_DEF_SEQ to null here.
5703 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
5704 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
5705 append_pattern_def_seq instead of new_pattern_def_seq.
5706 (vect_recog_divmod_pattern): Do both of the above.
5707 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
5710 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
5712 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
5713 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
5714 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
5715 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
5716 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
5717 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
5718 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
5719 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
5720 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
5721 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
5722 parameter with a single stmt_vec_info.
5723 (vect_recog_func_ptr): Likewise.
5724 (vect_recog_gather_scatter_pattern): Likewise, folding in...
5725 (vect_try_gather_scatter_pattern): ...this.
5726 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
5727 the stmt_vec_info of the statement to be matched. Don't clear
5728 STMT_VINFO_RELATED_STMT.
5729 (vect_pattern_recog): Update call accordingly.
5731 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
5733 PR tree-optimization/85694
5734 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
5735 (uavgM3_ceil): Document new optabs.
5736 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
5737 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
5739 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
5740 (savg_ceil_optab): New optabs.
5741 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
5742 (vect_vect_recog_func_ptrs): Add it.
5743 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
5744 constant directly from the associated lhs.
5746 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
5748 * tree-vect-patterns.c (vect_split_statement): New function.
5749 (vect_convert_input): Use it to try to split an existing cast.
5751 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
5753 * poly-int.h (print_hex): New function.
5754 * dumpfile.h (dump_dec, dump_hex): Declare.
5755 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
5756 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
5757 min_input_precision, operation_precision and operation_sign.
5758 * tree-vect-patterns.c (vect_get_range_info): New function.
5759 (vect_same_loop_or_bb_p, vect_single_imm_use)
5760 (vect_operation_fits_smaller_type): Delete.
5761 (vect_look_through_possible_promotion): Add an optional
5762 single_use_p parameter.
5763 (vect_recog_over_widening_pattern): Rewrite to use new
5764 stmt_vec_info infomration. Handle one operation at a time.
5765 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
5766 (vect_truncatable_operation_p, vect_set_operation_type)
5767 (vect_set_min_input_precision): New functions.
5768 (vect_determine_min_output_precision_1): Likewise.
5769 (vect_determine_min_output_precision): Likewise.
5770 (vect_determine_precisions_from_range): Likewise.
5771 (vect_determine_precisions_from_users): Likewise.
5772 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
5773 (vect_vect_recog_func_ptrs): Put over_widening first.
5775 (vect_pattern_recog): Call vect_determine_precisions.
5777 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
5779 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
5780 statements that have been replaced by further pattern statements.
5781 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
5783 2018-07-03 Richard Biener <rguenther@suse.de>
5785 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
5786 always set *dt. Dump vectype in vectype overload.
5787 * dumpfile.h (dump_gimple_expr): New function.
5788 (dump_gimple_expr_loc): Likewise.
5789 * dumpfile.c (dump_gimple_expr): New function.
5790 (dump_gimple_expr_loc): Likewise.
5792 2018-07-02 Jeff Law <law@redhat.com>
5794 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
5795 the H8/300, H8/300H and H8/S variants into a single pattern.
5796 (movhi_h8300, movqi_h8300hs): Similarly.
5797 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
5798 (QHI mode iterator): New.
5800 * config/h8300/h8300.md: Remove trailing whitespace.
5802 2018-07-02 Jim Wilson <jimw@sifive.com>
5804 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
5805 instead of emit_insn for interrupt returns.
5806 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
5807 (riscv_sret, riscv_uret): Likewise.
5809 2018-07-02 David Malcolm <dmalcolm@redhat.com>
5811 * pretty-print.c (selftest::test_pp_format): Move save and restore
5812 of quotes to class auto_fix_quotes, and add an instance.
5813 * selftest.c: Include "intl.h".
5814 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
5815 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
5816 * selftest.h (selftest::auto_fix_quotes): New class.
5818 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
5820 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
5821 (aarch64_sve_prepare_conditional_op): Remove.
5822 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
5823 Allow aarch64_simd_reg_or_zero as select operand; remove
5824 the aarch64_sve_prepare_conditional_op call.
5825 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
5826 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
5827 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
5828 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
5829 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
5830 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
5831 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
5832 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
5833 and a splitters to match all of the *_any patterns.
5834 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
5836 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
5837 (SVE_COND_FP_BINARY_REV): Remove.
5838 (sve_int_op_rev, sve_fp_op_rev): New.
5839 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
5840 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
5841 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
5842 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
5843 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
5844 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
5845 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
5846 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
5847 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
5849 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
5850 Remove match_dup 1 from the inner unspec.
5851 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
5853 * config/aarch64/aarch64.md (movprfx): New attr.
5854 (length): Default movprfx to 8.
5855 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
5856 (*madd<SVE_I>, *msub<SVE_I): Likewise.
5857 (*<su>mul<SVE_I>3_highpart): Likewise.
5858 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
5859 (*v<ASHIFT><SVE_I>3): Likewise.
5860 (*<su><MAXMIN><SVE_I>3): Likewise.
5861 (*<su><MAXMIN><SVE_F>3): Likewise.
5862 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
5863 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
5864 (*div<SVE_F>4): Likewise.
5866 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
5868 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
5871 2018-07-02 Richard Biener <rguenther@suse.de>
5873 PR tree-optimization/86363
5874 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
5875 memset argument refers to a non-variable address.
5877 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
5879 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
5880 {PLUS,MINUS}_EXPR code to...
5881 (adjust_symbolic_bound): ...here,
5882 (combine_bound): ...here,
5883 (set_value_range_with_overflow): ...and here.
5885 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
5887 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
5889 (extract_range_from_abs_expr): ...here.
5891 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
5893 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
5894 -fno-omit-frame-pointer when not optimizing.
5896 2018-07-02 Martin Liska <mliska@suse.cz>
5899 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
5900 (propagate_nothrow): Likewise.
5902 2018-07-02 Martin Liska <mliska@suse.cz>
5905 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
5907 2018-07-02 David Malcolm <dmalcolm@redhat.com>
5909 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
5910 function in r262149, changing "loc" param from source_location to
5911 const dump_location_t &.
5912 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
5913 declaration, as above.
5915 2018-07-01 Paul Koning <ni1d@arrl.net>
5917 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
5918 -munit-asm, -mgnu-asm, -mdec-asm.
5919 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
5920 (pdp11_output_labelref): New.
5921 (pdp11_output_def): New.
5922 (pdp11_output_addr_vec_elt): New.
5923 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
5924 %# and %@ format codes.
5925 (pdp11_option_override): New.
5926 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
5927 (pdp11_output_ident): New.
5928 (pdp11_asm_named_section): New.
5929 (pdp11_asm_init_sections): New.
5930 (pdp11_file_start): New.
5931 (pdp11_file_end): New.
5932 (output_ascii): Use .ascii/.asciz for -mdec-asm.
5933 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
5934 %o, like %c but octal.
5935 (pdp11_option_override): New.
5936 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
5938 (DATA_SECTION_ASM_OP): Ditto.
5939 (READONLY_DATA_SECTION_ASM_OP): New.
5940 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
5941 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
5942 (ASM_OUTPUT_LABELREF): Ditto.
5943 (ASM_OUTPUT_DEF): Ditto.
5944 (ASM_OUTPUT_EXTERNAL): New.
5945 (ASM_OUTPUT_SOURCE_FILENAME): New.
5946 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
5947 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
5948 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
5949 %# and %@ format codes.
5950 * config/pdp11/pdp11.opt (mgnu-asm): New.
5951 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
5952 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
5953 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
5955 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
5957 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
5958 dereferencing path[] beyond its length.
5960 (debug_all_paths): New.
5961 (rewire_first_differing_edge): New.
5962 (adjust_paths_after_duplication): New.
5963 (duplicate_thread_path): Call adjust_paths_after_duplication.
5965 (thread_through_all_blocks): Add new argument to
5966 duplicate_thread_path.
5968 2018-06-30 Jim Wilson <jimw@sifive.com>
5970 * config/riscv/predicates.md (p2m1_shift_operand): New.
5971 (high_mask_shift_operand): New.
5972 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
5973 pattern using p2m1_shift_operand.
5974 (lshsi3_zero_extend_3+2): New combiner pattern using
5975 high_mask_shift_operand.
5977 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
5979 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
5981 (vect_recog_rotate_pattern): ...here.
5982 (vect_convert_input): Try to insert casts of invariants in the
5984 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
5985 preheader to be empty.
5987 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
5989 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
5990 vector type. If given, install it in the new statement's
5992 (vect_element_precision): New function.
5993 (vect_unpromoted_value): New struct.
5994 (vect_unpromoted_value::vect_unpromoted_value): New function.
5995 (vect_unpromoted_value::set_op): Likewise.
5996 (vect_look_through_possible_promotion): Likewise.
5997 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
5998 (vect_widened_op_tree, vect_convert_input): Likewise.
5999 (vect_convert_inputs, vect_convert_output): Likewise.
6000 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
6001 to handle the optional cast of the multiplication result and
6002 vect_widened_op_tree to detect the widened multiplication itself.
6003 Do not require the input and output of promotion casts to have
6004 the same sign, but base the signedness of the operation on the
6005 input rather than the result. If the pattern includes two
6006 promotions, check that those promotions have the same sign.
6007 Do not restrict the MULT_EXPR handling to a double-width result;
6008 handle quadruple-width results and wider. Use vect_convert_inputs
6009 to convert the inputs to the common type.
6010 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
6011 to handle the optional cast of the ABS result. Also allow a sign
6012 change or a sign extension between the ABS and MINUS.
6013 Use vect_widened_op_tree to detect the widened subtraction and use
6014 vect_convert_inputs to convert the inputs to the common type.
6015 (vect_handle_widen_op_by_const): Delete.
6016 (vect_recog_widen_op_pattern): New function.
6017 (vect_recog_widen_mult_pattern): Use it.
6018 (vect_recog_widen_shift_pattern): Likewise.
6019 (vect_recog_widen_sum_pattern): Use
6020 vect_look_through_possible_promotion to handle the promoted
6023 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
6025 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
6026 the containing gimple_seq *.
6027 * gimple-iterator.h (gsi_for_stmt): Declare it.
6028 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
6029 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
6030 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
6031 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
6032 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
6034 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
6036 (vect_mark_pattern_stmts): ...here. Handle cases in which the
6037 statement being replaced is part of an existing pattern
6038 definition sequence, inserting the new pattern statements before
6040 (vect_pattern_recog_1): Don't return a bool. If the statement
6041 is already part of a pattern, instead apply pattern matching
6042 to the pattern definition statements. Don't clear the
6043 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
6044 (vect_pattern_recog): Don't break after the first match;
6045 continue processing the pattern definition statements instead.
6046 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
6048 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
6050 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
6051 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
6052 (vect_recog_widen_sum_pattern): Use it.
6054 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
6056 * tree-vect-loop.c (vectorizable_reduction): Assert that the
6057 phi is not a pattern statement and has not been replaced by
6058 a pattern statement.
6059 * tree-vect-patterns.c (type_conversion_p): Don't check
6060 STMT_VINFO_IN_PATTERN_P.
6061 (vect_recog_vector_vector_shift_pattern): Likewise.
6062 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
6063 the pattern statement rather than the original statement; check
6064 directly for a WIDEN_MULT_EXPR here.
6065 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
6066 vect_is_simple_use to return the pattern statement rather
6067 than the original statement; use is_pattern_stmt_p to check
6068 for such a pattern statement.
6069 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
6070 to return the pattern statement rather than the original statement;
6071 don't do the same transformation here.
6072 (vect_is_simple_use): If the defining statement has been replaced
6073 by a pattern statement, return the pattern statement instead.
6074 Remove the corresponding (local) transformation from the vectype
6077 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
6079 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
6080 end and default to null.
6081 * tree-vect-loop.c (vect_create_epilog_for_reduction)
6082 (vectorizable_reduction): Update calls accordingly, dropping the
6083 gimple ** argument if the passed-back statement isn't needed.
6084 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
6085 (vect_recog_rotate_pattern): Likewise.
6086 (vect_recog_mask_conversion_pattern): Likewise.
6087 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
6088 (vect_mask_constant_operand_p): Likewise.
6089 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
6090 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
6091 (get_group_load_store_type, get_load_store_type): Likewise.
6092 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
6093 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
6094 (vectorizable_conversion, vectorizable_assignment): Likewise.
6095 (vectorizable_shift, vectorizable_operation): Likewise.
6096 (vectorizable_store, vect_is_simple_cond): Likewise.
6097 (vectorizable_condition, vectorizable_comparison): Likewise.
6098 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
6099 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
6100 and move it to the end. Cope with null def_stmt_outs.
6102 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
6104 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
6106 2018-06-29 Jeff Law <law@redhat.com>
6108 * config/v850/v850.c (v850_legitimate_address_p): Handle large
6109 displacements for TARGET_V850E2V3 and newer.
6110 (TARGET_LRA_P): Remove. Defaults to LRA now.
6111 * config/v850/v850.md (sign23byte_load): Remove.
6112 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
6113 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
6115 2018-06-29 Martin Liska <mliska@suse.cz>
6118 * coverage.c (coverage_init): Mangle full path name.
6119 * doc/invoke.texi: Document the change.
6120 * gcov-io.c (mangle_path): New.
6121 * gcov-io.h (mangle_path): Likewise.
6122 * gcov.c (mangle_name): Use mangle_path for path mangling.
6124 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6126 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
6127 if starting source register is not even.
6129 2018-06-29 Martin Liska <mliska@suse.cz>
6131 PR tree-optimization/86263
6132 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
6133 Make edge redirection.
6135 2018-06-29 David Malcolm <dmalcolm@redhat.com>
6137 * dumpfile.c (dump_loc): Add indentation based on scope depth.
6138 (dump_scope_depth): New variable.
6139 (get_dump_scope_depth): New function.
6140 (dump_begin_scope): New function.
6141 (dump_end_scope): New function.
6142 * dumpfile.h (get_dump_scope_depth): New declaration.
6143 (dump_begin_scope): New declaration.
6144 (dump_end_scope): New declaration.
6145 (class auto_dump_scope): New class.
6146 (AUTO_DUMP_SCOPE): New macro.
6147 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
6150 2018-06-29 Richard Biener <rguenther@suse.de>
6152 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
6153 compute_all_dependences succeeds.
6154 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
6155 exceed --param loop-max-datarefs-for-datadeps.
6157 2018-06-29 Jakub Jelinek <jakub@redhat.com>
6159 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
6161 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
6164 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
6165 alternative 0 in preferred_for_speed attribute.
6167 2018-06-28 Paul Koning <ni1d@arrl.net>
6169 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
6170 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
6171 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
6172 * config/pdp11/pdp11.md: Correct "length" attribute calculation
6173 for shift insn patterns.
6175 2018-06-28 David Malcolm <dmalcolm@redhat.com>
6177 * cgraph.c (cgraph_node::get_body): Replace assignments to
6178 "dump_file" with calls to set_dump_file.
6179 * dumpfile.c (alt_dump_file): Make static, and group with...
6180 (alt_flags): ...this definition.
6181 (dumps_are_enabled): New variable.
6182 (refresh_dumps_are_enabled): New function.
6183 (set_dump_file): New function.
6184 (set_alt_dump_file): New function.
6185 (gcc::dump_manager::dump_start): Replace assignments to
6186 "dump_file" and "alt_dump_file" with calls to set_dump_file and
6188 (gcc::dump_manager::dump_finish): Likewise.
6189 * dumpfile.h (alt_dump_file): Delete decl.
6190 (dumps_are_enabled): New variable decl.
6191 (set_dump_file): New function decl.
6192 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
6194 * tree-nested.c (lower_nested_functions): Replace assignments to
6195 "dump_file" with calls to set_dump_file.
6197 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
6199 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
6200 goto_locus of each outgoing edge of each basic block.
6202 2018-06-28 Richard Biener <rguenther@suse.de>
6204 * dwarf2out.c (decl_scope_table): Remove.
6205 (push_decl_scope): Likewise.
6206 (pop_decl_scope): Likewise.
6207 (gen_type_die_for_member): Do not call push/pop_decl_scope.
6208 (gen_struct_or_union_type_die): Likewise.
6209 (gen_tagged_type_die): Likewise.
6210 (dwarf2out_init): Do not initialize decl_scope_table.
6211 (dwarf2out_c_finalize): Do not free it.
6213 2018-06-28 Richard Biener <rguenther@suse.de>
6215 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
6216 deciding whether to not re-use a DIE.
6218 2018-06-28 Richard Biener <rguenther@suse.de>
6220 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
6221 DW_AT_abstract_origin attribute.
6223 2018-06-28 Martin Liska <mliska@suse.cz>
6225 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
6226 Use newly introduced constants.
6227 * tree-switch-conversion.h (struct jump_table_cluster):
6228 Define max_ratio_for_size and max_ratio_for_speed.
6230 2018-06-28 Martin Liska <mliska@suse.cz>
6232 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
6233 Add new checking assert to catch invalid state.
6234 (jump_table_cluster::can_be_handled): Handle single case
6236 (jump_table_cluster::is_beneficial): Bail out for such case.
6237 (bit_test_cluster::find_bit_tests):
6238 Add new checking assert to catch invalid state.
6239 (bit_test_cluster::can_be_handled): Handle single case
6241 (bit_test_cluster::is_beneficial): Bail out for such case.
6242 (switch_decision_tree::analyze_switch_statement):
6245 2018-06-28 Martin Liska <mliska@suse.cz>
6247 * common.opt: Introduce -completion option.
6248 * gcc.c (driver_handle_option): Handle it.
6249 (driver::main): Print completions if completion
6251 * opt-suggestions.c (option_proposer::get_completions):
6253 (option_proposer::suggest_completion): Likewise.
6254 (option_proposer::find_param_completions): Likewise.
6255 (verify_autocompletions): Likewise.
6256 (test_completion_valid_options): Likewise.
6257 (test_completion_valid_params): Likewise.
6258 (in_completion_p): Likewise.
6259 (empty_completion_p): Likewise.
6260 (test_completion_partial_match): Likewise.
6261 (test_completion_garbage): Likewise.
6262 (opt_proposer_c_tests): Likewise.
6263 * opt-suggestions.h: Declare new functions.
6264 * opts.c (common_handle_option): Handle OPT__completion_.
6265 * selftest-run-tests.c (selftest::run_tests): Add
6266 opt_proposer_c_tests.
6267 * selftest.c (assert_str_startswith): New.
6268 * selftest.h (assert_str_startswith): Likewise.
6269 (opt_proposer_c_tests): New.
6270 (ASSERT_STR_STARTSWITH): Likewise.
6272 2018-06-28 Martin Liska <mliska@suse.cz>
6274 * Makefile.in: Add opt-suggestions.o.
6275 * gcc-main.c: Include opt-suggestions.h.
6276 * gcc.c (driver::driver): Likewise.
6277 (driver::~driver): Remove m_option_suggestions.
6278 (driver::build_option_suggestions): Moved to option_proposer.
6279 (driver::suggest_option): Likewise.
6280 (driver::handle_unrecognized_options): Use option_proposer.
6281 * gcc.h (class driver): Add new memver m_option_proposer.
6282 * opt-suggestions.c: New file.
6283 * opt-suggestions.h: New file.
6285 2018-06-28 Martin Liska <mliska@suse.cz>
6287 * vec.h (class auto_string_vec): New (moved from auto_argvec).
6288 (auto_string_vec::~auto_string_vec): Likewise.
6290 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
6292 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
6293 prevent_decl_creation_for_types fields up and add reset_location field.
6294 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
6295 statement if id->reset_location is true.
6296 (copy_edges_for_bb): Do not set goto_locus on the new edges if
6297 id->reset_location is true.
6298 (copy_phis_for_bb): Force input_location on the arguments if
6299 id->reset_location is true.
6300 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
6301 is set on the function to be inlined.
6303 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
6305 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
6307 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
6309 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
6310 registers for Pmode.
6311 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
6312 hard registers for the clobbered pseudo.
6314 2018-06-27 Paul Koning <ni1d@arrl.net>
6316 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
6317 mutually exclusive options.
6318 * config/pdp11/constraints.md (h): New constraint.
6319 (O): Update definition to match shift code generation.
6320 (D): New constraint.
6321 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
6323 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
6325 (output_jump): Change arguments.
6326 (pdp11_fixed_cc_regs): New function.
6327 (pdp11_cc_mode): Ditto.
6328 (pdp11_expand_shift): Ditto.
6329 (pdp11_assemble_shift): Ditto.
6330 (pdp11_small_shift): Ditto.
6331 (pdp11_branch_cost): Remove.
6332 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
6334 (pdp11_register_move_cost): Update for CC registers.
6335 (pdp11_rtx_costs): Add case for LSHIFTRT.
6336 (pdp11_output_jump): Add CCNZ mode conditional branches.
6337 (notice_update_cc_on_set): Remove.
6338 (pdp11_cc_mode): New function.
6339 (simple_memory_operand): Correct pre/post decrement case.
6340 (no_side_effect_operand): New function.
6341 (pdp11_regno_reg_class): Add CC_REGS class.
6342 (pdp11_fixed_cc_regs): New function.
6343 (pdp11_small_shift): New function.
6344 (pdp11_expand_shift): New function to expand shift insns.
6345 (pdp11_assemble_shift): New function to output shifts.
6346 (pdp11_branch_cost): Remove.
6347 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
6348 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
6349 (WCHAR_TYPE): Ditto.
6350 (PTRDIFF_TYPE): Ditto.
6351 (ADJUST_INSN_LENGTH): New macro.
6352 (FIXED_REGISTERS): Add CC registers.
6353 (CALL_USED_REGISTERS): Ditto.
6355 (REG_CLASS_NAMES): Ditto.
6356 (REG_CLASS_CONTENTS): Ditto.
6357 (SELECT_CC_MODE): Use new function.
6358 (TARGET_FLAGS_REGNUM): New macro.
6359 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
6360 (cc0_reg_rtx): Remove.
6361 (CC_STATUS_MDEP): Remove.
6362 (CC_STATUS_MDEFP_INIT): Remove.
6363 (CC_IN_FPU): Remove.
6364 (NOTICE_UPDATE_CC): Remove.
6365 (REGISTER_NAMES): Add CC registers.
6366 (BRANCH_COST): Change to constant 1.
6367 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
6369 * config/pdp11/pdp11.opt (mbcopy): Remove.
6370 (mbcopy-builtin): Remove.
6371 (mbranch-cheap): Remove.
6372 (mbranch-expensive): Remove.
6373 * config/pdp11/predicates.md (expand_shift_operand): Update to
6374 match shift code generation.
6375 (ccnz_operator): New predicate.
6376 * doc/invoke.texi (PDP-11 Options): Remove deleted options
6377 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
6378 Remove non-existent option -mabshi, -mno-abshi. Document mutually
6380 * doc/md.texi (PDP-11): Document new D and h constraints. Update
6381 description of O constraint.
6383 2018-06-27 Jeff Law <law@redhat.com>
6384 Austin Law <austinklaw@gmail.com>
6386 * config/v850/v850.md (addsi3_set_flags): New pattern.
6387 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
6388 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
6389 (zero_extendhisi2_v850_set_flags): Likewise.
6390 (zero_extendqisi2_v850_set_flags): Likewise.
6391 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
6392 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
6393 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
6395 * config/v850/v850-protos.h (notice_update_cc): Remove.
6396 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
6397 (v850_print_operand): Handle 'D' and "d".
6398 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
6399 Add handling of arithmetic/logical operations compared against zero.
6400 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
6401 Do not look at v850_compare_op, instead get mode from last argument.
6402 (v850_gen_compare): Remove
6403 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
6404 after reload for prologue insns.
6405 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
6407 (construct_save_jarl): Likewise.
6408 (TARGET_FLAGS_REGNUM): Define.
6409 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
6410 (NOTICE_UPDATE_CC): Remove.
6411 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
6412 than cc0. Conditionalize on reload_completed.
6413 (cmpsi_insn, setfcc_insn): Likewise.
6414 (tst1 splitter): Turn into define_and_split which sets the flags
6416 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
6417 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
6418 (cstoresf4, cstoredf4): Clobber the flags.
6419 (cmpsi, cmpsf, cmpdf): Remove expanders.
6420 (setf_insn): Remove pattern.
6421 (addsi3): Turn into define_and_split which clobbers the flags after
6422 reload and a suitable pattern (addsi3_clobber_flags) for use after
6424 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
6425 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
6426 (ashrsi3, ashrsi3_v850e2): Likewise.
6427 (bins): Clobber the flags.
6428 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
6429 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
6430 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
6431 (call_value_internal_short, call_value_internal_long): Likewise.
6432 (callt_save_interrupt, callt_return_interrupt): Likewise.
6433 (save_interrupt, return_interrupt): Likewise.
6434 (callt_save_all_interrupt, save_all_interrupt): Likewise.
6435 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
6436 (restore_all_interrupt, _restore_all_interrupt): Likewise.
6437 (All FP comparisons): Only allow after reload has completed.
6439 (divh, divhu): Tweak output template.
6440 (branch_z_normal, branch_z_invert): Remove
6441 (branch_nz_normal, branch_nz_invert): Likewise.
6442 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
6444 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
6445 * config/v850/v850.c (notice_update_cc): Remove.
6446 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
6447 (CC_NO_CARRY): Likewise.
6448 (NOTICE_UPDATE_CC): Define to nothing.
6449 * config/v850/v850.md: Remove block comment on cc0 handling
6450 Remove "cc" attribute from all patterns. Remove cc_status handling
6451 from all patterns. Minor formatting fixes.
6453 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6455 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
6456 (cortex-a76.cortex-a55): Likewise.
6457 * config/aarch64/aarch64-tune.md: Regenerate.
6458 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
6459 cortex-a76.cortex-a55.
6461 2018-06-27 Jeff Law <law@redhat.com>
6463 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
6464 (MULTILIB_DIRNAMES): Similarly.
6466 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
6468 * gimple.h (gimple_return_retbnd): Delete.
6469 (gimple_return_set_retbnd): Likewise.
6470 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
6471 gimple_return_set_retbnd.
6472 * gimple-pretty-print.c (dump_gimple_return): Remove call to
6473 gimple_return_retbnd and adjust.
6474 * tree-inline.h (struct copy_body_data): Remove retbnd field.
6475 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
6476 Explicitly return NULL in a couple more cases. Move assertion
6477 on debug statements and remove unreachable code.
6478 (reset_debug_binding): Do not test id->retbnd.
6479 (expand_call_inline): Do not set it.
6481 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
6483 * configure.ac: Add --disable-gcov option.
6484 * configure: Regenerate.
6485 * Makefile.in: Honour @enable_gcov@.
6486 * doc/install.texi: Document --disable-gcov.
6488 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6490 * config/arm/arm-cpus.in (cortex-a76): New entry.
6491 (cortex-a76.cortex-a55): Likewise.
6492 * config/arm/arm-tables.opt: Regenerate.
6493 * config/arm/arm-tune.md: Likewise.
6494 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
6495 * doc/invoke.texi (ARM Options): Document cortex-a76 and
6496 cortex-a76.cortex-a55.
6498 2018-06-27 Tamar Christina <tamar.christina@arm.com>
6501 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
6503 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
6505 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
6507 (EPILOGUE_USES): Likewise.
6509 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
6511 * tree-inline.c (remap_location): New function extracted from...
6512 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
6513 (copy_phis_for_bb): ...here. Call remap_location.
6514 (copy_cfg_body): Adjust call to copy_edges_for_bb.
6516 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
6518 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
6519 unaligned vsx for 16B memset.
6521 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
6524 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
6525 ieee128_float_type_node to long_double_type_node unless
6526 TARGET_LONG_DOUBLE_128 is set.
6528 2018-06-26 David Malcolm <dmalcolm@redhat.com>
6530 * cfgloop.c (get_loop_location): Convert return type from
6531 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
6532 by implicit construction from rtx_insn *, and using
6533 dump_user_location_t::from_function_decl for the fallback case.
6534 * cfgloop.h (get_loop_location): Convert return type from
6535 location_t to dump_user_location_t.
6536 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
6537 dump_printf_loc to pass in a dump_location_t rather than a
6538 location_t, via the gimple stmt.
6539 * coverage.c (get_coverage_counts): Update calls to
6540 dump_printf_loc to pass in dump_location_t rather than a
6542 * doc/optinfo.texi (Dump types): Convert example of
6543 dump_printf_loc from taking "locus" to taking "insn". Update
6544 description of the "_loc" calls to cover dump_location_t.
6545 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
6547 (dump_user_location_t::dump_user_location_t): New constructors,
6548 from gimple *stmt and rtx_insn *.
6549 (dump_user_location_t::from_function_decl): New function.
6550 (dump_loc): Make static.
6551 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
6552 const dump_location_t &.
6553 (dump_generic_expr_loc): Delete.
6554 (dump_printf_loc): Convert param "loc" from location_t to
6555 const dump_location_t &.
6556 (selftest::test_impl_location): New function.
6557 (selftest::dumpfile_c_tests): New function.
6558 * dumpfile.h: Include "profile-count.h".
6559 (class dump_user_location_t): New class.
6560 (struct dump_impl_location_t): New struct.
6561 (class dump_location_t): New class.
6562 (dump_printf_loc): Convert 2nd param from source_location to
6563 const dump_location_t &.
6564 (dump_generic_expr_loc): Delete.
6565 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
6566 const dump_location_t &.
6567 * gimple-fold.c (fold_gimple_assign): Update call to
6568 dump_printf_loc to pass in a dump_location_t rather than a
6569 location_t, via the gimple stmt.
6570 (gimple_fold_call): Likewise.
6571 * gimple-loop-interchange.cc
6572 (loop_cand::analyze_iloop_reduction_var): Update for change to
6573 check_reduction_path.
6574 (tree_loop_interchange::interchange): Update for change to
6576 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
6577 change in return-type of find_loop_location.
6578 (graphite_regenerate_ast_isl): Likewise.
6579 * graphite-optimize-isl.c (optimize_isl): Likewise.
6580 * graphite.c (graphite_transform_loops): Likewise.
6581 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
6582 pass in a dump_location_t rather than a location_t, via the
6584 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
6585 * ipa.c (walk_polymorphic_call_targets): Likewise.
6586 * loop-unroll.c (report_unroll): Convert "locus" param from
6587 location_t to dump_location_t.
6588 (decide_unrolling): Update for change to get_loop_location's
6590 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
6591 location_t to dump_user_location_t.
6592 (grid_find_single_omp_among_assignments_1): Updates calls to
6593 dump_printf_loc to pass in a dump_location_t rather than a
6594 location_t, via the gimple stmt.
6595 (grid_parallel_clauses_gridifiable): Convert "tloc" from
6596 location_t to dump_location_t. Updates calls to dump_printf_loc
6597 to pass in a dump_location_t rather than a location_t, via the
6599 (grid_inner_loop_gridifiable_p): Likewise.
6600 (grid_dist_follows_simple_pattern): Likewise.
6601 (grid_gfor_follows_tiling_pattern): Likewise.
6602 (grid_target_follows_gridifiable_pattern): Likewise.
6603 (grid_attempt_target_gridification): Convert initialization
6604 of local "grid" from memset to zero-initialization; FIXME: does
6605 this require C++11? Update call to dump_printf_loc to pass in a
6606 optinfo_location rather than a location_t, via the gimple stmt.
6607 * profile.c (read_profile_edge_counts): Updates call to
6608 dump_printf_loc to pass in a dump_location_t rather than a
6610 (compute_branch_probabilities): Likewise.
6611 * selftest-run-tests.c (selftest::run_tests): Call
6613 * selftest.h (dumpfile_c_tests): New decl.
6614 * tree-loop-distribution.c (pass_loop_distribution::execute):
6615 Update for change in return type of find_loop_location.
6616 * tree-parloops.c (parallelize_loops): Likewise.
6617 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
6618 "locus" from location_t to dump_user_location_t.
6619 (canonicalize_loop_induction_variables): Likewise.
6620 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
6621 for change in return type of find_loop_location.
6622 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
6623 to dump_printf_loc to pass in a dump_location_t rather than a
6624 location_t, via the stmt.
6625 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
6627 * tree-vect-loop-manip.c (find_loop_location): Convert return
6628 type from source_location to dump_user_location_t.
6629 (vect_do_peeling): Update for above change.
6630 (vect_loop_versioning): Update for change in type of
6632 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
6633 from location_t to dump_user_location_t.
6634 (vect_estimate_min_profitable_iters): Update for change in type
6636 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
6637 location_t to dump_location_t.
6638 (vect_slp_bb): Update for change in type of vect_location.
6639 * tree-vectorizer.c (vect_location): Convert from source_location
6640 to dump_user_location_t.
6641 (try_vectorize_loop_1): Update for change in vect_location's type.
6642 (vectorize_loops): Likewise.
6643 (increase_alignment): Likewise.
6644 * tree-vectorizer.h (vect_location): Convert from source_location
6645 to dump_user_location_t.
6646 (find_loop_location): Convert return type from source_location to
6647 dump_user_location_t.
6648 (check_reduction_path): Convert 1st param from location_t to
6649 dump_user_location_t.
6650 * value-prof.c (check_counter): Update call to dump_printf_loc to
6651 pass in a dump_user_location_t rather than a location_t; update
6652 call to error_at for change in type of "locus".
6653 (check_ic_target): Update call to dump_printf_loc to
6654 pass in a dump_user_location_t rather than a location_t, via the
6657 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
6659 * config/s390/s390.h (enum processor_flags): Do not use
6660 default tune parameter when -march was specified.
6662 2018-06-26 Jakub Jelinek <jakub@redhat.com>
6665 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
6666 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
6669 2018-06-26 Richard Biener <rguenther@suse.de>
6671 PR tree-optimization/86287
6673 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
6674 (vect_analyze_loop): Initialize n_stmts.
6676 2018-06-26 Richard Biener <rguenther@suse.de>
6679 * fold-const.c (fold_convertible_p): Pointer extension
6682 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
6685 * dwarf2out.c (loc_list_has_views): Adjust comments.
6686 (dw_loc_list): Split single cross-partition range with
6689 2018-06-25 Jeff Law <law@redhat.com>
6691 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
6692 on -mbig-switch by default.
6694 * config/v850/predicates.md (const_float_1_operand): Fix match_code
6696 (const_float_0_operand): Remove unused predicate.
6697 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
6698 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
6699 (recipsf2): New expander. Original pattern now called
6701 (recipdf2, recipdf2_insn): Similarly.
6702 (rsqrtsf2, rsqrtsf2_insn): Similarly
6703 (rsqrtdf2, rsqrtdf2_insn): Similarly
6705 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
6707 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
6708 Simplify logic for FreeBSD (twice).
6710 2018-06-25 Martin Sebor <msebor@redhat.com>
6712 PR tree-optimization/86204
6713 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
6714 a strnlen result if it's less than the length of the string.
6716 2018-06-25 Martin Sebor <msebor@redhat.com>
6718 PR tree-optimization/85700
6719 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
6720 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
6721 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
6723 2018-06-25 Martin Sebor <msebor@redhat.com>
6725 * doc/extend.texi (Zero-length arrays): Update and clarify.
6727 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
6729 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
6730 added IEEE/IBM long double multilib support on PowerPC little
6731 endian Linux systems.
6732 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
6733 (MULTILIB_DEFAULTS): Likewise.
6734 * config/rs6000/rs6000.c (rs6000_option_override_internal):
6736 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
6737 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
6738 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
6740 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
6743 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
6744 (REORDER_45): Likewise.
6746 2018-06-25 Jeff Law <law@redhat.com>
6748 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
6749 dividend to 32 bits. Adjust length.
6750 (udivmodhi4): Cleanup output template. Fix length.
6752 2018-06-25 Carl Love <cel@us.ibm.com>
6754 * config/rs6000/vsx.md: Change word selector to prefered location.
6756 2018-06-25 Richard Biener <rguenther@suse.de>
6758 PR tree-optimization/86304
6759 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
6760 epilogue-if-converted loops as well.
6762 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
6764 * lto-section-out.c (lto_begin_section): Do not print section
6765 name for noaddr and unnumbered dumps.
6767 2018-06-25 Richard Biener <rguenther@suse.de>
6769 * tree-vectorizer.h (struct vec_info_shared): New structure
6770 with parts split out from struct vec_info and loop_nest from
6771 struct _loop_vec_info.
6772 (struct vec_info): Adjust accordingly.
6773 (struct _loop_vec_info): Likewise.
6774 (LOOP_VINFO_LOOP_NEST): Adjust.
6775 (LOOP_VINFO_DATAREFS): Likewise.
6776 (LOOP_VINFO_DDRS): Likewise.
6777 (struct _bb_vec_info): Likewise.
6778 (BB_VINFO_DATAREFS): Likewise.
6779 (BB_VINFO_DDRS): Likewise.
6780 (struct _stmt_vec_info): Add dr_aux member.
6781 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
6782 (DR_MISALIGNMENT_UNINITIALIZED): New.
6783 (set_dr_misalignment): Adjust.
6784 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
6785 (vect_analyze_loop): Adjust prototype.
6786 (vect_analyze_loop_form): Likewise.
6787 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
6788 Compute dependences lazily.
6789 (vect_record_base_alignments): Use shared datarefs/ddrs.
6790 (vect_verify_datarefs_alignment): Likewise.
6791 (vect_analyze_data_refs_alignment): Likewise.
6792 (vect_analyze_data_ref_accesses): Likewise.
6793 (vect_analyze_data_refs): Likewise.
6794 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
6795 constructor parameter for shared part.
6796 (vect_analyze_loop_form): Pass in shared part and adjust.
6797 (vect_analyze_loop_2): Pass in storage for the number of
6798 stmts. Move loop nest finding to the caller. Compute
6800 (vect_analyze_loop): Pass in shared part.
6801 (vect_transform_loop): Verify shared datarefs are unchanged.
6802 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
6803 constructor parameter for shared part.
6804 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
6805 (vect_slp_bb): Verify shared datarefs are unchanged before
6807 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
6809 (new_stmt_vec_info): Initialize DR_AUX misalignment to
6810 DR_MISALIGNMENT_UNINITIALIZED.
6811 * tree-vectorizer.c (vec_info::vec_info): Add constructor
6812 parameter for shared part.
6813 (vec_info::~vec_info): Adjust.
6814 (vec_info_shared::vec_info_shared): New.
6815 (vec_info_shared::~vec_info_shared): Likewise.
6816 (vec_info_shared::save_datarefs): Likewise.
6817 (vec_info_shared::check_datarefs): Likewise.
6818 (try_vectorize_loop_1): Construct shared part live for analyses
6819 of a single loop for multiple vector sizes.
6820 * tree-parloops.c (gather_scalar_reductions): Adjust.
6822 2018-06-25 Richard Biener <rguenther@suse.de>
6824 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
6825 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
6826 (vect_analyze_data_refs): Remove similar code from here and
6827 simplify accordingly.
6829 2018-06-25 Richard Biener <rguenther@suse.de>
6831 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
6832 for reverse storage order accesses rather than asserting
6833 they cannot happen here.
6835 2018-06-25 Tom de Vries <tdevries@suse.de>
6838 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
6839 Use data16 instead of .byte for insn prefix.
6841 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
6844 * parser.c (make_char_string_pack): Pass this literal chars
6845 through cpp_interpret_string.
6846 (cp_parser_userdef_numeric_literal): Check the result of
6847 make_char_string_pack.
6849 2018-06-24 Maya Rashish <coypu@sdf.org>
6851 * ginclude/stddef.h: Simplify conditions around avoiding
6852 re-definition of __size_t.
6854 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
6856 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
6857 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
6859 2018-06-22 Maya Rashish <coypu@sdf.org>
6861 * doc/invoke.texi (mno-fancy-math-387): Update for changes
6862 made to OpenBSD and NetBSD through the years.
6864 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
6866 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
6867 behavior of vec_pack (vector double, vector double) to match
6868 behavior of vec_float2 (vector double, vector double).
6870 2018-06-22 Olivier Hainque <hainque@adacore.com>
6872 * gimplify.c (gimplify_function_tree): Prevent creation
6873 of a trampoline for the address of the current function
6874 passed to entry/exit instrumentation hooks.
6876 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
6879 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
6882 2018-06-22 Martin Liska <mliska@suse.cz>
6884 PR tree-optimization/86263
6885 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
6886 Bail out if is_enabled is false.
6887 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
6889 (jump_table_cluster::is_enabled): New function.
6891 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
6893 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
6894 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
6895 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
6896 (lto_input_ts_binfo_tree_pointers): Likewise.
6897 * tree-streamer-out.c (streamer_write_tree_bitfields,
6898 write_ts_binfo_tree_pointers): Likewise.
6899 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
6901 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
6903 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
6905 2018-06-22 Martin Liska <mliska@suse.cz>
6907 * symbol-summary.h (get): Make it pure and inline move
6908 functionality from ::get function.
6909 (get): Remove and inline into ::get and ::get_create.
6910 (get_create): Move code from ::get function.
6912 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6915 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
6916 -x assembler-with-cpp.
6918 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6920 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
6921 _FILE_OFFSET_BITS=64 for C++.
6923 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
6925 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
6926 conversion insn that shows up when pr85657-3.c is compiled using
6927 IEEE 128-bit long double.
6928 (neg<mode>2_internal): Use the correct mode to check whether the
6929 mode is IBM extended.
6930 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
6931 multiply and divide external functions from being created more
6934 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
6936 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
6938 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
6939 the edge can be forwarded.
6940 (cfg_layout_merge_blocks): Likewise.
6942 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
6944 * except.c (finish_eh_generation): Commit edge insertions only after
6945 the EH edges have been redirected from post-landing to landing pads.
6947 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
6949 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
6950 create_tmp_var_for to create the FRAME decl.
6951 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
6953 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
6955 * tree-inline.c (copy_edges_for_bb): Minor tweak.
6956 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
6957 debug statement when resetting its value.
6958 (expand_call_inline): Copy the locus of the call onto the assignment
6959 of the return value, if any. Use local variable in more cases.
6961 2018-06-21 Martin Liska <mliska@suse.cz>
6963 * ipa-pure-const.c (propagate_nothrow): Use
6964 funct_state_summaries->get.
6965 (dump_malloc_lattice): Likewise.
6966 (propagate_malloc): Likewise.
6968 2018-06-21 Richard Biener <rguenther@suse.de>
6970 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
6971 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
6972 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
6974 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
6975 BLOCK_ABSTRACT_ORIGIN unconditionally.
6977 2018-06-21 David Malcolm <dmalcolm@redhat.com>
6979 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
6981 * ipa-reference.c (ipa_reference_c_finalize): Delete
6982 ipa_ref_opt_sum_summaries and set it to NULL.
6984 2018-06-21 Tom de Vries <tdevries@suse.de>
6986 PR tree-optimization/85859
6987 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
6988 test with comment from bb_no_side_effects_p.
6990 2018-06-21 Richard Biener <rguenther@suse.de>
6992 PR tree-optimization/86232
6993 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
6994 max for constant niter.
6996 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
6998 * config/aarch64/aarch64-simd.md
6999 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
7001 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
7003 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
7004 Make opernads of the unspec commutative.
7006 2018-06-21 Richard Biener <rguenther@suse.de>
7008 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
7009 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
7010 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
7011 (vect_analyze_data_ref_dependence): Re-order checks to deal with
7013 (vect_record_base_alignments): Do not record base alignment
7014 for gathers or scatters.
7015 (vect_compute_data_ref_alignment): Drop return value that is always
7016 true. Bail out early for gathers or scatters.
7017 (vect_enhance_data_refs_alignment): Bail out early for gathers
7019 (vect_find_same_alignment_drs): Likewise.
7020 (vect_analyze_data_refs_alignment): Remove dead code.
7021 (vect_slp_analyze_and_verify_node_alignment): Likewise.
7022 (vect_analyze_data_refs): For possible gathers or scatters do
7023 not create an alternate DR, just check their possible validity
7024 and mark them. Adjust DECL_NONALIASED handling to not rely
7026 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
7027 update inits of gathers or scatters.
7028 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
7029 Also copy gather/scatter flag to pattern vinfo.
7031 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
7033 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
7034 behavior of vec_packsu (vector unsigned long long, vector unsigned
7035 long long) to match behavior of vec_packs with same signature.
7037 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
7038 Thomas Schwinge <thomas@codesourcery.com>
7039 Cesar Philippidis <cesar@codesourcery.com>
7041 * gimplify.c (gimplify_scan_omp_clauses): Add support for
7042 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
7043 (gimplify_adjust_omp_clauses): Likewise.
7044 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
7045 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
7046 (gimplify_omp_target_update): Update handling of acc update and
7048 * omp-low.c (install_var_field): Remove unused parameter
7049 base_pointers_restrict.
7050 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
7051 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
7053 (omp_target_base_pointers_restrict_p): Delete.
7054 (scan_omp_target): Update call to scan_sharing_clauses.
7055 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
7057 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
7058 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
7059 (convert_local_omp_clauses): Likewise.
7060 * tree-pretty-print.c (dump_omp_clause): Likewise.
7061 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
7063 (omp_clause_code_name): Likewise.
7065 2018-06-20 Jakub Jelinek <jakub@redhat.com>
7068 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
7071 PR tree-optimization/86231
7072 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
7073 anti-range don't overwrite *vr0min before using it to compute *vr0max.
7075 2018-06-20 Tom de Vries <tdevries@suse.de>
7077 PR tree-optimization/86097
7078 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
7079 iv type if signedness of iv type is not the same as that of *nit.
7081 2018-06-20 Jakub Jelinek <jakub@redhat.com>
7083 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
7084 EDGE_EH edges, verify they are all EDGE_EH.
7086 2018-06-20 Maya Rashish <coypu@sdf.org>
7088 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
7090 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7092 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
7093 * config/aarch64/aarch64.c (xgene1_tunings): Add
7094 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
7095 (aarch64_mode_valid_for_sched_fusion_p):
7096 Allow 16-byte modes.
7097 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
7098 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
7100 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
7102 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
7103 * config/aarch64/iterators.md (VQ2): New mode iterator.
7105 2018-06-20 Martin Liska <mliska@suse.cz>
7107 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
7108 Change default ratio from 10 to 8.
7110 2018-06-20 Martin Liska <mliska@suse.cz>
7112 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
7114 (bit_test_cluster::find_bit_tests): Likewise.
7115 (switch_decision_tree::analyze_switch_statement): Find clusters.
7116 * tree-switch-conversion.h (struct jump_table_cluster): Document
7119 2018-06-20 Martin Liska <mliska@suse.cz>
7121 * tree-switch-conversion.c (switch_conversion::collect):
7122 Record m_uniq property.
7123 (switch_conversion::expand): Bail out for special conditions.
7124 (group_cluster::~group_cluster): New.
7125 (group_cluster::group_cluster): Likewise.
7126 (group_cluster::dump): Likewise.
7127 (jump_table_cluster::emit): New.
7128 (switch_decision_tree::fix_phi_operands_for_edges): New.
7129 (struct case_node): Remove struct.
7130 (jump_table_cluster::can_be_handled): New.
7131 (case_values_threshold): Moved to header.
7132 (reset_out_edges_aux): Likewise.
7133 (jump_table_cluster::is_beneficial): New.
7134 (bit_test_cluster::can_be_handled): Likewise.
7135 (add_case_node): Remove.
7136 (bit_test_cluster::is_beneficial): New.
7137 (case_bit_test::cmp): New.
7138 (bit_test_cluster::emit): New.
7139 (expand_switch_as_decision_tree_p): Remove.
7140 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
7141 (fix_phi_operands_for_edge): Likewise.
7142 (switch_decision_tree::analyze_switch_statement): New.
7143 (compute_cases_per_edge): Move ...
7144 (switch_decision_tree::compute_cases_per_edge): ... here.
7145 (try_switch_expansion): Likewise.
7146 (switch_decision_tree::try_switch_expansion): Likewise.
7147 (record_phi_operand_mapping): Likewise.
7148 (switch_decision_tree::record_phi_operand_mapping): Likewise.
7149 (emit_case_decision_tree): Likewise.
7150 (switch_decision_tree::emit): Likewise.
7151 (balance_case_nodes): Likewise.
7152 (switch_decision_tree::balance_case_nodes): Likewise.
7153 (dump_case_nodes): Likewise.
7154 (switch_decision_tree::dump_case_nodes): Likewise.
7155 (emit_jump): Likewise.
7156 (switch_decision_tree::emit_jump): Likewise.
7157 (emit_cmp_and_jump_insns): Likewise.
7158 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
7159 (emit_case_nodes): Likewise.
7160 (switch_decision_tree::emit_case_nodes): Likewise.
7161 (conditional_probability): Remove.
7162 * tree-switch-conversion.h (enum cluster_type): New.
7164 (struct cluster): Likewise.
7165 (cluster::cluster): Likewise.
7166 (struct simple_cluster): Likewise.
7167 (simple_cluster::simple_cluster): Likewise.
7168 (struct group_cluster): Likewise.
7169 (struct jump_table_cluster): Likewise.
7170 (struct bit_test_cluster): Likewise.
7171 (struct min_cluster_item): Likewise.
7172 (struct case_tree_node): Likewise.
7173 (case_tree_node::case_tree_node): Likewise.
7174 (jump_table_cluster::case_values_threshold): Likewise.
7175 (struct case_bit_test): Likewise.
7176 (struct switch_decision_tree): Likewise.
7177 (struct switch_conversion): Likewise.
7178 (switch_decision_tree::reset_out_edges_aux): Likewise.
7180 2018-06-20 Martin Liska <mliska@suse.cz>
7182 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
7183 (hoist_edge_and_branch_if_true): Likewise.
7184 (expand_switch_using_bit_tests_p): Likewise.
7185 (struct case_bit_test): Likewise.
7186 (case_bit_test_cmp): Likewise.
7187 (emit_case_bit_tests): Likewise.
7188 (switch_conversion::switch_conversion): New class.
7189 (struct switch_conv_info): Remove old struct.
7190 (collect_switch_conv_info): More to ...
7191 (switch_conversion::collect): ... this.
7192 (check_range): Likewise.
7193 (switch_conversion::check_range): Likewise.
7194 (check_all_empty_except_final): Likewise.
7195 (switch_conversion::check_all_empty_except_final): Likewise.
7196 (check_final_bb): Likewise.
7197 (switch_conversion::check_final_bb): Likewise.
7198 (create_temp_arrays): Likewise.
7199 (switch_conversion::create_temp_arrays): Likewise.
7200 (free_temp_arrays): Likewise.
7201 (gather_default_values): Likewise.
7202 (switch_conversion::gather_default_values): Likewise.
7203 (build_constructors): Likewise.
7204 (switch_conversion::build_constructors): Likewise.
7205 (constructor_contains_same_values_p): Likewise.
7206 (switch_conversion::contains_same_values_p): Likewise.
7207 (array_value_type): Likewise.
7208 (switch_conversion::array_value_type): Likewise.
7209 (build_one_array): Likewise.
7210 (switch_conversion::build_one_array): Likewise.
7211 (build_arrays): Likewise.
7212 (switch_conversion::build_arrays): Likewise.
7213 (gen_def_assigns): Likewise.
7214 (switch_conversion::gen_def_assigns): Likewise.
7215 (prune_bbs): Likewise.
7216 (switch_conversion::prune_bbs): Likewise.
7217 (fix_phi_nodes): Likewise.
7218 (switch_conversion::fix_phi_nodes): Likewise.
7219 (gen_inbound_check): Likewise.
7220 (switch_conversion::gen_inbound_check): Likewise.
7221 (process_switch): Use the newly created class.
7222 (switch_conversion::expand): New.
7223 (switch_conversion::~switch_conversion): New.
7224 * tree-switch-conversion.h: New file.
7226 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
7228 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
7229 tree-vect-patterns.c.
7230 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
7231 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
7232 (vect_recog_sad_pattern): Likewise.
7233 (vect_recog_widen_sum_pattern): Likewise.
7234 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
7235 (vect_recog_widen_shift_pattern): Remove the type_in argument.
7236 (vect_recog_rotate_pattern): Likewise.
7237 (vect_recog_mult_pattern): Likewise.
7238 (vect_recog_vector_vector_shift_pattern): Likewise.
7239 (vect_recog_divmod_pattern): Likewise.
7240 (vect_recog_mixed_size_cond_pattern): Likewise.
7241 (vect_recog_bool_pattern): Likewise.
7242 (vect_recog_mask_conversion_pattern): Likewise.
7243 (vect_try_gather_scatter_pattern): Likewise.
7244 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
7245 (vect_recog_over_widening_pattern): Likewise.
7246 (vect_recog_gather_scatter_pattern): Likewise.
7247 (vect_recog_func_ptr): Move from tree-vectorizer.h
7248 (vect_vect_recog_func_ptrs): Move further down the file.
7249 (vect_recog_func): Likewise. Remove the third argument.
7250 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
7251 (vect_pattern_recog_1): Expect the pattern function to do any
7252 necessary target tests. Also expect it to provide a vector type.
7253 Remove the type_in handling.
7255 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
7257 * tree-vect-patterns.c (vect_pattern_detected): New function.
7258 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
7259 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
7260 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
7261 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
7262 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
7263 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
7264 (vect_recog_mask_conversion_pattern)
7265 (vect_try_gather_scatter_pattern): Likewise.
7267 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
7269 * tree-vect-patterns.c (vect_get_internal_def): New function.
7270 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
7271 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
7272 (search_type_for_mask_1): Use it.
7274 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
7276 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
7277 redundant WIDEN_SUM_EXPR handling.
7278 (vect_recog_sad_pattern): Likewise.
7280 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
7282 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
7283 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
7284 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
7285 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
7286 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
7288 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
7290 * tree-vect-stmts.c (vectorizable_call): Make sure that we
7291 use the stmt_vec_info of the original bb statement for the
7292 new zero assignment, even if the call is part of a pattern.
7294 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
7296 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
7297 that the sequence is attached to the original statement rather
7298 than the pattern statement.
7299 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
7300 PATTERN_DEF_SEQ from the original statement rather than
7301 the main pattern statement.
7302 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
7303 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
7304 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
7306 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
7308 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
7309 definition statements before the early exit for statements that aren't
7311 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
7313 (vect_transform_loop): ...here. Process pattern definition
7314 statements without first checking whether the main pattern
7315 statement is live or relevant.
7317 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
7319 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
7320 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
7322 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
7324 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
7325 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
7326 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
7327 (expand_block_compare): Change select_block_compare_mode call.
7328 (expand_strncmp_align_check): Use new functions, fix comment.
7329 (emit_final_str_compare_gpr): New function.
7330 (expand_strn_compare): Refactor and clean up code.
7331 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
7333 2018-06-19 Tony Reix <tony.reix@atos.com>
7334 Damien Bergamini <damien.bergamini@atos.com>
7335 David Edelsohn <dje.gcc@gmail.com>
7337 * collect2.c (static_obj): New variable.
7338 (static_libs): New variable.
7339 (is_in_list): Uncomment declaration.
7340 (main): Track AIX libraries linked statically.
7341 (is_in_list): Uncomment definition.
7342 (scan_prog_file): Don't add AIX shared libraries initializer
7343 to constructor list if linking statically.
7345 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
7347 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
7349 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
7351 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
7353 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
7356 2018-06-19 Martin Liska <mliska@suse.cz>
7358 * config/i386/i386.c (ix86_can_inline_p): Do not use
7359 ipa_fn_summaries::get_create.
7360 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
7362 (devirtualization_time_bonus): Likewise.
7363 (ipcp_propagate_stage): Likewise.
7364 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
7365 (edge_set_predicate): Likewise.
7366 (evaluate_conditions_for_known_args): Likewise.
7367 (evaluate_properties_for_edge): Likewise.
7368 (ipa_call_summary::reset): Tranform to ...
7369 (ipa_call_summary::~ipa_call_summary): ... this.
7370 (ipa_fn_summary::reset): Transform to ...
7371 (ipa_fn_summary::~ipa_fn_summary): ... this.
7372 (ipa_fn_summary_t::remove): Rename to ...
7373 (ipa_fn_summary_t::remove_callees): ... this.
7374 (ipa_fn_summary_t::duplicate): Use placement new
7375 instead of memory copy.
7376 (ipa_call_summary_t::duplicate): Likewise.
7377 (ipa_call_summary_t::remove): Remove.
7378 (dump_ipa_call_summary): Change get_create to get.
7379 (ipa_dump_fn_summary): Dump only when summary exists.
7380 (analyze_function_body): Use symbol_summary::get instead
7382 (compute_fn_summary): Likewise.
7383 (estimate_edge_devirt_benefit): Likewise.
7384 (estimate_edge_size_and_time): Likewise.
7385 (inline_update_callee_summaries): Likewise.
7386 (remap_edge_change_prob): Likewise.
7387 (remap_edge_summaries): Likewise.
7388 (ipa_merge_fn_summary_after_inlining): Likewise.
7389 (write_ipa_call_summary): Likewise.
7390 (ipa_fn_summary_write): Likewise.
7391 (ipa_free_fn_summary): Likewise.
7392 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
7393 (struct ipa_call_summary): Likewise.
7394 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
7396 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
7397 (estimate_size_after_inlining): Likewise.
7398 (estimate_growth): Likewise.
7399 (growth_likely_positive): Likewise.
7400 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
7401 (inline_call): Likewise.
7402 * ipa-inline.c (caller_growth_limits): Likewise.
7403 (can_inline_edge_p): Likewise.
7404 (can_inline_edge_by_limits_p): Likewise.
7405 (compute_uninlined_call_time): Likewise.
7406 (compute_inlined_call_time): Likewise.
7407 (want_inline_small_function_p): Likewise.
7408 (edge_badness): Likewise.
7409 (update_caller_keys): Likewise.
7410 (update_callee_keys): Likewise.
7411 (inline_small_functions): Likewise.
7412 (inline_to_all_callers_1): Likewise.
7413 (dump_overall_stats): Likewise.
7414 (early_inline_small_functions): Likewise.
7415 (early_inliner): Likewise.
7416 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
7417 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
7418 * ipa-pure-const.c (malloc_candidate_p): Likewise.
7419 * ipa-split.c (execute_split_functions): Likewise.
7420 * symbol-summary.h: Likewise.
7421 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
7423 2018-06-19 Richard Biener <rguenther@suse.de>
7425 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
7426 (vectorize_loops): ... here. Fix dbgcnt handling.
7427 (try_vectorize_loop): Wrap try_vectorize_loop_1.
7429 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
7432 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
7433 ieee128 argument takes up only one (vector) register, not two (floating
7436 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
7438 * gimplify.c (gimplify_init_constructor): Really never clear for an
7439 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
7441 2018-06-19 Richard Biener <rguenther@suse.de>
7443 PR tree-optimization/86179
7444 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
7445 after failed recognition.
7447 2018-06-18 Martin Sebor <msebor@redhat.com>
7450 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
7451 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
7452 Handle integer subtraction.
7453 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
7454 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
7456 2018-06-18 David Malcolm <dmalcolm@redhat.com>
7458 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
7459 param from rtx to rtx_insn *.
7460 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
7462 (frv_ifcvt_modify_insn): Likwise.
7463 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
7464 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
7465 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
7466 as_a <rtx_insn *> cast to local "unprotected_region" once
7467 it's been established that it's not NULL or pc_rtx.
7468 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
7469 param "sethi" from rtx to rtx_insn *.
7470 (nds32_group_float_insns): Likewise for param "insn".
7471 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
7473 (vax_output_int_subtract): Likewise.
7474 * config/vax/vax.c (vax_output_int_add): Likewise for param
7476 (vax_output_int_subtract): Likewise.
7477 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
7478 (emit_pattern_after): Likewise for param "after".
7479 (emit_insn_after): Likewise.
7480 (emit_jump_insn_after): Likewise.
7481 (emit_call_insn_after): Likewise.
7482 (emit_debug_insn_after): Likewise.
7483 (emit_pattern_before): Likewise for param "before".
7484 (emit_insn_before): Likewise.
7485 (emit_jump_insn_before): Likewise.
7486 * final.c (get_insn_template): Likewise for param "insn", removing
7488 * output.h (get_insn_template): Likewise for 2nd param.
7489 * rtl.h (emit_insn_before): Likewise.
7490 (emit_jump_insn_before): Likewise.
7491 (emit_debug_insn_before_noloc): Likewise.
7492 (emit_insn_after): Likewise.
7493 (emit_jump_insn_after): Likewise.
7494 (emit_call_insn_after): Likewise.
7495 (emit_debug_insn_after): Likewise.
7496 (set_insn_deleted): Likewise for param.
7498 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
7501 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
7502 floating point modes, so that IFmode is numerically greater than
7503 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
7504 to declare the ordering. This prevents IFmode from being
7505 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
7506 machine. Include rs6000-modes.h to share the fractional values
7507 between genmodes* and the rest of the compiler.
7511 * config/rs6000/rs6000-modes.h: New file.
7512 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
7513 meaning of rs6000_long_double_size so that 126..128 selects an
7514 appropriate 128-bit floating point type.
7515 (rs6000_option_override_internal): Likewise.
7516 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
7517 (TARGET_LONG_DOUBLE_128): Change the meaning of
7518 rs6000_long_double_size so that 126..128 selects an appropriate
7519 128-bit floating point type.
7520 (LONG_DOUBLE_TYPE_SIZE): Update comment.
7521 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
7522 source and destination to match the standard usage.
7523 (truncifkf2): Likewise.
7524 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
7525 ISA 2.07 to use an explicit clobber, instead of passing in a
7527 (copysign<mode>3_soft): Likewise.
7529 2018-06-18 David Malcolm <dmalcolm@redhat.com>
7531 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
7532 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
7533 (vect_slp_analyze_instance_dependence): Likewise.
7534 (vect_enhance_data_refs_alignment): Likewise.
7535 (vect_analyze_data_refs_alignment): Likewise.
7536 (vect_slp_analyze_and_verify_instance_alignment
7537 (vect_analyze_data_ref_accesses): Likewise.
7538 (vect_prune_runtime_alias_test_list): Likewise.
7539 (vect_analyze_data_refs): Likewise.
7540 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
7541 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
7542 (vect_analyze_scalar_cycles_1): Likewise.
7543 (vect_get_loop_niters): Likewise.
7544 (vect_analyze_loop_form_1): Likewise.
7545 (vect_update_vf_for_slp): Likewise.
7546 (vect_analyze_loop_operations): Likewise.
7547 (vect_analyze_loop): Likewise.
7548 (vectorizable_induction): Likewise.
7549 (vect_transform_loop): Likewise.
7550 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
7551 * tree-vect-slp.c (vect_analyze_slp): Likewise.
7552 (vect_make_slp_decision): Likewise.
7553 (vect_detect_hybrid_slp): Likewise.
7554 (vect_slp_analyze_operations): Likewise.
7555 (vect_slp_bb): Likewise.
7556 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
7557 (vectorizable_bswap): Likewise.
7558 (vectorizable_call): Likewise.
7559 (vectorizable_simd_clone_call): Likewise.
7560 (vectorizable_conversion): Likewise.
7561 (vectorizable_assignment): Likewise.
7562 (vectorizable_shift): Likewise.
7563 (vectorizable_operation): Likewise.
7564 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
7566 2018-06-18 Martin Sebor <msebor@redhat.com>
7568 PR tree-optimization/81384
7569 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
7570 * builtins.c (expand_builtin_strnlen): New function.
7571 (expand_builtin): Call it.
7572 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
7573 * builtins.def (BUILT_IN_STRNLEN): New.
7574 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
7575 Warn for bounds in excess of maximum object size.
7576 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
7577 single-value ranges. Handle strnlen.
7578 (handle_builtin_strlen): Handle strnlen.
7579 (strlen_check_and_optimize_stmt): Same.
7580 * doc/extend.texi (Other Builtins): Document strnlen.
7582 2018-06-18 Maya Rashish <coypu@sdf.org>
7584 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
7585 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
7586 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
7588 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
7590 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
7592 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7594 * tree.c (escaped_string::escape): Replace cast to char * by
7595 const_cast<char *> (unescaped).
7597 2018-06-18 Nick Clifton <nickc@redhat.com>
7600 * tree.c (escaped_string): New class. Converts an unescaped
7601 string into its escaped equivalent.
7602 (warn_deprecated_use): Use the new class to convert the
7603 deprecation message, if present.
7604 (test_escaped_strings): New self test.
7605 (test_c_tests): Add test_escaped_strings.
7606 * doc/extend.texi (deprecated): Add a note that the
7607 deprecation message is affected by the -fmessage-length
7608 option, and that control characters will be escaped.
7609 (#pragma GCC error): Document this pragma.
7610 (#pragma GCC warning): Likewise.
7611 * doc/invoke.texi (-fmessage-length): Document this option's
7612 effect on the #warning and #error preprocessor directives and
7613 the deprecated attribute.
7615 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
7617 * tree.c (decl_value_expr_lookup): Revert latest change.
7618 (decl_value_expr_insert): Likewise.
7620 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
7622 * gimplify.c (nonlocal_vlas): Delete.
7623 (nonlocal_vla_vars): Likewise.
7624 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
7626 (gimplify_body): Do not create and destroy nonlocal_vlas.
7627 * tree-nested.c: Include diagnostic.h.
7628 (use_pointer_in_frame): Tweak.
7629 (lookup_field_for_decl): Add assertion and declare the transformation.
7630 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
7631 internal error when the reference is in a wrong context. Do not
7632 create a debug decl by default.
7633 (note_nonlocal_block_vlas): Delete.
7634 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
7635 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
7636 create a debug decl by default.
7637 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
7638 call is in a wrong context.
7639 (fixup_vla_decls): New function.
7640 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
7641 debug variables were created.
7642 * tree.c (decl_value_expr_lookup): Add checking assertion.
7643 (decl_value_expr_insert): Likewise.
7645 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
7648 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
7649 * tree-scalar-evolution.c (interpret_expr): Likewise.
7650 (expression_expensive_p): Likewise.
7651 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
7652 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
7653 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
7654 (ssa_defined_by_minus_one_stmt_p): New.
7656 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
7659 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
7660 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
7661 * dojump.c (do_jump): Likewise.
7662 * expr.c (expand_expr_real_2): Check operand type's sign.
7663 * fold-const.c (const_unop): Handle ABSU_EXPR.
7664 (fold_abs_const): Likewise.
7665 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
7666 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
7667 (strip_sign_op_1): Likesise.
7668 * match.pd: Add new pattern to generate ABSU_EXPR.
7669 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
7670 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
7671 * tree-eh.c (operation_could_trap_helper_p): Likewise.
7672 * tree-inline.c (estimate_operator_cost): Likewise.
7673 * tree-pretty-print.c (dump_generic_node): Likewise.
7674 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
7675 * tree.def (ABSU_EXPR): New.
7677 2018-06-16 Jakub Jelinek <jakub@redhat.com>
7680 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
7681 documented as preserved for backward compatibility only.
7682 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
7684 PR rtl-optimization/86108
7685 * bb-reorder.c (create_forwarder_block): Renamed to ...
7686 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
7687 jump from new landing pad to the second part.
7688 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
7691 2018-06-15 Jakub Jelinek <jakub@redhat.com>
7694 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
7695 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
7696 Only call store_expr for halves if the mode is the same.
7699 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
7700 Fix up comment formatting.
7702 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
7704 * typed-splay-tree.h (typed_splay_tree::remove): New function.
7705 (typed_splay_tree::closure,
7706 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
7707 (typed_splay_tree::typed_splay_tree,
7708 typed_splay_tree::operator =): Declared private.
7709 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
7710 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
7711 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
7712 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
7713 typed_splay_tree::splay_tree_splay,
7714 typed_splay_tree::splay_tree_foreach_helper,
7715 typed_splay_tree::splay_tree_insert,
7716 typed_splay_tree::splay_tree_remove,
7717 typed_splay_tree::splay_tree_lookup,
7718 typed_splay_tree::splay_tree_predecessor,
7719 typed_splay_tree::splay_tree_successor,
7720 typed_splay_tree::splay_tree_min,
7721 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
7722 (typed_splay_tree::root, typed_splay_tree::comp,
7723 typed_splay_tree::delete_key,
7724 typed_splay_tree::delete_value): New data members.
7725 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
7726 typed_splay_tree::remove.
7728 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
7730 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
7731 -mginv and -mno-ginv to the assembler.
7732 * config/mips/mips.opt (-mcrc): New option.
7734 * doc/invoke.text (-mcrc): Document.
7737 2018-06-15 Nick Clifton <nickc@redhat.com>
7740 * tree.c (escaped_string): New class. Converts an unescaped
7741 string into its escaped equivalent.
7742 (warn_deprecated_use): Use the new class to convert the
7743 deprecation message, if present.
7744 (test_escaped_strings): New self test.
7745 (test_c_tests): Add test_escaped_strings.
7746 * doc/extend.texi (deprecated): Add a note that the
7747 deprecation message is affected by the -fmessage-length
7748 option, and that control characters will be escaped.
7749 (#pragma GCC error): Document this pragma.
7750 (#pragma GCC warning): Likewise.
7751 * doc/invoke.texi (-fmessage-length): Document this option's
7752 effect on the #warning and #error preprocessor directives and
7753 the deprecated attribute.
7755 2018-06-15 Richard Biener <rguenther@suse.de>
7757 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
7758 here, also noting vector size used.
7759 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
7760 size used in MSG_OPTIMIZED_LOCATIONS dump.
7761 (pass_slp_vectorize::execute): Adjust.
7763 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
7766 * config/arc/arc.c (arc_return_address_register): Fix
7769 2018-06-15 Richard Biener <rguenther@suse.de>
7772 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
7773 leave useless conversion stripping to force_gimple_operand_gsi.
7774 (gimplify_build2): Likewise.
7775 (gimplify_build1): Likewise.
7777 2018-06-15 Richard Biener <rguenther@suse.de>
7780 * tree-cfg.c (move_stmt_op): unshare invariant addresses
7781 before adjusting their block.
7783 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
7785 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
7786 multilibs for *-*-rtems*.
7787 * config/riscv/t-rtems: New file.
7789 2018-06-14 Jakub Jelinek <jakub@redhat.com>
7792 * match.pd ((A +- CST1) +- CST2): Punt if last resort
7793 unsigned_type_for returns NULL.
7796 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
7797 subregs of multi-word pseudos unless the float mode has word size.
7799 2018-06-14 Richard Biener <rguenther@suse.de>
7802 * tree-vect-generic.c (build_word_mode_vector_type): Remove
7803 duplicate and harmful type_hash_canon.
7804 * tree.c (type_hash_canon): Assert we didn't find ourselves.
7806 2018-06-14 Richard Biener <rguenther@suse.de>
7809 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
7812 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
7814 * config/rtems.h (STDINT_LONG32): Define.
7816 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
7817 Prachi Godbole <prachi.godbole@imgtec.com>
7819 * config/mips/mips-cpus.def: Define P6600.
7820 * config/mips/mips-tables.opt: Regenerate.
7821 * config/mips/mips.c (mips_ucbranch_type): New enum.
7822 (mips_rtx_cost_data): Add support for P6600.
7823 (mips_issue_rate): Likewise.
7824 (mips_multipass_dfa_lookahead): Likewise.
7825 (mips_avoid_hazard): Likewise.
7826 (mips_reorg_process_insns): Likewise.
7827 (mips_classify_branch_p6600): New function.
7828 * config/mips/mips.h (TUNE_P6600): New define.
7829 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
7830 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
7831 * config/mips/mips.md: Include p6600.md.
7832 (processor): Add p6600.
7833 * config/mips/p6600.md: New file.
7834 * doc/invoke.texi: Add p6600 to supported architectures.
7836 2018-06-13 Martin Sebor <msebor@redhat.com>
7838 PR tree-optimization/86114
7839 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
7841 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
7843 2018-06-13 Richard Biener <rguenther@suse.de>
7845 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
7846 Properly set vector type of the intermediate stmt.
7847 * tree-vect-stmts.c (vectorizable_operation): The destination
7848 var always has vectype_out type.
7850 2018-06-13 Jeff Law <law@redhat.com>
7852 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
7853 integer 0 for argument to print_rtl_with_bb.
7854 (rl78_reorg): Likewise.
7856 2018-06-13 David Malcolm <dmalcolm@redhat.com>
7858 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
7859 from rtx to rtx_insn *.
7860 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
7862 (add_sched_insns_for_speculation): Likewise for local "target",
7863 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
7864 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
7865 from rtx_insn ** to rtx_code_label **.
7866 (reorg_emit_nops): Likewise.
7867 (c6x_reorg): Likewise for local "call_labels".
7868 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
7870 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
7871 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
7872 the loops over LABEL_REFS.
7873 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
7875 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
7876 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
7877 (split_branches): Strengthen local "olabel" from rtx to
7878 rtx_insn *, adding a safe_as_a cast.
7879 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
7881 (add_insn_after): Likewise for first two params.
7882 (add_insn_before): Likewise.
7883 (remove_insn): Likewise for param.
7884 (emit_pattern_before_noloc): Likewise for second and third params.
7885 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
7886 (emit_call_insn_before_noloc): Likewise.
7887 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
7889 (emit_barrier_before): Likewise.
7890 (emit_label_before): Strengthen "label" param from "rtx" to
7891 "rtx_code_label *". Strengthen "before" param from "rtx" to
7893 (emit_insn_after_1): Strengthen "after" param from "rtx" to
7895 (emit_pattern_after_noloc): Likewise.
7896 (emit_insn_after_noloc): Likewise.
7897 (emit_jump_insn_after_noloc): Likewise.
7898 (emit_call_insn_after_noloc): Likewise.
7899 (emit_debug_insn_after_noloc): Likewise.
7900 (emit_barrier_after): Likewise.
7901 (emit_label_after): Likewise for both params.
7902 (emit_pattern_after_setloc): Likewise for "after" param. Convert
7903 "loc" param from "int" to "location_t".
7904 (emit_insn_after_setloc): Likewise.
7905 (emit_jump_insn_after_setloc): Likewise.
7906 (emit_call_insn_after_setloc): Likewise.
7907 (emit_debug_insn_after_setloc): Likewise.
7908 (emit_pattern_before_setloc): Likewise for "before" param. Convert
7909 "loc" param from "int" to "location_t".
7910 (emit_pattern_before): Convert NULL_RTX to NULL.
7911 (emit_insn_before_setloc): Convert "loc" param from "int" to
7913 (emit_jump_insn_before_setloc): Likewise.
7914 (emit_call_insn_before_setloc): Likewise.
7915 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
7916 rtx_insn *. Convert "loc" param from "int" to "location_t".
7917 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
7918 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
7919 Convert 3rd param from "int" to "location_t".
7920 (emit_barrier_before, emit_barrier_after, next_real_insn):
7921 Strengthen param from rtx to rtx_insn *.
7922 (emit_label_before): Strengthen 1st param from "rtx" to
7923 "rtx_code_label *". Strengthen 2nd param from "rtx" to
7925 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
7926 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
7927 Strengthen 2nd param from "rtx" to "rtx_insn *".
7928 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
7929 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
7930 Likewise. Convert 3rd param from "int" to "location_t".
7931 (emit_label_after): Strengthen 1st param from "rtx" to
7933 (next_real_insn, remove_insn): Strengthen param from "rtx" to
7935 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
7936 from "rtx" to "rtx_insn *".
7938 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
7940 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
7941 bodies streamed in with -Q.
7942 * dumpfile.c (dump_files): Add lto-stream-out dump file.
7943 * dumpfile.h (tree_dump_index): Add lto_stream_out.
7944 * gimple-streamer-out.c: Include gimple-pretty-print.h
7945 (output_bb): Dump stmts streamed.
7946 * lto-section-out.c: Include print-tree.h
7947 (lto_begin_section): Dump sections created.
7948 (lto_output_decl_index): Dump decl encoded.
7949 * lto-streamer-out.c: Include print-tree.h
7950 (create_output_block): Dump output block created.
7951 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
7952 (output_function): Dump function output.
7953 (output_constructor): Dump constructor streamed.
7954 (write_global_stream): Output indexes encoded.
7955 (produce_asm_for_decls): Dump streams encoded.
7956 * lto-streamer.c (streamer_dump_file): New global var.
7957 * lto-streamer.h (streamer_dump_file): Declare.
7958 * passes.c (ipa_write_summaries): Initialize streamer dump.
7959 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
7962 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
7965 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
7966 offsets for register save directives. Emit a second batch of save
7967 directives, if need be, when the function accesses prior frames.
7969 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
7971 * config/arc/fpu.md (fmasf4): Force operand to register.
7972 (fnmasf4): Likewise.
7974 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
7976 * config/arc/arc-protos.h (arc_pad_return): Remove.
7977 * config/arc/arc.c (machine_function): Remove force_short_suffix
7979 (arc_print_operand): Adjust printing of '&'.
7980 (arc_verify_short): Remove conditional printing of short suffix.
7981 (arc_final_prescan_insn): Remove reference to size_reason.
7982 (pad_return): New function.
7983 (arc_reorg): Call pad_return.
7984 (arc_pad_return): Remove.
7985 (arc_init_machine_status): Remove reference to force_short_suffix.
7986 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
7987 (attr length): When attribute iscompact is true force to 2
7988 regardless; in the case of maybe check if we want to force the
7989 instruction to have 4 bytes length.
7990 (nopv): Change it to generate 4 byte long nop as well.
7991 (blockage): New pattern.
7992 (simple_return): Remove call to arc_pad_return.
7993 (p_return_i): Likewise.
7995 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
7997 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
7999 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
8001 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
8004 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
8006 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
8007 for ARC700 and ARCv2.
8009 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
8012 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
8013 operands[2] instead of operands[1].
8016 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
8018 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
8019 case, check whether the outer register overlaps an unallocatable
8020 register, not just whether it fits the required class.
8022 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
8024 * poly-int.h (can_div_trunc_p): Add new overload in which all values
8026 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
8027 (memrefs_conflict_p): Likewise.
8028 (init_alias_analysis): Likewise.
8029 * cfgexpand.c (expand_debug_expr): Likewise.
8030 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
8031 * cse.c (fold_rtx): Likewise.
8032 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
8033 * expr.c (emit_block_move_hints): Likewise.
8034 (clear_storage_hints, push_block, emit_push_insn): Likewise.
8035 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
8036 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
8037 (emit_group_store): Likewise.
8038 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
8039 to read the PRE/POST_MODIFY increment.
8040 * calls.c (store_one_arg): Use strip_offset.
8041 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
8043 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
8045 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
8046 (simplify_binary_operation_1): Extend CONST_INT handling to
8048 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
8049 than a HOST_WIDE_INT.
8050 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
8052 (adjust_mems, add_stores): Update accodingly.
8053 (vt_canonicalize_addr): Track polynomial offsets.
8054 (emit_note_insn_var_location): Likewise.
8055 (vt_add_function_parameter): Likewise.
8056 (vt_initialize): Likewise.
8058 2018-06-12 Jeff Law <law@redhat.com>
8060 * config.gcc (alpha*-*-freebsd*): Remove.
8061 * config/alpha/freebsd.h: Remove.
8063 2018-06-12 David Malcolm <dmalcolm@redhat.com>
8066 * spellcheck-tree.c (levenshtein_distance): Rename to...
8067 (get_edit_distance): ...this, and update for underlying renaming.
8068 * spellcheck-tree.h (levenshtein_distance): Rename to...
8069 (get_edit_distance): ...this.
8070 * spellcheck.c (levenshtein_distance): Rename to...
8071 (get_edit_distance): ...this. Convert from Levenshtein distance
8072 to Damerau-Levenshtein distance by supporting transpositions of
8073 adjacent characters. Rename "v1" to "v_next" and "v0" to
8075 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
8076 (selftest::test_edit_distance_unit_test_oneway): ...this, and
8077 update for underlying renaming.
8078 (selftest::levenshtein_distance_unit_test): Rename to...
8079 (selftest::test_get_edit_distance_unit): ...this, and update for
8080 underlying renaming.
8081 (selftest::test_find_closest_string): Add example from PR 69968
8082 where transposition helps
8083 (selftest::test_metric_conditions): Update for renaming.
8084 (selftest::test_metric_conditions): Likewise.
8085 (selftest::spellcheck_c_tests): Likewise.
8086 * spellcheck.h (levenshtein_distance): Rename both overloads to...
8087 (get_edit_distance): ...this.
8088 (best_match::consider): Update for renaming.
8090 2018-06-12 Martin Sebor <msebor@redhat.com>
8092 PR tree-optimization/85259
8093 * builtins.c (compute_objsize): Handle constant offsets.
8094 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
8095 true iff a warning has been issued.
8096 * gimple.h (gimple_nonartificial_location): New function.
8097 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
8098 gimple_nonartificial_location and handle -Wno-system-headers.
8099 (handle_builtin_stxncpy): Same.
8101 2018-06-12 Martin Sebor <msebor@redhat.com>
8104 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
8106 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
8108 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8109 BUILTIN_VEC_XST entries for pointer to double and long long.
8111 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
8114 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
8116 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
8119 2018-06-12 Martin Liska <mliska@suse.cz>
8121 * doc/options.texi: Document IntegerRange.
8123 2018-06-12 Martin Liska <mliska@suse.cz>
8125 * config/i386/i386.opt: Make MPX-related options as Deprecated.
8126 * opt-functions.awk: Handle Deprecated flag.
8127 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
8129 (read_cmdline_option): Report warning for a deprecated option.
8130 * opts.h (struct cl_option): Add new field cl_deprecated.
8131 (CL_ERR_DEPRECATED): New.
8133 2018-06-12 Martin Liska <mliska@suse.cz>
8135 * doc/options.texi: Document Deprecated option flag.
8137 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
8139 * config/arc/arc-arch.h (arc_extras): New enum.
8140 (arc_cpu_t):Add field extra.
8141 (arc_cpu_types): Consider the extras.
8142 * config/arc/arc-cpus.def: Add extras info.
8143 * config/arc/arc-opts.h (processor_type): Consider extra field.
8144 * config/arc/arc.c (arc_override_options): Handle extra field.
8146 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
8148 * config/arc/arc-arch.h: Update ARC_OPTX macro.
8149 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
8151 * config/arc/arc.c (arc_init): Update pic warning.
8152 (irq_range): Update irq range parsing warnings.
8153 (arc_override_options): Update various warning messages.
8154 (arc_handle_aux_attribute): Likewise.
8156 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
8158 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
8160 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
8162 * doc/sourcebuild.texi: Document usage of line number 0 in verify
8163 compiler messages directives.
8165 2018-06-12 Matthew Fortune <mfortune@gmail.com>
8167 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
8168 * config/mips/mips-tables.opt: Regenerate.
8169 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
8171 * doc/invoke.texi: Document -march=i6500.
8173 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
8175 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
8176 (i6400_gpmul): Add cpu_unit.
8177 (i6400_gpdiv): Likewise.
8178 (i6400_msa_add_d): Update reservations.
8179 (i6400_msa_int_add) Likewise.
8180 (i6400_msa_short_logic3) Likewise.
8181 (i6400_msa_short_logic2) Likewise.
8182 (i6400_msa_short_logic) Likewise.
8183 (i6400_msa_move) Likewise.
8184 (i6400_msa_cmp) Likewise.
8185 (i6400_msa_short_float2) Likewise.
8186 (i6400_msa_div_d) Likewise.
8187 (i6400_msa_long_logic1) Likewise.
8188 (i6400_msa_long_logic2) Likewise.
8189 (i6400_msa_mult) Likewise.
8190 (i6400_msa_long_float2) Likewise.
8191 (i6400_msa_long_float4) Likewise.
8192 (i6400_msa_long_float5) Likewise.
8193 (i6400_msa_long_float8) Likewise.
8194 (i6400_fpu_fadd): Include frint type.
8195 (i6400_fpu_store): New define_insn_reservation.
8196 (i6400_fpu_load): Likewise.
8197 (i6400_fpu_move): Likewise.
8198 (i6400_fpu_fcmp): Likewise.
8199 (i6400_fpu_fmadd): Likewise.
8200 (i6400_int_mult): Include imul3nc type and update reservation.
8201 (i6400_int_div): Include idiv3 type and update reservation.
8202 (i6400_int_load): Update to check type not move_type.
8203 (i6400_int_store): Likewise.
8204 (i6400_int_prefetch): Set zero latency.
8206 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
8208 * gcc.c: Document new %@{...} sequence.
8209 (LINK_COMMAND_SPEC): Use it for the -L switches.
8210 (cpp_unique_options): Use it for the -I switches.
8211 (at_file_argbuf): New global variable.
8212 (in_at_file): Likewise.
8213 (alloc_args): Create at_file_argbuf.
8214 (clear_args): Truncate at_file_argbuf.
8215 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
8216 (open_at_file): New function.
8217 (close_at_file): Likewise.
8218 (create_at_file): Delete.
8219 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
8222 (validate_switches_from_spec): Deal with %@{...} sequence.
8223 (validate_switches): Likewise.
8224 (driver::finalize): Call clear_args.
8226 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
8228 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
8230 2018-06-11 Martin Sebor <msebor@redhat.com>
8232 * doc/invoke.texi (-Wall): List -Wc++17-compat.
8233 (Wno-class-memaccess): Add @opindex.
8234 (Wno-templates, Wno-multiple-inheritance): Same.
8235 (Wno-virtual-inheritance, Wno-namespaces): Same.
8236 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
8237 (Wno-format-overflow, Wno-format-truncation): Same.
8238 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
8239 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
8240 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
8241 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
8242 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
8243 (Wno-misspelled-isr): Same.
8245 2018-06-11 Martin Sebor <msebor@redhat.com>
8247 * PR tree-optimization/86083
8248 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
8250 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
8252 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
8254 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
8257 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
8258 on the correct operand.
8259 (*movdi_internal64): Ditto.
8261 2018-06-11 Martin Liska <mliska@suse.cz>
8263 PR tree-optimization/86089
8264 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
8266 2018-06-11 Julia Koval <julia.koval@intel.com>
8268 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
8269 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
8270 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
8272 2018-06-11 Olivier Hainque <hainque@adacore.com>
8274 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
8275 for Ada with strict dwarf2.
8277 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
8280 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
8283 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
8285 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
8287 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
8289 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
8292 2018-06-08 Martin Liska <mliska@suse.cz>
8294 * tree-cfg.h (debug_function): Fix argument type to match
8297 2018-06-08 Martin Liska <mliska@suse.cz>
8299 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
8300 Remove usage of MPX-related (and removed) fields.
8301 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
8303 2018-06-08 David Malcolm <dmalcolm@redhat.com>
8305 * cfg.c (debug): Use TDF_NONE rather than 0.
8306 * cfghooks.c (debug): Likewise.
8307 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
8308 (struct dump_option_value_info): Convert to...
8309 (struct kv_pair): ...this template type.
8310 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
8312 (optinfo_verbosity_options): Likewise.
8313 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
8315 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
8316 than int for "optgroup_flags" param.
8317 (dump_generic_expr_loc): Use dump_flags_t rather than int for
8319 (dump_dec): Likewise.
8320 (dump_finish): Use TDF_NONE rather than 0.
8321 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
8322 rather than int for "optgroup_flags" param. Use TDF_NONE rather
8323 than 0. Update for change to option_ptr.
8324 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
8325 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
8326 0. Update for changes to optinfo_verbosity_options and
8328 (opt_info_switch_p): Convert optgroup_flags from int to
8330 (dump_basic_block): Use dump_flags_t rather than int
8331 for "dump_kind" param.
8332 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
8333 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
8334 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
8335 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
8336 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
8337 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
8338 TDF_NONE): Convert from macros to...
8339 (enum dump_flag): ...this new enum.
8340 (dump_flags_t): Update to use enum.
8341 (operator|, operator&, operator~, operator|=, operator&=):
8342 Implement for dump_flags_t.
8343 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
8344 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
8345 Convert from macros to...
8346 (enum optgroup_flag): ...this new enum.
8347 (optgroup_flags_t): New typedef.
8348 (operator|, operator|=): Implement for optgroup_flags_t.
8349 (struct dump_file_info): Convert field "alt_flags" to
8350 dump_flags_t. Convert field "optgroup_flags" to
8352 (dump_basic_block): Use dump_flags_t rather than int for param.
8353 (dump_generic_expr_loc): Likewise.
8354 (dump_dec): Likewise.
8355 (dump_register): Convert param "optgroup_flags" to
8357 (opt_info_enable_passes): Likewise.
8358 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
8360 * gimple-pretty-print.c (debug): Likewise.
8361 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
8362 (merged_store_group::apply_stores): Likewise.
8363 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
8364 * gimple.c (verify_gimple_pp): Likewise.
8365 * graphite-poly.c (print_pbb_body): Likewise.
8366 * passes.c (pass_manager::register_one_dump_file): Convert
8367 local "optgroup_flags" to optgroup_flags_t.
8368 * print-tree.c (print_node): Use TDF_NONE rather than 0.
8370 (debug_body): Likewise.
8371 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
8372 to optgroup_flags_t.
8373 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
8375 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
8376 (convert_mult_to_fma): Likewise.
8377 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
8378 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
8379 * tree-vect-data-refs.c (dump_lower_bound): Convert param
8380 "dump_kind" to dump_flags_t.
8382 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
8384 * config/rs6000/rs6000.c (min, max): Delete.
8386 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
8388 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
8389 -mabi=spe and -mabi=no-spe.
8391 2018-06-08 Martin Liska <mliska@suse.cz>
8393 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
8394 where we expect an existing summary.
8396 2018-06-08 Martin Liska <mliska@suse.cz>
8398 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
8399 * ipa-inline.h (estimate_edge_growth): Likewise.
8401 2018-06-08 Martin Liska <mliska@suse.cz>
8403 * cgraph.c (function_version_hasher::hash): Use
8404 cgraph_node::get_uid ().
8405 (function_version_hasher::equal):
8406 * cgraph.h (cgraph_node::get_uid): New method.
8407 * ipa-inline.c (update_caller_keys): Use
8408 cgraph_node::get_uid ().
8409 (update_callee_keys): Likewise.
8410 * ipa-utils.c (searchc): Likewise.
8411 (ipa_reduced_postorder): Likewise.
8412 * lto-cgraph.c (input_node): Likewise.
8413 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
8414 * symbol-summary.h (symtab_insertion): Likewise.
8415 (symtab_removal): Likewise.
8416 (symtab_duplication): Likewise.
8417 * tree-pretty-print.c (dump_function_header): Likewise.
8418 * tree-sra.c (convert_callers_for_node): Likewise.
8420 2018-06-08 Martin Liska <mliska@suse.cz>
8422 * cgraph.c (symbol_table::create_edge): Always assign a new
8424 (symbol_table::free_edge): Do not recycle numbers.
8425 * cgraph.h (cgraph_edge::get): New method.
8426 * symbol-summary.h (symtab_removal): Use it.
8427 (symtab_duplication): Likewise.
8428 (call_summary::hashable_uid): Remove.
8430 2018-06-08 Martin Liska <mliska@suse.cz>
8432 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
8433 (initialize_growth_caches): Remove.
8434 (free_growth_caches): Likewise.
8435 (do_estimate_edge_time): Use edge_growth_cache.
8436 (do_estimate_edge_size): Likewise.
8437 (do_estimate_edge_hints): Likewise.
8438 * ipa-inline.c (reset_edge_caches): Likewise.
8439 (recursive_inlining): Likewise.
8440 (inline_small_functions): Likewise.
8441 * ipa-inline.h (initialize_growth_caches): Remove.
8442 (estimate_edge_size): Likewise.
8443 (estimate_edge_time): Likewise.
8444 (estimate_edge_hints): Likewise.
8445 (reset_edge_growth_cache): Likewise.
8446 * symbol-summary.h (call_summary::remove): New method.
8448 2018-06-08 Martin Liska <mliska@suse.cz>
8450 * ipa-cp.c (class edge_clone_summary): New summary.
8451 (grow_edge_clone_vectors): Remove.
8452 (ipcp_edge_duplication_hook): Remove.
8453 (class edge_clone_summary_t): New call_summary class.
8454 (ipcp_edge_removal_hook): Remove.
8455 (edge_clone_summary_t::duplicate): New function.
8456 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
8457 (create_specialized_node): Likewise.
8458 (ipcp_driver): Initialize edge_clone_summaries and do not
8461 2018-06-08 Martin Liska <mliska@suse.cz>
8463 * symbol-summary.h (get): New function.
8464 (call_summary::m_initialize_when_cloning): New class member.
8466 2018-06-08 Martin Liska <mliska@suse.cz>
8468 * cgraph.c (cgraph_node::remove): Do not recycle uid.
8469 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
8470 (symbol_table::allocate_cgraph_symbol): Do not set uid.
8471 * passes.c (uid_hash_t): Record removed_nodes by their uids.
8472 (remove_cgraph_node_from_order): Use the removed_nodes set.
8473 (do_per_function_toporder): Likwise.
8474 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
8475 instead of summary_uid.
8476 (symtab_removal): Likewise.
8477 (symtab_duplication): Likewise.
8479 2018-06-08 Martin Liska <mliska@suse.cz>
8481 * ipa-cp.c (ipcp_store_bits_results): Use
8482 ipcp_transformation_sum.
8483 (ipcp_store_vr_results): Likewise.
8484 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
8486 (ipcp_transformation_initialize): ... this.
8487 (ipa_set_node_agg_value_chain):
8488 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
8489 (write_ipcp_transformation_info): Likewise.
8490 (read_ipcp_transformation_info): Likewise.
8491 (ipcp_update_bits): Likewise.
8492 (ipcp_update_vr): Likewise.
8493 (ipcp_transform_function): Likewise.
8494 * ipa-prop.h: Rename ipcp_transformation_summary to
8495 ipcp_transformation.
8496 (class ipcp_transformation_t): New function summary.
8497 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
8498 (ipa_get_agg_replacements_for_node): Likewise.
8500 2018-06-08 Martin Liska <mliska@suse.cz>
8502 * ipa-pure-const.c (struct funct_state_d): Do it class instead
8504 (class funct_state_summary_t): New function_summary class.
8505 (has_function_state): Remove.
8506 (get_function_state): Likewise.
8507 (set_function_state): Likewise.
8508 (add_new_function): Likewise.
8509 (funct_state_summary_t::insert): New function.
8510 (duplicate_node_data): Remove.
8511 (remove_node_data): Remove.
8512 (funct_state_summary_t::duplicate): New function.
8513 (register_hooks): Create new funct_state_summaries.
8514 (pure_const_generate_summary): Use it.
8515 (pure_const_write_summary): Likewise.
8516 (pure_const_read_summary): Likewise.
8517 (propagate_pure_const): Likewise.
8518 (propagate_nothrow): Likewise.
8519 (dump_malloc_lattice): Likewise.
8520 (propagate_malloc): Likewise.
8521 (execute): Do not register hooks, just remove summary
8523 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
8526 2018-06-08 Martin Liska <mliska@suse.cz>
8528 * ipa-reference.c (remove_node_data): Remove.
8529 (duplicate_node_data): Likewise.
8530 (class ipa_ref_var_info_summary_t): New class.
8531 (class ipa_ref_opt_summary_t): Likewise.
8532 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
8533 (get_reference_optimization_summary): Use
8534 ipa_ref_opt_sum_summaries.
8535 (set_reference_vars_info): Remove.
8536 (set_reference_optimization_summary): Likewise.
8537 (ipa_init): Create summaries.
8538 (init_function_info): Use function summary.
8539 (ipa_ref_opt_summary_t::duplicate): New function.
8540 (ipa_ref_opt_summary_t::remove): New function.
8541 (get_read_write_all_from_node): Fix GNU coding style.
8542 (propagate): Use function summary.
8543 (write_node_summary_p): Fix GNU coding style.
8544 (stream_out_bitmap): Likewise.
8545 (ipa_reference_read_optimization_summary): Use function summary.
8546 (ipa_reference_c_finalize): Do not release hooks.
8548 2018-06-08 Martin Liska <mliska@suse.cz>
8550 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
8551 (analyze_function_body): Extract multiple calls of get_create.
8552 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
8553 * ipa-inline.c (recursive_inlining): Use ::get method.
8554 * ipa-inline.h (estimate_edge_growth): Likewise.
8556 2018-06-08 Martin Liska <mliska@suse.cz>
8558 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
8560 (hsa_function_summary::hsa_function_summary): Use the new enum
8562 (hsa_gpu_implementation_p): Use hsa_summaries::get.
8563 * hsa-gen.c (hsa_get_host_function): Likewise.
8564 (get_brig_function_name): Likewise.
8565 * ipa-hsa.c (process_hsa_functions): Likewise.
8566 (ipa_hsa_write_summary): Likewise.
8567 * symbol-summary.h (symtab_duplication): Use ::get function/
8568 (get): New function.
8570 2018-06-08 Martin Liska <mliska@suse.cz>
8572 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
8574 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
8575 (hsa_register_kernel): Likewise.
8576 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
8577 * hsa-gen.c (hsa_get_host_function): Likewise.
8578 (get_brig_function_name): Likewise.
8579 (generate_hsa): Likewise.
8580 (pass_gen_hsail::execute): Likewise.
8581 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
8582 (devirtualization_time_bonus): Likewise.
8583 (ipcp_propagate_stage): Likewise.
8584 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
8585 (edge_set_predicate): Likewise.
8586 (evaluate_conditions_for_known_args): Likewise.
8587 (evaluate_properties_for_edge): Likewise.
8588 (ipa_fn_summary::reset): Likewise.
8589 (ipa_fn_summary_t::duplicate): Likewise.
8590 (dump_ipa_call_summary): Likewise.
8591 (ipa_dump_fn_summary): Likewise.
8592 (analyze_function_body): Likewise.
8593 (compute_fn_summary): Likewise.
8594 (estimate_edge_devirt_benefit): Likewise.
8595 (estimate_edge_size_and_time): Likewise.
8596 (estimate_calls_size_and_time): Likewise.
8597 (estimate_node_size_and_time): Likewise.
8598 (inline_update_callee_summaries): Likewise.
8599 (remap_edge_change_prob): Likewise.
8600 (remap_edge_summaries): Likewise.
8601 (ipa_merge_fn_summary_after_inlining): Likewise.
8602 (ipa_update_overall_fn_summary): Likewise.
8603 (read_ipa_call_summary): Likewise.
8604 (inline_read_section): Likewise.
8605 (write_ipa_call_summary): Likewise.
8606 (ipa_fn_summary_write): Likewise.
8607 (ipa_free_fn_summary): Likewise.
8608 * ipa-hsa.c (process_hsa_functions): Likewise.
8609 (ipa_hsa_write_summary): Likewise.
8610 (ipa_hsa_read_section): Likewise.
8611 * ipa-icf.c (sem_function::merge): Likewise.
8612 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
8613 (do_estimate_edge_time): Likewise.
8614 (estimate_size_after_inlining): Likewise.
8615 (estimate_growth): Likewise.
8616 (growth_likely_positive): Likewise.
8617 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
8618 (inline_call): Likewise.
8619 * ipa-inline.c (caller_growth_limits): Likewise.
8620 (can_inline_edge_p): Likewise.
8621 (can_inline_edge_by_limits_p): Likewise.
8622 (compute_uninlined_call_time): Likewise.
8623 (compute_inlined_call_time): Likewise.
8624 (want_inline_small_function_p): Likewise.
8625 (edge_badness): Likewise.
8626 (update_caller_keys): Likewise.
8627 (update_callee_keys): Likewise.
8628 (recursive_inlining): Likewise.
8629 (inline_small_functions): Likewise.
8630 (inline_to_all_callers_1): Likewise.
8631 (dump_overall_stats): Likewise.
8632 (early_inline_small_functions): Likewise.
8633 (early_inliner): Likewise.
8634 * ipa-inline.h (estimate_edge_growth): Likewise.
8635 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
8636 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
8637 * ipa-prop.h (IPA_NODE_REF): Likewise.
8638 (IPA_EDGE_REF): Likewise.
8639 * ipa-pure-const.c (malloc_candidate_p): Likewise.
8640 (propagate_malloc): Likewise.
8641 * ipa-split.c (execute_split_functions): Likewise.
8642 * symbol-summary.h: Rename get to get_create.
8644 (get_create): Likewise.
8645 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
8647 2018-06-08 Martin Liska <mliska@suse.cz>
8649 * symbol-summary.h (release): Move definition out of class
8651 (symtab_removal): Likewise.
8652 (symtab_duplication): Likewise.
8654 2018-06-08 Martin Liska <mliska@suse.cz>
8656 * symbol-summary.h (function_summary): Move constructor
8657 implementation out of class declaration.
8658 (release): Likewise.
8659 (symtab_insertion): Likewise.
8660 (symtab_removal): Likewise.
8661 (symtab_duplication): Likewise.
8664 2018-06-08 Martin Liska <mliska@suse.cz>
8666 * Makefile.in: Remove support for MPX (macros, related functions,
8667 fields in cgraph_node, ...).
8668 * builtin-types.def (BT_BND): Likewise.
8669 (BT_FN_BND_CONST_PTR): Likewise.
8670 (BT_FN_CONST_PTR_BND): Likewise.
8671 (BT_FN_VOID_PTR_BND): Likewise.
8672 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
8673 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
8674 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
8675 (expand_builtin_mempcpy_with_bounds): Likewise.
8676 (expand_builtin_memset_with_bounds): Likewise.
8677 (expand_builtin_memset_args): Likewise.
8678 (std_expand_builtin_va_start): Likewise.
8679 (expand_builtin): Likewise.
8680 (expand_builtin_with_bounds): Likewise.
8681 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
8682 (DEF_LIB_BUILTIN_CHKP): Likewise.
8683 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
8684 (DEF_CHKP_BUILTIN): Likewise.
8685 (BUILT_IN_MEMCPY): Likewise.
8686 (BUILT_IN_MEMMOVE): Likewise.
8687 (BUILT_IN_MEMPCPY): Likewise.
8688 (BUILT_IN_MEMSET): Likewise.
8689 (BUILT_IN_STPCPY): Likewise.
8690 (BUILT_IN_STRCAT): Likewise.
8691 (BUILT_IN_STRCHR): Likewise.
8692 (BUILT_IN_STRCPY): Likewise.
8693 (BUILT_IN_STRLEN): Likewise.
8694 (BUILT_IN_MEMCPY_CHK): Likewise.
8695 (BUILT_IN_MEMMOVE_CHK): Likewise.
8696 (BUILT_IN_MEMPCPY_CHK): Likewise.
8697 (BUILT_IN_MEMSET_CHK): Likewise.
8698 (BUILT_IN_STPCPY_CHK): Likewise.
8699 (BUILT_IN_STRCAT_CHK): Likewise.
8700 (BUILT_IN_STRCPY_CHK): Likewise.
8701 * calls.c (store_bounds): Likewise.
8702 (emit_call_1): Likewise.
8703 (special_function_p): Likewise.
8704 (maybe_warn_nonstring_arg): Likewise.
8705 (initialize_argument_information): Likewise.
8706 (finalize_must_preallocate): Likewise.
8707 (compute_argument_addresses): Likewise.
8708 (expand_call): Likewise.
8709 * cfgexpand.c (expand_call_stmt): Likewise.
8710 (expand_return): Likewise.
8711 (expand_gimple_stmt_1): Likewise.
8712 (pass_expand::execute): Likewise.
8713 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
8714 (cgraph_node::remove): Likewise.
8715 (cgraph_node::dump): Likewise.
8716 (cgraph_node::verify_node): Likewise.
8717 * cgraph.h (chkp_function_instrumented_p): Likewise.
8718 (symtab_node::get_alias_target): Likewise.
8719 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
8720 (cgraph_local_p): Likewise.
8721 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
8722 (cgraph_edge::rebuild_references): Likewise.
8723 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
8724 (walk_polymorphic_call_targets): Likewise.
8725 (cgraph_node::expand_thunk): Likewise.
8726 (symbol_table::output_weakrefs): Likewise.
8727 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
8728 (ix86_handle_option): Likewise.
8729 * config/i386/constraints.md: Likewise.
8730 * config/i386/i386-builtin-types.def (BND): Likewise.
8734 * config/i386/i386-builtin.def (BDESC_END): Likewise.
8735 (BDESC_FIRST): Likewise.
8737 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
8738 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
8739 * config/i386/i386.c (enum reg_class): Likewise.
8740 (ix86_target_string): Likewise.
8741 (ix86_option_override_internal): Likewise.
8742 (ix86_conditional_register_usage): Likewise.
8743 (ix86_valid_target_attribute_inner_p): Likewise.
8744 (ix86_set_indirect_branch_type): Likewise.
8745 (ix86_set_current_function): Likewise.
8746 (ix86_function_arg_regno_p): Likewise.
8747 (init_cumulative_args): Likewise.
8748 (ix86_function_arg_advance): Likewise.
8749 (ix86_function_arg): Likewise.
8750 (ix86_pass_by_reference): Likewise.
8751 (ix86_function_value_regno_p): Likewise.
8752 (ix86_function_value_1): Likewise.
8753 (ix86_function_value_bounds): Likewise.
8754 (ix86_return_in_memory): Likewise.
8755 (ix86_setup_incoming_vararg_bounds): Likewise.
8756 (ix86_va_start): Likewise.
8757 (indirect_thunk_need_prefix): Likewise.
8758 (print_reg): Likewise.
8759 (ix86_print_operand): Likewise.
8760 (ix86_expand_call): Likewise.
8761 (ix86_output_function_return): Likewise.
8762 (reg_encoded_number): Likewise.
8763 (BDESC_VERIFYS): Likewise.
8764 (ix86_init_mpx_builtins): Likewise.
8765 (ix86_init_builtins): Likewise.
8766 (ix86_emit_cmove): Likewise.
8767 (ix86_emit_move_max): Likewise.
8768 (ix86_expand_builtin): Likewise.
8769 (ix86_builtin_mpx_function): Likewise.
8770 (ix86_get_arg_address_for_bt): Likewise.
8771 (ix86_load_bounds): Likewise.
8772 (ix86_store_bounds): Likewise.
8773 (ix86_load_returned_bounds): Likewise.
8774 (ix86_store_returned_bounds): Likewise.
8775 (ix86_class_likely_spilled_p): Likewise.
8776 (ix86_hard_regno_mode_ok): Likewise.
8777 (x86_order_regs_for_local_alloc): Likewise.
8778 (ix86_mitigate_rop): Likewise.
8779 (ix86_bnd_prefixed_insn_p): Likewise.
8780 (ix86_mpx_bound_mode): Likewise.
8781 (ix86_make_bounds_constant): Likewise.
8782 (ix86_initialize_bounds): Likewise.
8783 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
8784 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
8785 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
8786 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
8787 (TARGET_CHKP_BOUND_MODE): Likewise.
8788 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
8789 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
8790 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
8791 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
8792 * config/i386/i386.h (TARGET_MPX): Likewise.
8793 (TARGET_MPX_P): Likewise.
8794 (VALID_BND_REG_MODE): Likewise.
8795 (FIRST_BND_REG): Likewise.
8796 (LAST_BND_REG): Likewise.
8797 (enum reg_class): Likewise.
8798 (BND_REG_P): Likewise.
8799 (BND_REGNO_P): Likewise.
8800 (BNDmode): Likewise.
8801 (ADJUST_INSN_LENGTH): Likewise.
8802 * config/i386/i386.md: Likewise.
8803 * config/i386/i386.opt: Likewise.
8804 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
8805 (defined): Likewise.
8806 (LINK_MPX): Likewise.
8807 (MPX_SPEC): Likewise.
8808 (LIBMPX_SPEC): Likewise.
8809 (LIBMPXWRAPPERS_SPEC): Likewise.
8810 (CHKP_SPEC): Likewise.
8811 * config/i386/predicates.md: Likewise.
8812 * dbxout.c (dbxout_type): Likewise.
8813 * doc/extend.texi: Likewise.
8814 * doc/invoke.texi: Likewise.
8815 * doc/md.texi: Likewise.
8816 * doc/tm.texi: Likewise.
8817 * doc/tm.texi.in: Likewise.
8818 * dwarf2out.c (is_base_type): Likewise.
8819 (gen_formal_types_die): Likewise.
8820 (gen_subprogram_die): Likewise.
8821 (gen_type_die_with_usage): Likewise.
8822 (gen_decl_die): Likewise.
8823 (dwarf2out_late_global_decl): Likewise.
8824 * expr.c (expand_assignment): Likewise.
8825 (emit_storent_insn): Likewise.
8826 (store_expr_with_bounds): Likewise.
8827 (store_expr): Likewise.
8828 (expand_expr_real_1): Likewise.
8829 * expr.h (store_expr_with_bounds): Likewise.
8830 * function.c (use_register_for_decl): Likewise.
8831 (struct bounds_parm_data): Likewise.
8832 (assign_parms_augmented_arg_list): Likewise.
8833 (assign_parm_find_entry_rtl): Likewise.
8834 (assign_parm_is_stack_parm): Likewise.
8835 (assign_parm_load_bounds): Likewise.
8836 (assign_bounds): Likewise.
8837 (assign_parms): Likewise.
8838 (expand_function_start): Likewise.
8839 * gcc.c (CHKP_SPEC): Likewise.
8840 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
8841 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
8842 (wrestrict_dom_walker::check_call): Likewise.
8843 * gimple.c (gimple_build_call_from_tree): Likewise.
8844 * gimple.h (enum gf_mask): Likewise.
8845 (gimple_call_with_bounds_p): Likewise.
8846 (gimple_call_set_with_bounds): Likewise.
8847 * gimplify.c (gimplify_init_constructor): Likewise.
8848 * ipa-cp.c (initialize_node_lattices): Likewise.
8849 (propagate_constants_across_call): Likewise.
8850 (find_more_scalar_values_for_callers_subset): Likewise.
8851 * ipa-hsa.c (process_hsa_functions): Likewise.
8852 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
8853 * ipa-icf.c (sem_function::merge): Likewise.
8854 * ipa-inline.c (early_inliner): Likewise.
8855 * ipa-pure-const.c (warn_function_noreturn): Likewise.
8856 (warn_function_cold): Likewise.
8857 (propagate_pure_const): Likewise.
8858 * ipa-ref.h (enum GTY): Likewise.
8859 * ipa-split.c (find_retbnd): Likewise.
8860 (consider_split): Likewise.
8861 (split_function): Likewise.
8862 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
8863 * ipa.c (walk_polymorphic_call_targets): Likewise.
8864 (symbol_table::remove_unreachable_nodes): Likewise.
8865 (process_references): Likewise.
8866 (cgraph_build_static_cdtor_1): Likewise.
8867 * lto-cgraph.c (lto_output_node): Likewise.
8868 (output_refs): Likewise.
8869 (compute_ltrans_boundary): Likewise.
8870 (input_overwrite_node): Likewise.
8871 (input_node): Likewise.
8872 (input_cgraph_1): Likewise.
8873 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
8874 * passes.c (pass_manager::execute_early_local_passes): Likewise.
8875 (class pass_chkp_instrumentation_passes): Likewise.
8876 (make_pass_chkp_instrumentation_passes): Likewise.
8877 * passes.def: Likewise.
8878 * rtl.h (struct GTY): Likewise.
8879 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
8880 * stor-layout.c (layout_type): Likewise.
8881 * symtab.c: Likewise.
8882 * target.def: Likewise.
8883 * targhooks.c (default_chkp_bound_type): Likewise.
8884 (default_chkp_bound_mode): Likewise.
8885 (default_builtin_chkp_function): Likewise.
8886 (default_chkp_function_value_bounds): Likewise.
8887 (default_chkp_make_bounds_constant): Likewise.
8888 (default_chkp_initialize_bounds): Likewise.
8889 * targhooks.h (default_chkp_bound_type): Likewise.
8890 (default_chkp_bound_mode): Likewise.
8891 (default_builtin_chkp_function): Likewise.
8892 (default_chkp_function_value_bounds): Likewise.
8893 (default_chkp_make_bounds_constant): Likewise.
8894 (default_chkp_initialize_bounds): Likewise.
8895 * toplev.c (compile_file): Likewise.
8896 (process_options): Likewise.
8897 * tree-core.h (DEF_BUILTIN): Likewise.
8898 (DEF_BUILTIN_CHKP): Likewise.
8899 * tree-inline.c (declare_return_variable): Likewise.
8900 (remap_gimple_stmt): Likewise.
8901 (copy_bb): Likewise.
8902 (initialize_inlined_parameters): Likewise.
8903 (expand_call_inline): Likewise.
8904 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
8905 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
8906 (make_pass_ipa_chkp_produce_thunks): Likewise.
8907 (make_pass_chkp): Likewise.
8908 (make_pass_chkp_opt): Likewise.
8909 (make_pass_chkp_instrumentation_passes): Likewise.
8910 * tree-pretty-print.c (dump_generic_node): Likewise.
8911 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
8912 * tree-ssa-dce.c (propagate_necessity): Likewise.
8913 (eliminate_unnecessary_stmts): Likewise.
8914 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
8915 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
8916 * tree-ssa-sccvn.h: Likewise.
8917 * tree-ssa-strlen.c (get_string_length): Likewise.
8918 (valid_builtin_call): Likewise.
8919 (adjust_last_stmt): Likewise.
8920 (handle_builtin_strchr): Likewise.
8921 (handle_builtin_strcpy): Likewise.
8922 (handle_builtin_stxncpy): Likewise.
8923 (handle_builtin_memcpy): Likewise.
8924 (handle_builtin_strcat): Likewise.
8925 (strlen_check_and_optimize_stmt): Likewise.
8926 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
8927 * tree-streamer-in.c: Likewise.
8928 * tree-streamer.c (record_common_node): Likewise.
8929 * tree.c (tree_code_size): Likewise.
8930 (wide_int_to_tree_1): Likewise.
8931 (type_contains_placeholder_1): Likewise.
8932 (build_common_tree_nodes): Likewise.
8933 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
8934 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
8935 (POINTER_BOUNDS_P): Likewise.
8936 (BOUNDED_TYPE_P): Likewise.
8937 (BOUNDED_P): Likewise.
8938 (CALL_WITH_BOUNDS_P): Likewise.
8939 (pointer_bounds_type_node): Likewise.
8940 * value-prof.c (gimple_ic): Likewise.
8941 * var-tracking.c (vt_add_function_parameters): Likewise.
8942 * varasm.c (make_decl_rtl): Likewise.
8943 (assemble_start_function): Likewise.
8944 (output_constant): Likewise.
8945 (maybe_assemble_visibility): Likewise.
8946 * varpool.c (ctor_for_folding): Likewise.
8947 * chkp-builtins.def: Remove.
8948 * ipa-chkp.c: Remove.
8949 * ipa-chkp.h: Remove.
8950 * rtl-chkp.c: Remove.
8951 * rtl-chkp.h: Remove.
8952 * tree-chkp-opt.c: Remove.
8953 * tree-chkp.c: Remove.
8954 * tree-chkp.h: Remove.
8956 2018-06-07 Carl Love <cel@us.ibm.com>
8958 * config/rs6000/vsx.md (vextract_fp_from_shorth,
8959 vextract_fp_from_shortl): Add BE support.
8961 2018-06-07 Paul Koning <ni1d@arrl.net>
8963 * compare-elim.c (try_merge_compare): Don't merge compare if
8964 address contains a side effect.
8965 (try_eliminate_compare): Likewise.
8967 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
8969 * config.gcc: Support "tremont".
8970 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
8971 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
8973 * config/i386/i386.c (m_TREMONT): Define.
8974 (processor_target_table): Add "tremont".
8975 (PTA_TREMONT): Define.
8976 (ix86_lea_outperforms): Add TARGET_TREMONT.
8977 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
8978 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
8979 and M_INTEL_GOLDMONT_PLUS.
8980 (fold_builtin_cpu): Add "tremont".
8981 (ix86_add_stmt_cost): Add TARGET_TREMONT.
8982 (ix86_option_override_internal): Add "tremont".
8983 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
8984 (processor_type): Add PROCESSOR_TREMONT.
8985 * config/i386/x86-tune.def: Add m_TREMONT.
8986 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
8988 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
8990 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
8991 symbol defined for msp430i* devices to be lower case.
8993 2018-06-07 Richard Biener <rguenther@suse.de>
8995 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
8996 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
8997 Properly wrap signed arithmetic if overflow wraps.
8999 2018-06-07 Jakub Jelinek <jakub@redhat.com>
9001 PR tree-optimization/69615
9002 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
9003 of a cast from a same precision integral SSA_NAME in a bb dominated
9004 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
9005 cast to utype if rhs2 has already a compatible type.
9007 2018-06-07 Richard Biener <rguenther@suse.de>
9009 PR tree-optimization/85935
9010 * graphite-scop-detection.c (find_params_in_bb): Analyze
9011 condition operands with respect to the correct loop. Assert
9012 the analysis doesn't fail.
9014 2018-06-04 Carl Love <cel@us.ibm.com>
9016 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
9017 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
9018 as it is slightly cheaper.
9019 (first_match_or_eos_index_<mode>):
9020 Calculate index using natural element order.
9021 (first_match_index_<mode>):
9022 Calculate index using natural element order.
9023 (first_match_or_eos_index_<mode>):
9024 Calculate index using natural order.
9025 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
9026 for BE and LE modes.
9027 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
9028 P9V_BUILTIN_VCLZLSBB_V16QI.
9029 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
9032 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
9034 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
9035 indentation and line wrap for many prototypes. Add missing
9036 @smallexample directives around block of prototypes for vec_xl and
9039 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
9041 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
9042 track if we pass or return IEEE 128-bit floating point.
9043 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
9044 C++ mangling that is compatible with GCC 8.1.
9045 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
9046 (init_cumulative_args): Note if we pass or return IEEE 128-bit
9047 floating point types.
9048 (rs6000_function_arg_advance_1): Likewise.
9049 (rs6000_mangle_type): Optionally generate mangled names that match
9050 what GCC 8.1 generated for IEEE 128-bit floating point types.
9051 (rs6000_globalize_decl_name): If we have an external function that
9052 passes or returns IEEE 128-bit types, generate a weak reference
9053 from the mangled name used in GCC 8.1 to the current mangled
9055 (rs6000_init_builtins): Make __ibm128 use the long double type if
9056 long double is IBM extended double. Make __float128 use the long
9057 double type if long double is IEEE 128-bit.
9060 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
9061 macro for __ibm128 built-in functions.
9062 (PACK_IF): Add __ibm128 pack/unpack functions.
9063 (UNPACK_IF): Likewise.
9064 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
9065 enable long double built-in functions if long double is IEEE
9066 128-bit floating point.
9067 (rs6000_invalid_builtin): Update long double built-in function
9069 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
9070 functions, adjust the built-in function to use the long double
9071 built-in function if __ibm128 and long double are the same type.
9072 * doc/extend.texi (PowerPC builtins): Update documention for
9073 __builtin_{,un}pack_longdouble. Add documentation for
9074 __builtin_{,un}pack_ibm128.
9076 2018-06-06 Jim Wilson <jimw@sifive.com>
9078 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
9079 (struct machine_function): New field interrupt_mode.
9080 (riscv_handle_type_attribute): New function. Add forward declaration.
9081 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
9082 (riscv_expand_epilogue): Check interrupt_mode field.
9083 (riscv_set_current_function): Check interrupt attribute args and
9084 set interrupt_mode field.
9085 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
9086 (riscv_sret, riscv_uret): New.
9087 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
9088 new arguments to interrupt attribute.
9090 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
9093 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
9094 Don't handle -mcpu=power8 if -mpower9-vector is also used.
9096 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
9098 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
9099 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
9100 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
9101 several redundant entries.
9103 2018-06-06 David Malcolm <dmalcolm@redhat.com>
9105 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
9106 type from "rtx" to "rtx_insn *".
9107 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
9108 for local "call_insn", removing cast.
9109 (ix86_expand_call): Likewise, introducing a "call_insn" local.
9111 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
9113 PR tree-optimization/86066
9114 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
9115 for BIT_INSERT_EXPR stores.
9117 2018-06-06 Richard Biener <rguenther@suse.de>
9119 PR tree-optimization/86062
9120 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
9121 component refs ontop
9122 of to be offsetted base.
9124 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
9126 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
9127 to be static and remove check on interrupt attribute name.
9129 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
9131 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
9132 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
9134 2018-06-05 Steve Ellcey <sellcey@cavium.com>
9137 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
9139 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
9140 Remove second argument, change how error is called.
9141 (aarch64_layout_arg): Remove second argument from
9142 aarch64_err_no_fpadvsimd call.
9143 (aarch64_init_cumulative_args): Ditto.
9144 (aarch64_gimplify_va_arg_expr): Ditto.
9145 * config/aarch64/aarch64.md (mov<mode>): Ditto.
9147 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
9149 * config/i386/i386.md (simple_return_indirect_internal): New expander.
9150 (*simple_return_indirect_internal<mode>): Rename from
9151 simple_return_indirect_internal. Use W mode iterator.
9152 (rstorssp): New expander.
9153 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
9154 (clrssbsy): New expander.
9155 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
9157 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
9159 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
9161 (cmse_check_pointed_object): Likewise.
9163 2018-06-05 Martin Liska <mliska@suse.cz>
9165 PR gcov-profile/47618
9166 * doc/invoke.texi: Document how -fprofile-dir format
9169 2018-06-05 Richard Biener <rguenther@suse.de>
9171 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
9172 removal pretend DOM info isn't available so we do not update
9173 it and only remove edges, not dominated blocks. Actually free
9174 DOM info in case we removed something. Remove unreachable blocks.
9175 (mfb_keep_latches): Work with either DOM info or marked backedges.
9176 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
9177 first. Mark backedges if DOM info isn't available.
9178 (Re-)compute DOM info after cleanup_control_flow_pre.
9180 2018-06-05 Richard Biener <rguenther@suse.de>
9182 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
9183 (locus_discrim_hasher::hash): Adjust.
9184 (locus_discrim_hasher::equal): Likewise.
9185 (next_discriminator_for_locus): Work on line directly.
9186 (same_line_p): Pass in expanded locus1 as well.
9187 (assign_discriminators): Avoid redundant location expansions.
9189 2018-06-05 Richard Biener <rguenther@suse.de>
9191 PR tree-optimization/86046
9192 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
9193 if required after clearing TREE_ADDRESSABLE.
9195 2018-06-05 Richard Biener <rguenther@suse.de>
9197 PR tree-optimization/86047
9198 * tree-ssa-loop.c (for_each_index): Glob handling of all
9199 decls and constants and really handle all of them.
9201 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9204 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
9205 qualifier_void_pointer and qualifier_const_void_pointer.
9206 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
9207 (arm_init_builtins): Handle the above.
9208 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
9209 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
9210 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
9213 2018-06-05 Martin Liska <mliska@suse.cz>
9215 * auto-profile.c (read_autofdo_file): Do not use
9216 gcov_ctr_summary struct.
9217 (afdo_callsite_hot_enough_for_early_inline): Likewise.
9218 * coverage.c (struct counts_entry): Likewise.
9219 (read_counts_file): Read just single summary entry.
9220 (get_coverage_counts): Use gcov_summary struct.
9221 * coverage.h (get_coverage_counts): Likewise.
9222 * gcov-dump.c (dump_working_sets): Likewise.
9223 (tag_summary): Dump just single summary.
9224 * gcov-io.c (gcov_write_summary): Write just histogram
9226 (gcov_read_summary): Read just single summary.
9227 (compute_working_sets): Use gcov_summary struct.
9228 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
9229 of GCOV_COUNTERS_SUMMABLE.
9230 (GCOV_COUNTERS_SUMMABLE): Remove.
9231 (GCOV_FIRST_VALUE_COUNTER): Replace with
9232 GCOV_COUNTER_V_INTERVAL.
9233 (struct gcov_ctr_summary): Remove.
9234 (struct gcov_summary): Directly use fields of former
9236 (compute_working_sets): Use gcov_summary struct.
9237 * gcov.c (read_count_file): Do not use ctrs fields.
9238 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
9240 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
9242 * profile.c: Likewise.
9243 * profile.h: Likewise.
9245 2018-06-05 Martin Liska <mliska@suse.cz>
9247 PR gcov-profile/84846
9248 * gcov.c (output_lines): Print working directory only
9249 in intermediate format.
9251 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
9253 * config/s390/s390-builtin-types.def: Add void function type.
9254 * config/s390/s390-builtins.def: Use the function type for the
9257 2018-06-04 Jim Wilson <jimw@sifive.com>
9259 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
9261 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
9262 and maybe_eh_return. Change regno to unsigned int. Use new args to
9263 handle EH_RETURN_DATA_REGNO registers properly.
9264 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
9265 (riscv_expand_epilogue): Update comment. Change argument name and
9266 type. Update code to use new name and type. Pass new args to
9267 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
9269 * config/riscv/riscv.md (NORMAL_RETURN): New.
9270 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
9271 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
9272 (eh_return): Call gen_eh_return_internal and emit barrier.
9273 (eh_return_internal): Call riscv_expand_epilogue.
9275 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
9277 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
9278 bit_insertion field and declare can_be_merged_into method.
9279 (merged_store_group::can_be_merged_into): New method.
9280 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
9281 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
9282 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
9284 2018-06-04 Richard Biener <rguenther@suse.de>
9286 PR tree-optimization/85955
9287 * builtins.c (fold_builtin_sincos): Convert pointers to
9288 destination to appropriate type before dereferencing.
9290 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
9292 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
9294 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
9296 * expr.c (expand_expr_real_1): Force the operand into memory if
9297 its TYPE_MODE is BLKmode and if there is no integer mode for
9298 the number of bits being extracted.
9300 2018-06-04 Jakub Jelinek <jakub@redhat.com>
9304 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
9305 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
9307 2018-06-04 Richard Biener <rguenther@suse.de>
9309 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
9310 (cleanup_tree_cfg_noloop): ... single caller. Do
9311 start_recording_case_labels later.
9313 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
9315 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
9316 to _IMMINTRIN_H_INCLUDED.
9317 * config/i386/pconfigintrin.h: Ditto.
9318 * config/i386/waitpkgintrin.h: Ditto.
9319 * config/i386/immintrin.h: Add includes for sgxintrin.h,
9320 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
9321 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
9322 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
9323 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
9324 waitpkgintrin.h and cldemoteintrin.h.
9326 2018-06-04 Richard Biener <rguenther@suse.de>
9328 PR tree-optimization/86038
9329 * tracer.c (find_best_successor): Check probability for
9330 being initialized, bail out if not.
9332 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
9335 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
9336 of bits to ignore when comparing architectures.
9338 2018-06-04 Jakub Jelinek <jakub@redhat.com>
9340 PR tree-optimization/69615
9341 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
9342 maximum or minimum of the type, try to merge it also as if
9343 range1 is + [-, x - 1] or + [x + 1, -].
9346 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
9348 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
9350 PR tree-optimization/86034
9351 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
9352 the unsigned bitfield type in a bit insertion sequence if it does not
9353 have a larger precision than the bitfield size.
9354 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
9356 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
9358 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
9360 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
9362 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
9363 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
9364 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
9365 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
9367 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
9369 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
9370 Disable -fdelete-null-pointer-checks for ELF toolchain.
9372 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
9373 Kito Cheng <kito.cheng@gmail.com>
9375 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
9376 (nds32le-*-*, nds32be-*-*): Integrate checking process.
9377 (nds32*-*-*): Add glibc and uclibc conditions.
9378 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
9379 (TARGET_EXCEPT_UNWIND_INFO): Define.
9380 * config/nds32/elf.h: New file.
9381 * config/nds32/linux.h: New file.
9382 * config/nds32/nds32-elf.opt: New file.
9383 * config/nds32/nds32-linux.opt: New file.
9384 * config/nds32/nds32-fp-as-gp.c
9385 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
9386 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
9388 (nds32_asm_file_end): Ditto.
9389 (nds32_print_operand): Ditto.
9390 (nds32_insert_attributes): Ditto.
9391 (nds32_init_libfuncs): New function.
9392 (TARGET_HAVE_TLS): Define.
9393 (TARGET_INIT_LIBFUNCS): Define.
9394 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
9396 (TARGET_ELF): Apply different mcmodel setting.
9397 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
9398 been migrated into elf.h and linux.h files.
9399 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
9400 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
9401 (mcmodel): The content has been migrated into nds32-elf.opt and
9402 nds32-linux.opt files.
9403 * config/nds32/t-elf: New file.
9404 * config/nds32/t-linux: New file.
9406 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
9407 Shiva Chen <shiva0217@gmail.com>
9409 * config/nds32/constants.md (unspec_volatile_element): Add
9410 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
9411 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
9413 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
9414 (make_pass_nds32_fp_as_gp): Declare.
9415 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
9417 (nds32_asm_function_end_prologue): Remove unused asm output.
9418 (nds32_asm_function_begin_epilogue): Remove unused asm output.
9419 (nds32_asm_file_start): Output necessary fp_as_gp information.
9420 (nds32_option_override): Adjust register usage.
9421 (nds32_expand_prologue): Consider fp_as_gp situation.
9422 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
9423 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
9426 (simple_return): Ditto.
9427 (omit_fp_begin): Output special directive for fp_as_gp.
9428 (omit_fp_end): Output special directive for fp_as_gp.
9429 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
9430 mforbid-fp-as-gp): New options.
9432 2018-06-01 Mark Wielaard <mark@klomp.org>
9434 * dwarf2out.c (dwarf2out_finish): Remove generation of
9435 DW_AT_loclists_base.
9437 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
9439 * gimple-ssa-store-merging.c: Include gimple-fold.h.
9440 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
9441 (struct merged_store_group): Add bit_insertion field.
9442 (dump_char_array): Use standard hexadecimal format.
9443 (merged_store_group::merged_store_group): Set bit_insertion to false.
9444 (merged_store_group::apply_stores): Use optimal buffer size. Deal
9445 with BIT_INSERT_EXPR stores. Move up code updating the mask and
9446 also print the mask in the dump file.
9447 (pass_store_merging::gate): Minor tweak.
9448 (imm_store_chain_info::coalesce_immediate): Fix wrong association
9449 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
9450 stores with INTEGER_CST stores.
9451 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
9452 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
9453 and use it throughout. Generate bit insertion sequences if need be.
9454 (pass_store_merging::process_store): Remove redundant condition.
9455 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
9457 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
9459 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
9460 the 128-bit floating point types. Fix function comment.
9462 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9464 * config/aarch64/aarch64-simd.md
9465 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
9467 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
9470 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
9472 PR tree-optimization/85989
9473 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
9475 (backprop::intersect_uses): Check it when deciding whether this
9476 is a backedge reference.
9477 (backprop::process_block): Add each phi to m_visited_phis
9478 after visiting it, then clear it at the end.
9480 2018-06-01 Richard Biener <rguenther@suse.de>
9482 * tree-vectorizer.h (vect_dr_stmt): New function.
9483 (vect_get_load_cost): Adjust.
9484 (vect_get_store_cost): Likewise.
9485 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
9486 Use vect_dr_stmt instead of DR_SMTT.
9487 (vect_record_base_alignments): Likewise.
9488 (vect_calculate_target_alignment): Likewise.
9489 (vect_compute_data_ref_alignment): Likewise and make static.
9490 (vect_update_misalignment_for_peel): Likewise.
9491 (vect_verify_datarefs_alignment): Likewise.
9492 (vector_alignment_reachable_p): Likewise.
9493 (vect_get_data_access_cost): Likewise. Pass down
9494 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
9495 (vect_get_peeling_costs_all_drs): Likewise.
9496 (vect_peeling_hash_get_lowest_cost): Likewise.
9497 (vect_enhance_data_refs_alignment): Likewise.
9498 (vect_find_same_alignment_drs): Likewise.
9499 (vect_analyze_data_refs_alignment): Likewise.
9500 (vect_analyze_group_access_1): Likewise.
9501 (vect_analyze_group_access): Likewise.
9502 (vect_analyze_data_ref_access): Likewise.
9503 (vect_analyze_data_ref_accesses): Likewise.
9504 (vect_vfa_segment_size): Likewise.
9505 (vect_small_gap_p): Likewise.
9506 (vectorizable_with_step_bound_p): Likewise.
9507 (vect_prune_runtime_alias_test_list): Likewise.
9508 (vect_analyze_data_refs): Likewise.
9509 (vect_supportable_dr_alignment): Likewise.
9510 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
9511 (vect_gen_prolog_loop_niters): Likewise.
9512 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
9513 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
9515 (vect_recog_mask_conversion_pattern): Likewise.
9516 (vect_try_gather_scatter_pattern): Likewise.
9517 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
9518 to vect_get_store_cost.
9519 (vect_get_store_cost): Get stmt_info instead of DR.
9520 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
9521 (vect_get_load_cost): Get stmt_info instead of DR.
9523 2018-06-01 Richard Biener <rguenther@suse.de>
9526 * gimple-fold.c (var_decl_component_p): Also allow offsetted
9527 vars wrapped in MEM_REFs.
9529 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
9531 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
9532 Fix subreg tests so that we only return a choice between
9533 GENERAL_REGS and FP_REGS if the original classes included both.
9535 2018-06-01 Richard Biener <rguenther@suse.de>
9538 * tree-ssa-structalias.c (get_function_part_constraint):
9539 Handle NULL fi->decl.
9540 (find_func_aliases_for_call): Properly handle indirect
9541 fi from direct call.
9542 (find_func_clobbers): Likewise.
9543 (ipa_pta_execute): Likewise.
9544 (create_variable_info_for): For functions that are ifunc_resolver
9545 resolve to a varinfo that contains the result of the resolver call.
9546 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
9549 2018-05-31 Michael Collison <michael.collison@arm.com>
9551 * config/aarch64/aarch64.md:
9552 (*fix_to_zero_extenddfdi2): New pattern.
9553 * gcc.target/aarch64/fix_extend1.c: New testcase.
9555 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
9559 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
9560 and BUILT_IN_STRNCMP_EQ.
9561 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
9562 BUILT_IN_STRNCMP_EQ.
9563 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
9564 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
9565 (gimple_fold_builtin): Likewise.
9566 * tree-ssa-strlen.c (compute_string_length): New function.
9567 (determine_min_obsize): New function.
9568 (handle_builtin_string_cmp): New function to handle calls to
9569 string compare functions.
9570 (strlen_optimize_stmt): Add handling to builtin string compare
9572 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
9573 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
9574 * tree.c (build_common_builtin_nodes): Add new defines of
9575 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
9577 2018-05-31 Jakub Jelinek <jakub@redhat.com>
9580 * bb-reorder.c (pass_partition_blocks::gate): Return false for
9581 functions with naked attribute.
9583 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
9585 * config/i386/sse.md (avx_vec_concat<mode>):
9586 Substitute concat_tg_mode mode attribute with xtg_mode.
9587 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
9588 (concat_tg_mode): Remove mode attribute.
9590 2018-05-31 Martin Sebor <msebor@redhat.com>
9593 * calls.c (alloc_max_size): Correct a logic error/typo.
9594 Treat excessive arguments as infinite. Warn for invalid arguments.
9595 * doc/invoke.texi (-Walloc-size-larger-than): Update.
9597 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
9600 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
9601 and movx for Haswell.
9603 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
9604 Cesar Philippidis <cesar@codesourcery.com>
9607 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
9608 when emitting error on private/firstprivate reductions.
9609 * omp-low.c (lower_omp_target): Avoid reference-type processing
9610 on pointers for firstprivate clause.
9612 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
9614 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
9617 * config/aarch64/aarch64-simd.md
9618 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
9619 (aarch64_ld1_x3_<mode>): Likewise
9620 (aarch64_st1x2<VALLDIF:mode>): Likewise
9621 (aarch64_st1_x2_<mode>): Likewise
9622 (aarch64_st1x3<VALLDIF:mode>): Likewise
9623 (aarch64_st1_x3_<mode>): Likewise
9624 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
9625 (vld1_s8_x3): Likewise.
9626 (vld1_u16_x3): Likewise.
9627 (vld1_s16_x3): Likewise.
9628 (vld1_u32_x3): Likewise.
9629 (vld1_s32_x3): Likewise.
9630 (vld1_u64_x3): Likewise.
9631 (vld1_s64_x3): Likewise.
9632 (vld1_f16_x3): Likewise.
9633 (vld1_f32_x3): Likewise.
9634 (vld1_f64_x3): Likewise.
9635 (vld1_p8_x3): Likewise.
9636 (vld1_p16_x3): Likewise.
9637 (vld1_p64_x3): Likewise.
9638 (vld1q_u8_x3): Likewise.
9639 (vld1q_s8_x3): Likewise.
9640 (vld1q_u16_x3): Likewise.
9641 (vld1q_s16_x3): Likewise.
9642 (vld1q_u32_x3): Likewise.
9643 (vld1q_s32_x3): Likewise.
9644 (vld1q_u64_x3): Likewise.
9645 (vld1q_s64_x3): Likewise.
9646 (vld1q_f16_x3): Likewise.
9647 (vld1q_f32_x3): Likewise.
9648 (vld1q_f64_x3): Likewise.
9649 (vld1q_p8_x3): Likewise.
9650 (vld1q_p16_x3): Likewise.
9651 (vld1q_p64_x3): Likewise.
9652 (vst1_s64_x2): Likewise.
9653 (vst1_u64_x2): Likewise.
9654 (vst1_f64_x2): Likewise.
9655 (vst1_s8_x2): Likewise.
9656 (vst1_p8_x2): Likewise.
9657 (vst1_s16_x2): Likewise.
9658 (vst1_p16_x2): Likewise.
9659 (vst1_s32_x2): Likewise.
9660 (vst1_u8_x2): Likewise.
9661 (vst1_u16_x2): Likewise.
9662 (vst1_u32_x2): Likewise.
9663 (vst1_f16_x2): Likewise.
9664 (vst1_f32_x2): Likewise.
9665 (vst1_p64_x2): Likewise.
9666 (vst1q_s8_x2): Likewise.
9667 (vst1q_p8_x2): Likewise.
9668 (vst1q_s16_x2): Likewise.
9669 (vst1q_p16_x2): Likewise.
9670 (vst1q_s32_x2): Likewise.
9671 (vst1q_s64_x2): Likewise.
9672 (vst1q_u8_x2): Likewise.
9673 (vst1q_u16_x2): Likewise.
9674 (vst1q_u32_x2): Likewise.
9675 (vst1q_u64_x2): Likewise.
9676 (vst1q_f16_x2): Likewise.
9677 (vst1q_f32_x2): Likewise.
9678 (vst1q_f64_x2): Likewise.
9679 (vst1q_p64_x2): Likewise.
9680 (vst1_s64_x3): Likewise.
9681 (vst1_u64_x3): Likewise.
9682 (vst1_f64_x3): Likewise.
9683 (vst1_s8_x3): Likewise.
9684 (vst1_p8_x3): Likewise.
9685 (vst1_s16_x3): Likewise.
9686 (vst1_p16_x3): Likewise.
9687 (vst1_s32_x3): Likewise.
9688 (vst1_u8_x3): Likewise.
9689 (vst1_u16_x3): Likewise.
9690 (vst1_u32_x3): Likewise.
9691 (vst1_f16_x3): Likewise.
9692 (vst1_f32_x3): Likewise.
9693 (vst1_p64_x3): Likewise.
9694 (vst1q_s8_x3): Likewise.
9695 (vst1q_p8_x3): Likewise.
9696 (vst1q_s16_x3): Likewise.
9697 (vst1q_p16_x3): Likewise.
9698 (vst1q_s32_x3): Likewise.
9699 (vst1q_s64_x3): Likewise.
9700 (vst1q_u8_x3): Likewise.
9701 (vst1q_u16_x3): Likewise.
9702 (vst1q_u32_x3): Likewise.
9703 (vst1q_u64_x3): Likewise.
9704 (vst1q_f16_x3): Likewise.
9705 (vst1q_f32_x3): Likewise.
9706 (vst1q_f64_x3): Likewise.
9707 (vst1q_p64_x3): Likewise.
9709 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
9711 * config/msp430/msp430.c (msp430_output_labelref): Prepend
9712 user_label_prefix to name.
9714 * tree-core.h: Update comment about the format of NAME string
9715 passed to handler in attribute_spec.
9717 * config/msp430/msp430.md: Remove erroneous subreg expression from
9718 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
9719 zero_extend{q,h}isi2.
9721 2018-05-30 Borislav Petkov <bp@suse.de>
9723 * doc/extend.texi: Document some architecture specific
9724 constraints and sort entries.
9726 2018-05-30 Martin Sebor <msebor@redhat.com>
9729 * builtins.c (expand_builtin_stpcpy_1): New function.
9730 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
9731 only if the former succeeds.
9733 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
9735 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
9738 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
9740 * doc/invoke.texi (-flinker-output): Document
9742 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
9744 * passes.c (ipa_write_summaries): Only modify statements if body
9746 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
9747 incrementally linking.
9748 (ipa_passes): Likewise.
9749 * lto-cgraph.c (lto_output_node): When incrementally linking do not
9750 pass down resolution info.
9751 * common.opt (flag_incremental_link): Update info.
9752 * gcc.c (plugin specs): Turn flinker-output=* to
9753 -plugin-opt=-linker-output-known
9754 * toplev.c (compile_file): Also cut compilation when doing incremental
9756 * flag-types. (enum lto_partition_model): Add
9757 LTO_LINKER_OUTPUT_NOLTOREL.
9758 (invoke.texi): Add -flinker-output docs.
9759 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
9760 link same way as WPA; do not stream in dead initializers.
9762 * dwarf2out.c (dwarf2out_die_ref_for_decl,
9763 darf2out_register_external_decl): Support incremental link.
9765 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
9767 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
9769 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
9771 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
9772 it down to simple_object_copy_lto_debug_sections.
9773 (run_gcc): Determine incremental LTO link time and configure
9774 lto1 into non-wpa mode, disable renaming of debug sections.
9776 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
9778 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
9779 descriptions of various incorrectly documented functions.
9781 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
9784 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
9785 address check not strict.
9787 2018-05-30 Richard Biener <rguenther@suse.de>
9789 PR tree-optimization/85964
9790 * tracer.c (better_p): Drop initialized count check, we only
9791 call the function with initialized counts now.
9792 (find_best_successor): Do find a best edge if one
9793 has uninitialized count.
9794 (find_best_predecessor): Likewise. Do BB frequency check only
9795 if count is initialized.
9797 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
9799 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
9800 (aarch64_ldrstr_offset_compare): New.
9801 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
9802 load/store orderings.
9803 (aarch64_gen_adjusted_ldpstp): Likewise.
9805 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
9807 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
9808 Check for subset of GENERAL_REGS and FP_REGS.
9809 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
9812 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
9814 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
9815 and wi::to_poly_offset. Add the current offset and then check
9816 whether the sum fits, rather than using an unchecked addition of
9817 a checked term. Check for a shwi rather than a uhwi.
9818 * expr.c (get_bit_range): Use tree_to_poly_uint64.
9819 (store_constructor): Use poly_int_tree_p.
9820 (expand_expr_real_1): Likewise.
9821 * function.c (assign_temp): Likewise.
9822 * fold-const.c (const_binop): Use poly_int_tree_p and
9824 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
9826 * ipa-icf-gimple.c (func_checker::compare_operand): Use
9827 to_poly_offset for MEM offsets.
9828 * ipa-icf.c (sem_variable::equals): Likewise.
9829 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
9830 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
9831 wi::to_poly_offset for BIT_FIELD_REF offsets.
9832 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
9834 * var-tracking.c (emit_note_insn_var_location): Use
9835 tree_to_poly_uint64.
9837 2018-05-29 Jim Wilson <jimw@sifive.com>
9839 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
9841 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
9844 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
9845 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
9847 (sse4_1_round<mode>2): Use nonimmediate_operand
9848 for operand 1 predicate.
9850 2018-05-29 Martin Sebor <msebor@redhat.com>
9851 Richard Biener <rguenther@suse.de>
9854 * calls.c (get_size_range): Call determine_value_range instead
9855 of get_value_range..
9856 * tree-vrp.h (determine_value_range): Declared new function.
9857 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
9859 2018-05-29 Richard Biener <rguenther@suse.de>
9861 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
9862 sure to use non-pattern stmts for get_earlier_stmt arguments.
9863 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
9864 called on pattern stmts.
9865 (get_later_stmt): Likewise.
9867 2018-05-29 Martin Liska <mliska@suse.cz>
9869 PR gcov-profile/85759
9870 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
9873 2018-05-29 Jakub Jelinek <jakub@redhat.com>
9875 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
9877 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
9881 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
9882 VEC_PACK_FLOAT_EXPR): New tree codes.
9883 * tree-pretty-print.c (op_code_prio): Handle
9884 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
9885 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
9886 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
9887 * tree-inline.c (estimate_operator_cost): Likewise.
9888 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
9889 * fold-const.c (const_binop): Likewise.
9890 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
9891 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
9892 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
9893 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
9894 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
9895 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
9896 * expr.c (expand_expr_real_2): Likewise.
9897 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
9898 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
9899 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
9901 * optabs.c (expand_widen_pattern_expr): For
9902 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
9903 sign from result type rather than operand's type.
9904 (expand_binop_directly): For vec_packu_float_optab and
9905 vec_packs_float_optab allow result type to be different from operand's
9907 * optabs-tree.c (optab_for_tree_code): Handle
9908 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
9909 VEC_PACK_FLOAT_EXPR. Formatting fixes.
9910 * tree-vect-generic.c (expand_vector_operations_1): Handle
9911 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
9912 VEC_PACK_FLOAT_EXPR.
9913 * tree-vect-stmts.c (supportable_widening_operation): Handle
9915 (supportable_narrowing_operation): Handle FLOAT_EXPR.
9916 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
9917 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
9918 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
9919 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
9921 (vec_pack<floatprefix>_float_<mode>): New expander.
9922 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
9924 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
9925 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
9926 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
9927 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
9928 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
9930 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
9931 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
9932 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
9933 VEC_PACK_FLOAT_EXPR): Document.
9935 2018-05-29 Richard Biener <rguenther@suse.de>
9937 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
9939 (stmt_vec_info_vec): Make pointer.
9940 (init_stmt_vec_info_vec): Remove.
9941 (free_stmt_vec_info_vec): Likewise.
9942 (set_stmt_vec_info_vec): New function.
9943 (free_stmt_vec_infos): Likewise.
9944 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
9945 (set_vinfo_for_stmt): Likewise.
9946 (get_earlier_stmt): Likewise.
9947 (get_later_stmt): Likewise.
9948 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
9949 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
9950 (vec_info::~vec_info): Free stmt_vec_infos.
9951 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
9952 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
9953 (pass_slp_vectorize::execute): Likewise.
9954 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
9955 (free_stmt_vec_info_vec): Likewise.
9956 (set_stmt_vec_info_vec): New function.
9957 (free_stmt_vec_infos): Likewise.
9958 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
9959 the global stmt_vec_info_vec.
9960 * tree-parloops.c (gather_scalar_reductions): Use
9961 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
9964 2018-05-29 Richard Biener <rguenther@suse.de>
9966 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
9968 2018-05-29 Martin Liska <mliska@suse.cz>
9969 David Malcolm <dmalcolm@redhat.com>
9971 * vec.c (test_reverse): New.
9972 (vec_c_tests): Add new test.
9973 * vec.h (vl_ptr>::reverse): New function.
9975 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
9977 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
9979 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
9982 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
9984 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
9986 2018-05-28 Richard Biener <rguenther@suse.de>
9988 PR tree-optimization/85933
9989 * tree-vect-data-refs.c (vect_record_base_alignments): Only
9990 look at stmts marked as vectorizable.
9992 2018-05-28 Richard Biener <rguenther@suse.de>
9994 PR tree-optimization/85934
9995 * tree-vect-generic.c (expand_vector_operations_1): Hoist
9996 vector boolean check before scalar optimization.
9998 2018-05-28 Jakub Jelinek <jakub@redhat.com>
10000 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
10003 2018-05-28 Mark Wielaard <mark@klomp.org>
10005 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
10006 if it is an expression containing a minus sign.
10008 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
10010 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
10012 2018-05-27 Paul Koning <ni1d@arrl.net>
10014 * config/pdp11/pdp11.md (truncsihi2): Remove.
10016 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
10017 Chung-Ju Wu <jasonwucj@gmail.com>
10019 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
10021 (unaligned_store_dw): Ditto.
10022 * config/nds32/nds32-memory-manipulation.c
10023 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
10024 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
10025 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
10026 (emit_setmem_word_loop): Rename to ...
10027 (emit_setmem_doubleword_loop): ... this.
10028 (nds32_gen_dup_4_byte_to_word_value): New function.
10029 (nds32_gen_dup_8_byte_to_double_word_value): New function.
10030 (nds32_expand_setmem_loop): Refine implementation.
10031 (nds32_expand_setmem_loop_v3m): Ditto.
10032 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
10035 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
10037 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
10039 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
10041 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
10042 (nds32_init_machine_status): Initialize machine->attr_naked_p and
10043 machine->attr_no_prologue_p.
10044 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
10045 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
10046 (nds32_expand_epilogue): Consider attr_naked_p.
10047 (nds32_expand_epilogue_v3pop): Likewise.
10048 (nds32_can_use_return_insn): Likewise.
10049 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
10050 attr_no_prologue_p fields.
10051 * config/nds32/nds32.opt (mret-in-naked-func): New option.
10053 2018-05-27 Jakub Jelinek <jakub@redhat.com>
10056 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
10058 * config/i386/sse.md
10059 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
10061 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
10063 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
10065 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
10067 (*<floatsuffix>floatv2div2sf2): Rename to ...
10068 (*float<floatunssuffix>v2div2sf2): ... this.
10069 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
10070 (float<floatunssuffix>v2div2sf2_mask): ... this.
10071 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
10072 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
10073 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
10075 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
10077 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
10079 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
10081 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
10083 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
10085 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
10086 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
10087 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
10088 gen_ufix_truncv8dfv8si2.
10089 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
10090 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
10091 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
10092 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
10093 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
10094 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
10095 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
10096 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
10098 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
10102 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
10104 2018-05-25 Jim Wilson <jimw@sifive.com>
10106 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
10107 * config/riscv/riscv.c (struct machine_function): Add
10108 interrupt_handler_p and attribute_checked_p fields.
10109 (riscv_attribute_table): Add interrupt.
10110 (riscv_interrupt_type_p): New.
10111 (riscv_save_reg_p): Save extra regs for interrupt handler.
10112 (riscv_use_save_libcall): Return false for interrupt handler.
10113 (riscv_first_stack_step): Add forward declaration.
10114 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
10115 for interrupt handler with large frame. Use it for saved reg list.
10116 (riscv_expand_prologue): Move flag_stack_usage_info support to
10117 eliminate duplication.
10118 (riscv_expand_epilogue): Generate mret for interrupt handler.
10119 (riscv_epilogue_uses): New.
10120 (riscv_can_use_return_insn): Return false for interrupt handler.
10121 (riscv_function_ok_for_sibcall): Likewise.
10122 (riscv_set_current_function): Add interrupt handler support.
10123 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
10124 * config/riscv/riscv.md (UNSPECV_MRET): New.
10126 (riscv_frflags, riscv_fsflags): Use tab after opcode.
10128 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
10130 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
10132 PR tree-optimization/85712
10133 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
10134 this candidate has already been replaced in-situ by a copy.
10136 2018-05-25 Jason Merrill <jason@redhat.com>
10138 PR c++/80485 - inline function non-zero address.
10139 * symtab.c (nonzero_address): Check DECL_COMDAT.
10141 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
10144 * config/alpha/alpha.md (ashlsi3): New insn pattern.
10145 (*ashlsi_se): Rename from *ashldi_se. Define as sign
10146 extension of SImode operation. Use const123_operand predicate.
10147 (*saddsi_1): Remove.
10148 (*saddl_se_1): Ditto.
10149 (*ssubsi_1): Ditto.
10150 (*ssubl_se_1): Ditto.
10151 * config/alpha/predicates.md (const123_operand): New predicate.
10152 * config/alpha/constraints.md (P): Use IN_RANGE.
10154 2018-05-25 Richard Biener <rguenther@suse.de>
10156 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
10158 (ref_maybe_used_by_stmt_p): Likewise.
10159 (stmt_may_clobber_ref_p): Likewise.
10160 (stmt_may_clobber_ref_p_1): Likewise.
10161 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
10163 (ref_maybe_used_by_stmt_p): Likewise.
10164 (stmt_may_clobber_ref_p): Likewise.
10165 (stmt_may_clobber_ref_p_1): Likewise.
10166 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
10167 the alias oracle to disambiguate DRs with stmts DR analysis
10169 (vect_analyze_data_refs): Do not give up on not analyzable
10170 DRs for BB vectorization. Remove code truncating the dataref
10173 2018-05-25 Jakub Jelinek <jakub@redhat.com>
10176 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
10177 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
10178 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
10180 2018-05-25 Richard Biener <rguenther@suse.de>
10182 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
10183 function, combining stmt data ref gathering and fatal analysis
10185 (vect_analyze_data_refs): Remove now redudnant code and simplify.
10186 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
10187 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
10188 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
10189 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
10191 2018-05-25 Bin Cheng <bin.cheng@arm.com>
10193 PR tree-optimization/85720
10194 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
10195 SCC if all partitions are builtins.
10196 (version_loop_by_alias_check): New parameter. Generate cancelable
10197 runtime alias check if all partitions are builtins.
10198 (distribute_loop): Update call to above function.
10200 2018-05-25 Bin Cheng <bin.cheng@arm.com>
10202 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
10203 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
10204 (parm_default_def_partition_arg): Ditto.
10205 (set_parm_default_def_partition): Ditto.
10206 (get_parm_default_def_partitions): Ditto and make it static.
10207 (get_undefined_value_partitions): Ditto and make it static.
10208 (remove_ssa_form): Refactor call to init_var_map here.
10209 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
10210 computation for loop region.
10211 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
10212 (register_default_def): Delete.
10213 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
10214 (parm_default_def_partition_arg): Ditto.
10215 (set_parm_default_def_partition): Ditto.
10216 (get_parm_default_def_partitions): Ditto and make it static.
10217 (get_undefined_value_partitions): Ditto and make it static.
10218 (coalesce_with_default, coalesce_with_default): Update comment.
10219 (create_coalesce_list_for_region): New func factored out from
10220 create_outofssa_var_map.
10221 (populate_coalesce_list_for_outofssa): New func factored out from
10222 create_outofssa_var_map and coalesce_ssa_name.
10223 (create_outofssa_var_map): Delete.
10224 (coalesce_ssa_name): Refactor to support live range computation.
10225 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
10226 (get_parm_default_def_partitions): Delete.
10227 (get_undefined_value_partitions): Ditto.
10228 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
10229 computation for loop region.
10230 (new_tree_live_info, loe_visit_block): Ditto.
10231 (live_worklist, set_var_live_on_entry): Ditto.
10232 (calculate_live_on_exit, verify_live_on_entry): Ditto.
10233 * tree-ssa-live.h (struct _var_map): New fields.
10234 (init_var_map): Change decl.
10235 (region_contains_p): New.
10237 2018-05-25 Bin Cheng <bin.cheng@arm.com>
10239 * tree-ssa-live.h (live_merge_and_clear): Delete.
10241 2018-05-25 Richard Biener <rguenther@suse.de>
10244 * tree-dump.c (dequeue_and_dump): Remove access to removed
10245 operand 2 of a SWITCH_EXPR.
10247 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
10249 * doc/sourcebuild.texi (vect_double_cond_arith): Include
10250 multiplication and division.
10251 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
10252 (cond_udiv@var{m}, cond_umod@var{m}): Document.
10253 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
10254 (cond_udiv_optab, cond_umod_optab): New optabs.
10255 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
10256 (IFN_COND_RDIV): New internal functions.
10257 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
10258 TRUNC_MOD_EXPR and RDIV_EXPR.
10259 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
10260 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
10262 (SVE_INT_BINARY): Include mult.
10263 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
10264 (optab, sve_int_op): Handle mult.
10265 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
10267 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
10268 for SVE_INT_BINARY_SD.
10270 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
10272 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
10273 (optab, sve_int_op): Handle div and udiv.
10274 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
10275 for SVE_INT_BINARY_SD.
10276 (*<optab><mode>3): New insn for the same.
10278 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
10280 * tree-vect-patterns.c: Include predict.h.
10281 (vect_recog_divmod_pattern): Restrict check for division support
10282 to when optimizing for size.
10284 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
10286 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
10287 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
10288 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
10289 (gimple_match_op::set_op): Likewise.
10290 (gimple_resimplify4): Declare.
10291 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
10292 (expr::gen_transform): Likewise.
10293 (decision_tree::gen): Generate a simplification routine for 4 operands.
10294 * gimple-match-head.c (gimple_simplify): Add an overload for
10295 4 operands. In the top-level function, handle up to 4 call
10296 arguments and call gimple_resimplify4.
10297 (gimple_resimplify4): New function.
10298 (build_call_internal): Pass a fourth operand.
10299 (maybe_push_to_seq): Likewise.
10300 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
10301 Fold VEC_COND_EXPRs of an operation and a default value into
10302 an IFN_COND_* function if possible.
10303 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
10305 (SVE_COND_FP_BINARY): Include them.
10306 (optab, sve_fp_op): Handle them.
10307 (SVE_INT_BINARY_REV): New code iterator.
10308 (SVE_COND_FP_BINARY_REV): New int iterator.
10309 (commutative): New int attribute.
10310 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
10312 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
10314 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
10315 (*cond_<optab><mode>): New patterns for reversed operands.
10317 2018-05-25 Richard Biener <rguenther@suse.de>
10319 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
10320 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
10321 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
10322 (STMT_VINFO_GROUPED_ACCESS): Adjust.
10323 * tree-vect-data-refs.c (everywhere): Adjust users.
10324 * tree-vect-loop.c (everywhere): Likewise.
10325 * tree-vect-slp.c (everywhere): Likewise.
10326 * tree-vect-stmts.c (everywhere): Likewise.
10327 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
10329 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10331 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
10333 (gcc_cv_as_section_exclude): ... this.
10334 Try Solaris as #exclude syntax.
10335 * configure: Regenerate.
10336 * config.in: Regenerate.
10337 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
10339 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
10340 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
10342 * varasm.c (default_elf_asm_named_section): Don't check if
10343 HAVE_GAS_SECTION_EXCLUDE is defined.
10345 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
10347 * doc/md.texi: Update the documentation of the cond_* optabs
10348 to mention the new final operand. Fix GET_MODE_NUNITS call.
10349 Describe the scalar case too.
10350 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
10351 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
10353 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
10354 (get_conditional_internal_fn): Update comment.
10355 * tree-vect-loop.c (vectorizable_reduction): Pass the original
10356 accumulator value as a final argument to conditional functions.
10357 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
10358 a define_expand and add an "else" operand. Assert for now that
10359 the else operand is equal to operand 2. Use SVE_INT_BINARY and
10360 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
10361 (*cond_<optab><mode>): New patterns.
10362 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
10363 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
10364 (UNSPEC_COND_EOR): Delete.
10365 (optab): Remove associated mappings.
10366 (SVE_INT_BINARY): New code iterator.
10367 (sve_int_op): Remove int attribute and add "minus" to the code
10369 (SVE_COND_INT_OP): Delete.
10370 (SVE_COND_FP_OP): Rename to...
10371 (SVE_COND_FP_BINARY): ...this.
10373 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
10375 * optabs.c (can_reuse_operands_p): New function.
10376 (maybe_legitimize_operands): Try to reuse the results for
10379 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
10381 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
10382 Add {q} suffix to insn mnemonic.
10384 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
10386 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
10387 (msp430_warn_func_return): New.
10389 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
10391 * fold-const.c (tree_nonzero_bits): New function.
10392 * fold-const.h (tree_nonzero_bits): Likewise.
10393 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
10394 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
10396 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
10400 * varasm.c (assemble_alias): Check ifunc_resolver only on
10403 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
10406 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
10407 when memory input operand is handled.
10409 2018-05-24 Luis Machado <luis.machado@linaro.org>
10411 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
10413 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
10415 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
10417 * match.pd: Delay FMA folds until after vectorization.
10419 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
10422 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
10423 address check not strict.
10425 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
10427 * gimple-match.h (gimple_match_op): New class.
10428 (mprts_hook): Replace parameters with a gimple_match_op *.
10429 (maybe_build_generic_op): Likewise.
10430 (gimple_simplified_result_is_gimple_val): Replace parameters with
10431 a const gimple_match_op *.
10432 (gimple_simplify): Replace code_helper * and tree * parameters with
10433 a gimple_match_op * parameter.
10434 (gimple_resimplify1): Replace code_helper *, tree and tree *
10435 parameters with a gimple_match_op * parameter.
10436 (gimple_resimplify2): Likewise.
10437 (gimple_resimplify3): Likewise.
10438 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
10439 parameters with a gimple_match_op * parameter.
10440 * gimple-match-head.c (gimple_simplify): Change prototypes of
10441 auto-generated functions to take a gimple_match_op * instead of
10442 separate code_helper * and tree * parameters. Make the same
10443 change in the top-level overload and update calls to the
10444 gimple_resimplify routines. Update calls to the auto-generated
10445 functions and to maybe_push_res_to_seq in the publicly-facing
10446 operation-specific gimple_simplify overloads.
10447 (gimple_match_op::MAX_NUM_OPS): Define.
10448 (gimple_resimplify1): Replace rcode and ops with a single res_op
10449 parameter. Update call to gimple_simplify.
10450 (gimple_resimplify2): Likewise.
10451 (gimple_resimplify3): Likewise.
10452 (mprts_hook): Replace parameters with a gimple_match_op *.
10453 (maybe_build_generic_op): Likewise.
10454 (build_call_internal): Replace type, nargs and ops with
10455 a gimple_match_op *.
10456 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
10457 with a single gimple_match_op *. Update calls to mprts_hook,
10458 build_call_internal and gimple_simplified_result_is_gimple_val.
10459 Factor out code that is common to the tree_code and combined_fn cases.
10460 * genmatch.c (expr::gen_transform): Replace tem_code and
10461 tem_ops with a gimple_match_op called tem_op. Update calls
10462 to the gimple_resimplify functions and maybe_push_res_to_seq.
10463 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
10464 res_ops. Update call to the gimple_resimplify functions.
10465 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
10466 (decision_tree::gen): Make the functions take a gimple_match_op *
10467 called res_op instead of separate res_code and res_ops parameters.
10468 Update call accordingly.
10469 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
10470 and ops with a single res_op parameter. Update calls to
10471 maybe_build_generic_op and maybe_push_res_to_seq.
10472 (fold_stmt_1): Update calls to gimple_simplify and
10473 replace_stmt_with_simplification.
10474 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
10475 and gimple_simplified_result_is_gimple_val.
10476 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
10478 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
10479 with a gimple_match_op *.
10480 (vn_nary_build_or_lookup): Likewise. Update call to
10481 vn_nary_build_or_lookup_1.
10482 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
10483 gimple_match_op *. Update calls to the gimple_resimplify routines
10484 and to gimple_simplified_result_is_gimple_val.
10485 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
10486 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
10487 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
10488 (visit_nary_op): Likewise.
10489 (visit_reference_op_load): Likewise.
10491 2018-05-23 Luis Machado <luis.machado@linaro.org>
10493 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
10494 modifier for printing the step amount.
10496 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
10499 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
10502 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
10504 * doc/sourcebuild.texi (Endianness): New subsubsection.
10506 2018-05-23 Luis Machado <luis.machado@linaro.org>
10508 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
10509 <prefetch_dynamic_strides>: New const bool field.
10510 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
10511 prefetch_dynamic_strides.
10512 (exynosm1_prefetch_tune): Likewise.
10513 (thunderxt88_prefetch_tune): Likewise.
10514 (thunderx_prefetch_tune): Likewise.
10515 (thunderx2t99_prefetch_tune): Likewise.
10516 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
10518 (aarch64_override_options_internal): Update to set
10519 PARAM_PREFETCH_DYNAMIC_STRIDES.
10520 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
10521 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
10522 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
10523 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
10524 prefetch-dynamic-strides setting.
10526 2018-05-23 Luis Machado <luis.machado@linaro.org>
10528 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
10529 <minimum_stride>: New const int field.
10530 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
10531 minimum_stride field defaulting to -1.
10532 (exynosm1_prefetch_tune): Likewise.
10533 (thunderxt88_prefetch_tune): Likewise.
10534 (thunderx_prefetch_tune): Likewise.
10535 (thunderx2t99_prefetch_tune): Likewise.
10536 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
10537 <default_opt_level>: Set to 3.
10538 (aarch64_override_options_internal): Update to set
10539 PARAM_PREFETCH_MINIMUM_STRIDE.
10540 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
10541 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
10542 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
10543 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
10544 stride is constant and is below the minimum stride threshold.
10546 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10548 * config/arm/arm-cpus.in (mode26): Delete.
10549 (armv4): Delete mode26 reference.
10550 * config/arm/arm.c (arm_configure_build_target): Delete use of
10553 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
10555 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
10557 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
10558 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
10560 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
10561 (floatunsdidf2): Ditto.
10563 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
10565 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
10566 (fixuns_trunc<mode>si2_avx512f): Ditto.
10567 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
10568 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
10569 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
10571 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
10573 PR rtl-optimization/79985
10574 * df-scan.c (df_insn_refs_collect): Remove special case for
10575 global registers and asm statements.
10577 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
10579 * extend.texi (Global Register Variables): Rewrite the bullet list.
10580 Note that the register is available for allocation. Note that access
10581 via inline asm must use constraints. Add note about async-signal
10582 handlers. Remove paragraph about automagic register selection.
10584 2018-05-23 Richard Biener <rguenther@suse.de>
10586 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
10587 of fixed offset from memset VN.
10589 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
10591 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
10592 first_interp field.
10593 (alloc_cand_and_find_basis): Initialize first_interp field.
10594 (slsr_process_mul): Modify first_interp field.
10595 (slsr_process_add): Likewise.
10596 (slsr_process_cast): Modify first_interp field for each new
10598 (slsr_process_copy): Likewise.
10599 (dump_candidate): Dump first_interp field.
10600 (replace_mult_candidate): Process all interpretations, not just
10602 (replace_rhs_if_not_dup): Likewise.
10603 (replace_one_candidate): Likewise.
10605 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
10607 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
10609 (aarch64_needs_frame_chain): New function.
10610 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
10612 2018-05-23 Sudakshina Das <sudi.das@arm.com>
10615 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
10616 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
10617 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
10618 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
10619 as true for strict-align.
10620 (aarch64_can_inline_p): Perform checks even when callee has no
10621 attributes to check for strict alignment.
10622 * doc/extend.texi (AArch64 Function Attributes): Document
10624 * doc/invoke.texi: (AArch64 Options): Likewise.
10626 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
10628 PR tree-optimization/85853
10629 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
10630 the handling of the root of the node to...
10631 (vect_slp_analyze_node_operations_1): ...this new function,
10632 and run the whole thing with the child nodes' def types
10633 set according to their SLP node's def type.
10635 2018-05-23 Richard Biener <rguenther@suse.de>
10637 PR middle-end/85874
10638 * tree-data-ref.c (create_runtime_alias_checks): Defer
10639 and ignore overflow warnings.
10641 2018-05-23 Yury Gribov <tetra2005@gmail.com>
10643 PR tree-optimization/85822
10644 * tree-vrp.c (is_masked_range_test): Fix handling of negative
10647 2018-05-23 Richard Biener <rguenther@suse.de>
10649 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
10650 memset constants via native_interpret_expr.
10652 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
10655 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
10657 (cgraph_node::create_alias): Likewise.
10658 (cgraph_node::get_availability): Check ifunc_resolver instead
10659 of looking up ifunc attribute.
10660 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
10661 * varasm.c (do_assemble_alias): Likewise.
10662 (assemble_alias): Likewise.
10663 (default_binds_local_p_3): Likewise.
10664 * cgraph.h (cgraph_node): Add ifunc_resolver.
10665 (cgraph_node::only_called_directly_or_aliased_p): Return false
10666 for IFUNC resolver.
10667 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
10669 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
10670 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
10671 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
10672 instead of looking up ifunc attribute.
10674 2018-05-22 Luis Machado <luis.machado@linaro.org>
10676 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
10678 2018-05-22 Martin Sebor <msebor@redhat.com>
10680 PR middle-end/85359
10681 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
10682 only when expasion succeeds.
10683 (expand_builtin_strcmp): Same.
10684 (expand_builtin_strncmp): Same.
10686 2018-05-22 Martin Sebor <msebor@redhat.com>
10688 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
10690 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
10691 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10693 * config/aarch64/aarch64-ldpstp.md: Replace uses of
10694 aarch64_mem_pair_operand with memory_operand and delete operand swapping
10696 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
10697 Add check for legitimate_address.
10698 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
10699 (aarch64_swap_ldrstr_operands): New.
10700 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
10703 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
10704 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10706 * config/aarch64/aarch64.md: New patterns to generate stp
10708 (store_pair_sw, store_pair_dw): New patterns to generate stp for
10709 single words and double words.
10710 (load_pair_sw, load_pair_dw): Likewise.
10711 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
10713 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
10715 * config/aarch64/aarch64-ldpstp.md: Modify peephole
10716 for different mode ldpstp and add peephole for merged zero stores.
10717 Likewise for loads.
10718 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
10720 (aarch64_gen_store_pair): Rename calls to match new patterns.
10721 (aarch64_gen_load_pair): Rename calls to match new patterns.
10722 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
10723 (load_pair<DREG:mode><DREG2:mode>): ... This.
10724 (store_pair<mode>): Rename to...
10725 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
10726 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
10727 New mode iterators.
10728 (V_INT_EQUIV): Handle SImode.
10729 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
10732 2018-05-22 Martin Sebor <msebor@redhat.com>
10735 * calls.c (maybe_warn_nonstring_arg): Use string length to set
10736 or ajust the presumed bound on an operation to avoid unnecessary
10739 2018-05-22 Martin Sebor <msebor@redhat.com>
10741 PR tree-optimization/85826
10742 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
10743 assuming that a DECL necesarily has a constant size.
10745 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
10747 PR middle-end/85862
10748 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
10750 2018-05-22 Richard Biener <rguenther@suse.de>
10752 PR tree-optimization/85834
10753 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
10754 non-constant and non-zero memset arguments.
10756 2018-05-22 Martin Liska <mliska@suse.cz>
10759 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
10761 2018-05-22 Richard Biener <rguenther@suse.de>
10763 PR tree-optimization/85863
10764 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
10765 comparisons when vectype is specified.
10766 (vectorizable_condition): Do not specify vectype for
10767 vect_is_simple_cond when SLP vectorizing.
10769 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
10772 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
10773 define __ibm128 as long double.
10774 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
10775 as a distinct type when IEEE 128-bit support is enabled.
10776 (init_float128_ieee): Fix up conversions between IFmode and IEEE
10777 128-bit types to use the correct functions.
10778 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
10779 convert between 128-bit floating point types that have different
10780 modes but the same representation, instead of using gen_lowpart to
10782 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
10784 (IFKF_reg): New attributes to give the register constraints for
10786 (extend<mode>tf2_internal): New insns to mark an explicit
10787 conversion between 128-bit floating point types that have a
10788 different mode but share the same representation.
10790 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
10792 PR tree-optimization/85814
10793 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
10794 a null return from get_strinfo when unsharing the next
10795 strinfo in the chain.
10797 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
10800 * varasm.c (weak_finish): Clean up weak_decls.
10802 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10804 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
10805 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
10806 UNSPEC_UADALP values.
10807 * config/aarch64/iterators.md (ABAL): New int iterator.
10810 (sur): Add mappings for the above.
10811 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
10813 (aarch64_<sur>abal<mode>_4): Likewise.
10814 (aarch64_<sur>adalp<mode>_3): Likewise.
10815 (<sur>sadv16qi): New define_expand.
10817 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
10819 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
10820 (*movdf_internal): Ditto.
10821 (*rcpsf2_sse): Ditto.
10822 (*rsqrtsf2_sse): Ditto.
10823 (*sqrt<mode>2_sse): Ditto.
10825 2018-05-21 Tamar Christina <tamar.christina@arm.com>
10827 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
10829 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
10830 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
10831 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
10832 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
10834 * config/aarch64/arm_neon.h: Likewise.
10835 * config/aarch64/iterators.md (VQ_I): New.
10837 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
10839 * config.gcc: Add arc/t-multilib-linux to tmake_file for
10841 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
10844 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
10846 * config/nds32/constraints.md (S): New constraint.
10847 * config/nds32/nds32.md (call_internal): Use constraint S.
10848 (call_value_internal): Likewise.
10849 (sibcall_internal): Likewise.
10850 (sibcall_value_internal): Likewise.
10852 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
10853 Chung-Ju Wu <jasonwucj@gmail.com>
10855 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
10856 into consideration.
10858 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
10859 Chung-Ju Wu <jasonwucj@gmail.com>
10861 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
10862 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
10863 (nds32_rtx_costs_impl): Simplify.
10864 (nds32_address_cost_impl): Simplify.
10865 (nds32_init_rtx_costs): New function.
10866 (nds32_rtx_costs_speed_prefer): Likewise.
10867 (nds32_rtx_costs_size_prefer): Likewise.
10868 (nds32_address_cost_speed_prefer): Likewise.
10869 (nds32_address_cost_speed_fwprop): Likewise.
10870 (nds32_address_cost_size_prefer): Likewise.
10871 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
10872 * config/nds32/nds32.c (nds32_option_override): Use
10873 nds32_init_rtx_costs function.
10875 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
10877 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
10878 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
10879 (TARGET_PIPELINE_N8): Likewise.
10880 (TARGET_PIPELINE_N10): Likewise.
10881 (TARGET_PIPELINE_N13): Likewise.
10882 (TARGET_PIPELINE_GRAYWOLF): Likewise.
10884 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
10886 * config/nds32/nds32-fpu.md: Update copyright year.
10888 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
10890 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
10892 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
10894 * config/nds32/nds32.c
10895 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
10896 * config/nds32/nds32.opt (minline-asm-r15): New option.
10898 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
10900 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
10902 * config/nds32/nds32.md (abssi2): New pattern.
10904 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
10906 * config/i386/i386.md (rex64namesuffix): New mode attribute.
10907 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
10908 Merge insn pattern from sse_cvtsi2ss<round_name> and
10909 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
10910 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
10911 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
10912 using SWI48 mode iterator.
10913 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
10914 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
10915 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
10916 pattern from sse_cvttss2si<round_saeonly_name>
10917 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
10918 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
10919 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
10920 using SWI48 mode iterator.
10921 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
10922 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
10923 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
10924 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
10925 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
10926 using SWI48 mode iterator.
10927 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
10928 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
10929 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
10930 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
10931 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
10932 SWI48 mode iterator.
10933 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
10934 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
10935 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
10936 pattern from sse_cvttsd2si<round_saeonly_name>
10937 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
10939 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
10941 * config/nds32/nds32-md-auxiliary.c
10942 (nds32_valid_smw_lwm_base_p): Refine.
10943 (nds32_output_smw_single_word): Refine.
10944 (nds32_output_smw_double_word): New.
10945 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
10947 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
10949 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
10950 (nds32_output_stack_pop): Refine.
10951 (nds32_expand_unaligned_load): Refine.
10952 (nds32_expand_unaligned_store): Refine.
10954 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
10955 Chung-Ju Wu <jasonwucj@gmail.com>
10957 * config/nds32/constants.md: Add TP_REGNUM constant.
10958 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
10959 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
10961 * config/nds32/nds32-doubleword.md: Consider flag_pic.
10962 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
10963 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
10964 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
10965 and PIC code generation.
10966 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
10968 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
10970 * config/nds32/nds32.md: Support TLS and PIC.
10971 * config/nds32/nds32.c: Support TLS and PIC.
10972 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
10973 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
10976 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
10978 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
10979 mode with E_ prefix.
10981 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
10982 Chung-Ju Wu <jasonwucj@gmail.com>
10984 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
10985 * config/nds32/nds32-md-auxiliary.c
10986 (symbolic_reference_mentioned_p): New.
10987 (nds32_legitimize_ict_address): New.
10988 (nds32_expand_ict_move): New.
10989 (nds32_indirect_call_referenced_p): New.
10990 (nds32_symbol_binds_local_p): Delete.
10991 (nds32_long_call_p): Modify.
10992 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
10993 * config/nds32/nds32-protos.h
10994 (symbolic_reference_mentioned_p): Declare.
10995 (nds32_legitimize_ict_address): Declare.
10996 (nds32_expand_ict_move): Declare.
10997 (nds32_indirect_call_referenced_p): Declare.
10998 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
10999 (nds32_relax_group): Use nds32_ict_const_p as condition.
11000 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
11001 (nds32_asm_file_start): Output ict_model directive in asm code.
11002 (nds32_legitimate_address_p): Consider indirect call.
11003 (nds32_print_operand): Consider indirect call.
11004 (nds32_print_operand_address): Consider indirect call.
11005 (nds32_insert_attributes): Handle "indirect_call" attribute.
11006 (TARGET_LEGITIMATE_ADDRESS_P): Define.
11007 (TARGET_LEGITIMATE_CONSTANT_P): Define.
11008 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
11009 (TARGET_DELEGITIMIZE_ADDRESS): Define.
11010 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
11011 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
11012 (TARGET_ICT_MODEL_SMALL): Define.
11013 (TARGET_ICT_MODEL_LARGE): Define.
11014 * config/nds32/nds32.md (movsi): Consider ict model.
11015 (call, call_value): Consider ict model.
11016 (sibcall, sibcall_value): Consider ict model.
11017 * config/nds32/nds32.opt (mict-model): New option.
11018 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
11021 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
11022 Monk Chiang <sh.chiang04@gmail.com>
11023 Jim Wilson <jimw@sifive.com>
11025 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
11026 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
11027 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
11028 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
11029 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
11030 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
11031 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
11032 compute save_libcall_adjustment properly.
11033 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
11034 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
11035 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
11036 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
11037 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
11038 (ABI_SPEC): Handle mabi=ilp32e.
11039 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
11040 (RVE): Add RVE mask.
11041 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
11042 <-march>: Add rv32e as an example.
11044 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
11047 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
11048 (intra_create_variable_infos): Handle C++ constructors.
11050 2018-05-18 Martin Liska <mliska@suse.cz>
11052 * passes.def: Remove a redundant pass.
11054 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
11057 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
11059 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11061 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
11063 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
11065 (armv2, armv2a, armv3, armv3m): Delete architectures.
11066 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
11067 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
11068 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
11070 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
11071 (*mulsidi3adddi): Likewise.
11072 (mulsidi3): Likewise.
11073 (*mulsidi3_nov6): Likewise.
11074 (umulsidi3): Likewise.
11075 (umulsidi3_nov6): Likewise.
11076 (umaddsidi4): Likewise.
11077 (*umulsidi3adddi): Likewise.
11078 (smulsi3_highpart): Likewise.
11079 (*smulsi3_highpart_nov6): Likewise.
11080 (umulsi3_highpart): Likewise.
11081 (*umulsi3_highpart_nov6): Likewise.
11082 * config/arm/arm.h (arm_arch3m): Delete.
11083 * config/arm/arm.c (arm_arch3m): Delete.
11084 (arm_option_override_internal): Update armv3-related comment.
11085 (arm_configure_build_target): Delete use of isa_bit_mode32.
11086 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
11087 (arm_rtx_costs_internal): Delete check of arm_arch3m.
11088 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
11089 (mulsa3): Likewise.
11090 (mulusa3): Likewise.
11091 * config/arm/arm-protos.h (arm_arch3m): Delete.
11092 * config/arm/arm-tables.opt: Regenerate.
11093 * config/arm/arm-tune.md: Likewise.
11094 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
11095 deleted architectures.
11097 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11099 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
11100 (armv5t, armv5te): New features.
11101 (ARMv5, ARMv5e): Delete fgroups.
11102 (ARMv5t, ARMv5te): Adjust for above changes.
11103 (ARMv6m): Likewise.
11104 (armv5, armv5e): Delete arches.
11105 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
11107 (*call_reg_arm): Likewise.
11108 (*call_value_reg_armv5): Likewise.
11109 (*call_value_reg_arm): Likewise.
11110 (*call_symbol): Likewise.
11111 (*call_value_symbol): Likewise.
11112 (*sibcall_insn): Likewise.
11113 (*sibcall_value_insn): Likewise.
11114 (clzsi2): Likewise.
11115 (prefetch): Likewise.
11116 (define_split and define_peephole2 dependent on arm_arch5):
11118 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
11120 (TARGET_ARM_QBIT): Likewise.
11121 (TARGET_DSP_MULTIPLY): Likewise.
11122 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
11123 (arm_arch5, arm_arch5e): Delete.
11124 (arm_arch5t, arm_arch5te): Declare.
11125 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
11126 (arm_arch5t): Declare.
11127 (arm_option_reconfigure_globals): Update for the above.
11128 (arm_options_perform_arch_sanity_checks): Update comment, replace
11129 use of arm_arch5 with arm_arch5t.
11130 (use_return_insn): Likewise.
11131 (arm_emit_call_insn): Likewise.
11132 (output_return_instruction): Likewise.
11133 (arm_final_prescan_insn): Likewise.
11134 (arm_coproc_builtin_available): Likewise.
11135 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
11136 arm_arch5e with arm_arch5t and arm_arch5te.
11137 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
11138 (arm_arch5t, arm_arch5te): Declare.
11139 * config/arm/arm-tables.opt: Regenerate.
11140 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
11141 * config/arm/t-multilib: Likewise.
11142 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
11143 instead of arm_arch5.
11144 (*call_reg_thumb1): Likewise.
11145 (*call_value_reg_thumb1_v5): Likewise.
11146 (*call_value_reg_thumb1): Likewise.
11147 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
11149 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
11151 2018-05-18 Martin Liska <mliska@suse.cz>
11153 PR gcov-profile/84846
11154 * doc/gcov.texi: Document -t option of gcov tool.
11156 2018-05-18 Martin Liska <mliska@suse.cz>
11158 PR gcov-profile/84846
11159 * gcov.c (print_usage): Add new -t option.
11160 (process_args): Handle the option.
11161 (generate_results): Use stdout as output when requested by
11164 2018-05-18 Martin Liska <mliska@suse.cz>
11166 PR gcov-profile/84846
11167 * coverage.c (coverage_init): Write PWD to .gcno file.
11168 * doc/gcov.texi: Document how working directory is printed.
11169 * gcov-dump.c (dump_gcov_file): Print PWD.
11170 * gcov.c (output_intermediate_file): Likewise.
11171 (read_graph_file): Read PWD string.
11172 (output_lines): Print PWD.
11174 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11176 PR middle-end/85817
11177 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
11178 for retval and return false if all args to phi are zero.
11180 2018-05-18 Richard Biener <rguenther@suse.de>
11182 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
11184 (evrp_dom_walker::before_dom_children): Call it.
11186 2018-05-18 Richard Biener <rguenther@suse.de>
11188 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
11189 results when processing array refs with variable index.
11191 2018-05-18 Toon Moene <toon@moene.org>
11193 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
11194 directly after that of -floop-interchange. Indicate that both
11195 options are enabled by default when specifying -O3.
11197 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11199 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
11200 iterator. Delete separate integer-mode vec_set<mode> expander.
11201 (aarch64_simd_vec_setv2di): Delete.
11202 (vec_setv2di): Delete.
11203 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
11204 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
11205 the "w, r" alternative.
11207 2018-05-18 Martin Liska <mliska@suse.cz>
11209 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
11210 * tree-pass.h (make_pass_lower_switch_O0): New function.
11211 * tree-switch-conversion.c (node_has_low_bound): Remove.
11212 (node_has_high_bound): Likewise.
11213 (node_is_bounded): Likewise.
11214 (class pass_lower_switch): Make it a template type and create
11216 (pass_lower_switch::execute): Add template argument.
11217 (make_pass_lower_switch): New function.
11218 (make_pass_lower_switch_O0): New function.
11219 (do_jump_if_equal): Remove.
11220 (emit_case_nodes): Simplify to just handle all 3 cases and leave
11221 all the hard work to tree optimization passes.
11223 2018-05-18 Martin Liska <mliska@suse.cz>
11225 * dbgcnt.c (limit_low): Renamed from limit.
11226 (limit_high): New variable.
11227 (dbg_cnt_is_enabled): Check for upper limit.
11228 (dbg_cnt): Adjust dumping.
11229 (dbg_cnt_set_limit_by_index): Add new argument for high
11231 (dbg_cnt_set_limit_by_name): Likewise.
11232 (dbg_cnt_process_single_pair): Parse new format.
11233 (dbg_cnt_process_opt): Use strtok.
11234 (dbg_cnt_list_all_counters): Remove 'value' and add
11236 * doc/invoke.texi: Document changes.
11238 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
11240 * doc/sourcebuild.texi (scalar_all_fma): Document.
11241 * tree.def (FMA_EXPR): Delete.
11242 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
11243 * internal-fn.c (ternary_direct): New macro.
11244 (expand_ternary_optab_fn): Likewise.
11245 (direct_ternary_optab_supported_p): Likewise.
11246 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
11247 * builtins.c (fold_builtin_fma): Delete.
11248 (fold_builtin_3): Don't call it.
11249 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
11250 * expr.c (expand_expr_real_2): Likewise.
11251 * fold-const.c (operand_equal_p): Likewise.
11252 (fold_ternary_loc): Likewise.
11253 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
11254 * gimple.c (DEFTREECODE): Likewise.
11255 * gimplify.c (gimplify_expr): Likewise.
11256 * optabs-tree.c (optab_for_tree_code): Likewise.
11257 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11258 * tree-eh.c (operation_could_trap_p): Likewise.
11259 (stmt_could_throw_1_p): Likewise.
11260 * tree-inline.c (estimate_operator_cost): Likewise.
11261 * tree-pretty-print.c (dump_generic_node): Likewise.
11262 (op_code_prio): Likewise.
11263 * tree-ssa-loop-im.c (stmt_cost): Likewise.
11264 * tree-ssa-operands.c (get_expr_operands): Likewise.
11265 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
11266 * fold-const-call.h (fold_fma): Delete.
11267 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
11268 CFN_FNMA and CFN_FNMS.
11269 (fold_fma): Delete.
11270 * genmatch.c (combined_fn): New enum.
11271 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
11272 (commutative_op): New function.
11273 (commutate): Use it. Handle more than 2 operands.
11274 (dt_operand::gen_gimple_expr): Use commutative_op.
11275 (parser::parse_expr): Allow :c to be used with non-binary
11276 operators if the commutative operand is known.
11277 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
11278 CFN_FMS, CFN_FNMA and CFN_FNMS.
11279 (backprop::process_assign_use): Remove FMA_EXPR handling.
11280 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
11281 (gen_hsa_fma): New function.
11282 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
11283 IFN_FNMA and IFN_FNMS.
11284 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
11285 * gimple-fold.h (follow_all_ssa_edges): Declare.
11286 * gimple-fold.c (follow_all_ssa_edges): New function.
11287 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
11288 gimple_build interface and use follow_all_ssa_edges to fold the result.
11289 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
11290 instead of checking for optabs directly.
11291 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
11292 rather than FMA_EXPRs.
11293 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
11294 call to IFN_FMA instead of an FMA_EXPR.
11296 2018-05-17 Jim Wilson <jimw@sifive.com>
11298 * expr.c (do_tablejump): When converting index to Pmode, if we have a
11299 sign extended promoted subreg, and the range does not have the sign bit
11300 set, then do a sign extend.
11302 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
11303 test, check for sign extended subreg and/or constant operands, and
11304 do a sign extend in that case.
11306 2018-05-17 Steve Ellcey <sellcey@cavium.com>
11308 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
11309 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
11311 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
11312 Change logics_shift_reg to logics_shift_imm.
11313 (thunderx2t99_fp_loadpair_basic): Delete.
11314 (thunderx2t99_fp_storepair_basic): Delete.
11315 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
11316 (thunderx2t99_asimd_polynomial): Delete.
11317 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
11318 and neon_fp_mul_d_scalar_q.
11319 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
11320 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
11321 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
11322 (thunderx2t99_asimd_lut): Add missing tbl types.
11323 (thunderx2t99_asimd_ext): Delete.
11324 (thunderx2t99_asimd_load1_1_mult): Delete.
11325 (thunderx2t99_asimd_load1_2_mult): Delete.
11326 (thunderx2t99_asimd_load1_ldp): New.
11327 (thunderx2t99_asimd_load1): New.
11328 (thunderx2t99_asimd_load2): Add missing *load2* types.
11329 (thunderx2t99_asimd_load3): New.
11330 (thunderx2t99_asimd_load4): New.
11331 (thunderx2t99_asimd_store1_1_mult): Delete.
11332 (thunderx2t99_asimd_store1_2_mult): Delete.
11333 (thunderx2t99_asimd_store2_mult): Delete.
11334 (thunderx2t99_asimd_store2_onelane): Delete.
11335 (thunderx2t99_asimd_store_stp): New.
11336 (thunderx2t99_asimd_store1): New.
11337 (thunderx2t99_asimd_store2): New.
11338 (thunderx2t99_asimd_store3): New.
11339 (thunderx2t99_asimd_store4): New.
11341 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
11343 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
11344 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
11346 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
11347 Segher Boessenkool <segher@kernel.crashing.org>
11350 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
11353 2018-05-17 Richard Biener <rguenther@suse.de>
11355 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
11356 for pruning loop and prune defs feeding only already visited PHIs.
11358 2018-05-17 Richard Biener <rguenther@suse.de>
11360 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
11362 2018-05-17 Bin Cheng <bin.cheng@arm.com>
11363 Richard Biener <rguenther@suse.de>
11365 PR tree-optimization/85793
11366 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
11367 for VMAT_ELEMENTWISE.
11369 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
11371 * internal-fn.h (lookup_internal_fn): Declare
11372 * internal-fn.c (lookup_internal_fn): New function.
11373 * gimple.c (gimple_build_call_from_tree): Handle calls to
11374 internal functions.
11375 * gimple-pretty-print.c (dump_gimple_call): Print "." before
11376 internal function names.
11377 * tree-pretty-print.c (dump_generic_node): Likewise.
11378 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
11380 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
11382 * gimple-fold.h (gimple_build): Make the function forms take
11383 combined_fn rather than built_in_function.
11384 (gimple_simplify): Likewise.
11385 * gimple-match-head.c (gimple_simplify): Likewise.
11386 * gimple-fold.c (gimple_build): Likewise.
11387 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
11388 rather than gimple_build_call_internal.
11389 (get_initial_defs_for_reduction): Likewise.
11390 (vect_create_epilog_for_reduction): Likewise.
11391 (vectorizable_live_operation): Likewise.
11393 2018-05-17 Martin Liska <mliska@suse.cz>
11395 * gimple-ssa-sprintf.c (format_directive): Do not use
11396 space in between 'G_' and '('.
11398 2018-05-17 Jakub Jelinek <jakub@redhat.com>
11401 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
11402 even if the mask is not all ones.
11405 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
11407 (ix86_gimple_fold_builtin): Likewise.
11410 * config/i386/i386.c: Include tree-vector-builder.h.
11411 (ix86_vector_shift_count): New function.
11412 (ix86_fold_builtin): Fold shift builtins by scalar count.
11413 (ix86_gimple_fold_builtin): Likewise.
11415 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
11416 _mm512_setzero): New intrinsics.
11418 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
11419 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11421 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
11422 code generation for cases where splatting a value is not useful.
11423 * simplify-rtx.c (simplify_ternary_operation): Simplify
11424 vec_merge across a vec_duplicate and a paradoxical subreg forming
11425 a vector mode to a vec_concat.
11427 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
11429 * config.gcc: Support "goldmont-plus".
11430 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
11432 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
11433 PROCESSOR_GOLDMONT_PLUS.
11434 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
11435 (processor_target_table): Add "goldmont-plus".
11436 (PTA_GOLDMONT_PLUS): Define.
11437 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
11438 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
11439 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
11440 (fold_builtin_cpu): Add "goldmont-plus".
11441 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
11442 (ix86_option_override_internal): Add "goldmont-plus".
11443 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
11444 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
11445 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
11446 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
11448 2018-05-17 Richard Biener <rguenther@suse.de>
11450 PR tree-optimization/85757
11451 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
11452 remove defs that only feed that PHI from further processing.
11454 2018-05-16 Jim Wilson <jimw@sifive.com>
11456 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
11458 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
11460 2018-05-16 Mark Wielaard <mark@klomp.org>
11462 * dwarf2out.c (count_index_strings): New function.
11463 (output_indirect_strings): Call count_index_strings and generate
11464 header for dwarf_version >= 5.
11466 2018-05-16 Mark Wielaard <mark@klomp.org>
11468 * dwarf2out.c (dwarf_FORM): New function.
11469 (set_indirect_string): Use dwarf_FORM.
11470 (reset_indirect_string): Likewise.
11471 (size_of_die): Likewise.
11472 (value_format): Likewise.
11473 (output_die): Likewise.
11474 (add_skeleton_AT_string): Likewise.
11475 (output_macinfo_op): Likewise.
11476 (index_string): Likewise.
11477 (output_index_string_offset): Likewise.
11478 (output_index_string): Likewise.
11479 (count_index_strings): Likewise.
11481 2018-05-16 Carl Love <cel@us.ibm.com>
11483 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
11484 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
11486 2018-05-16 Martin Jambor <mjambor@suse.cz>
11488 * ipa-prop.c (ipa_free_all_edge_args): Remove.
11489 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
11491 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
11493 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
11494 (fnma<mode>4): Likewise.
11495 (fms<mode>4): Likewise.
11496 (fnms<mode>4): Likewise.
11497 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
11498 (aarch64_fnma<mode>4): Likewise.
11499 (aarch64_fms<mode>4): Likewise.
11500 (aarch64_fnms<mode>4): Likewise.
11501 (aarch64_fnmadd<mode>4): Likewise.
11503 2018-05-16 Jason Merrill <jason@redhat.com>
11505 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
11507 2018-05-16 Richard Biener <rguenther@suse.de>
11509 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
11510 (dump_stmt_cost): Declare.
11511 (add_stmt_cost): Dump cost we add.
11512 (add_stmt_costs): New function.
11513 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
11514 No longer exported.
11515 (vect_analyze_stmt): Adjust prototype.
11516 (vectorizable_condition): Likewise.
11517 (vectorizable_live_operation): Likewise.
11518 (vectorizable_reduction): Likewise.
11519 (vectorizable_induction): Likewise.
11520 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
11521 cost vector to pass to vectorizable_ and record afterwards.
11522 (vect_model_reduction_cost): Take cost vector argument and adjust.
11523 (vect_model_induction_cost): Likewise.
11524 (vectorizable_reduction): Likewise.
11525 (vectorizable_induction): Likewise.
11526 (vectorizable_live_operation): Likewise.
11527 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
11528 SLP_TREE_NUMBER_OF_VEC_STMTS.
11529 (vect_analyze_slp_cost_1): Remove.
11530 (vect_analyze_slp_cost): Likewise.
11531 (vect_slp_analyze_node_operations): Take visited args and
11532 a target cost vector. Avoid processing already visited stmt sets.
11533 (vect_slp_analyze_operations): Use a local cost vector to gather
11534 costs and register those of non-discarded instances.
11535 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
11536 (vect_schedule_slp_instance): Remove copying of
11537 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
11539 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
11540 adding cost. Record cost entry location.
11541 (vect_prologue_cost_for_slp_op): Function to compute cost of
11542 a constant or invariant generated for SLP vect in the prologue,
11543 split out from vect_analyze_slp_cost_1.
11544 (vect_model_simple_cost): Make static. Adjust for SLP costing.
11545 (vect_model_promotion_demotion_cost): Likewise.
11546 (vect_model_store_cost): Likewise, make static.
11547 (vect_model_load_cost): Likewise.
11548 (vectorizable_bswap): Add cost vector arg and adjust.
11549 (vectorizable_call): Likewise.
11550 (vectorizable_simd_clone_call): Likewise.
11551 (vectorizable_conversion): Likewise.
11552 (vectorizable_assignment): Likewise.
11553 (vectorizable_shift): Likewise.
11554 (vectorizable_operation): Likewise.
11555 (vectorizable_store): Likewise.
11556 (vectorizable_load): Likewise.
11557 (vectorizable_condition): Likewise.
11558 (vectorizable_comparison): Likewise.
11559 (can_vectorize_live_stmts): Likewise.
11560 (vect_analyze_stmt): Likewise.
11561 (vect_transform_stmt): Adjust calls to vectorizable_*.
11562 * tree-vectorizer.c: Include gimple-pretty-print.h.
11563 (dump_stmt_cost): New function.
11565 2018-05-16 Richard Biener <rguenther@suse.de>
11567 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
11568 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
11569 * tree-ssa-dse.c: Include tree-ssa-loop.h.
11570 (check_name): New callback.
11571 (dse_classify_store): Track cycles via a visited bitmap of PHI
11572 defs and simplify handling of in-loop and across loop dead stores
11573 and properly fail for loop-variant refs. Handle byte-tracking with
11574 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
11577 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
11579 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
11580 (vect_get_mask_type_for_stmt): Likewise.
11581 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
11583 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
11584 to determine the statement's vector type and the vector type that
11585 should be used for calculating nunits. Deal with cases in which
11586 the type has to be deferred.
11587 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
11588 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
11589 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
11590 (vect_determine_vf_for_stmt): New functions, split out from...
11591 (vect_determine_vectorization_factor): ...here.
11592 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
11593 (vect_get_mask_type_for_stmt): New functions, split out from
11594 vect_determine_vectorization_factor.
11596 2018-05-16 Richard Biener <rguenther@suse.de>
11598 * tree-cfg.c (verify_gimple_assign_ternary): Properly
11599 verify the [VEC_]COND_EXPR embedded comparison.
11601 2018-05-15 Martin Sebor <msebor@redhat.com>
11603 PR tree-optimization/85753
11604 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
11605 RECORD_TYPE in addition to ARRAY_TYPE.
11607 2018-05-15 Martin Sebor <msebor@redhat.com>
11609 PR middle-end/85643
11610 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
11612 2018-05-15 Richard Biener <rguenther@suse.de>
11614 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
11615 add by_clobber_p one. Change algorithm to collect all defs
11616 representing uses we need to walk and try reducing them to
11617 a single one before failing.
11618 (dse_dom_walker::dse_optimize_stmt): Adjust.
11620 2018-05-13 Mark Wielaard <mark@klomp.org>
11622 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
11623 (size_of_loc_descr): Likewise.
11624 (output_loc_operands): Likewise.
11625 (output_loc_operands_raw): Likewise.
11626 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
11627 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
11628 (hash_loc_operands): Likewise.
11629 (compare_loc_operands): Likewise.
11631 2018-05-14 Mark Wielaard <mark@klomp.org>
11633 * dwarf2out.c (count_index_addrs): New function.
11634 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
11636 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11638 PR tree-optimization/83648
11639 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
11640 return value as malloc candidate.
11642 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11645 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
11646 param as true in call to suggest_attribute.
11648 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
11650 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
11651 -mreadonly-in-sdata.
11653 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11655 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
11657 (aarch64_crypto_aesd_fused): Likewise.
11659 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
11661 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
11662 (movsi_aarch64): Likewise.
11663 (load_pairsi): Likewise.
11664 (load_pairdi): Likewise.
11665 (store_pairsi): Likewise.
11666 (store_pairdi): Likewise.
11667 (load_pairsf): Likewise.
11668 (load_pairdf): Likewise.
11669 (store_pairsf): Likewise.
11670 (store_pairdf): Likewise.
11671 (zero_extend): Likewise.
11672 (trunc): Swap alternatives.
11673 (fcvt_target): Add '?' to prefer w over r.
11675 2018-05-14 Jakub Jelinek <jakub@redhat.com>
11678 * config/i386/i386.md: Disallow non-commutative arithmetics in
11679 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
11680 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
11681 in the peephole2 before it.
11683 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
11685 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
11686 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
11687 (ix86_handle_option): Handle -mcldemote.
11688 * config.gcc: New header.
11689 * config/i386/cldemoteintrin.h: New file.
11690 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
11691 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
11693 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
11694 OPTION_MASK_ISA_CLDEMOTE.
11695 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
11696 (ix86_valid_target_attribute_inner_p): Ditto.
11697 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
11698 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
11699 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
11700 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
11701 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
11703 * config/i386/i386.opt: Add -mcldemote.
11704 * config/i386/x86intrin.h: New header.
11705 * doc/invoke.texi: Add -mcldemote.
11707 2018-05-14 Richard Biener <rguenther@suse.de>
11709 * doc/match-and-simplify.texi: Adjust :s documentation.
11711 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
11713 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
11714 intended memcpy size.
11715 (REORDER_45): Likewise.
11717 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
11719 * sort.cc: New file.
11720 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
11721 * vec.c (qsort_chk): Use gcc_qsort.
11722 * Makefile.in (OBJS-libcommon): Add sort.o.
11723 (build/sort.o): New target. Use it...
11724 (BUILD_RTL): ... here, and...
11725 (build/gencfn-macros): ... here, and...
11726 (build/genmatch): ... here.
11728 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
11729 Chung-Ju Wu <jasonwucj@gmail.com>
11731 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
11732 * config/nds32/nds32-graywolf.md: New file.
11733 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
11734 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
11736 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
11737 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
11738 * config/nds32/nds32.md (pipeline_model): Add graywolf.
11739 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
11740 * config/nds32/pipelines.md: Include n15 settings.
11742 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
11743 Chung-Ju Wu <jasonwucj@gmail.com>
11745 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
11746 * config/nds32/nds32-n13.md: New file.
11747 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
11748 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
11750 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
11751 * config/nds32/nds32.md (pipeline_model): Add n13.
11752 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
11753 * config/nds32/pipelines.md: Include n13 settings.
11755 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
11756 Chung-Ju Wu <jasonwucj@gmail.com>
11758 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
11759 * config/nds32/nds32-n10.md: New file.
11760 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
11761 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
11763 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
11764 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
11765 * config/nds32/nds32.md (pipeline_model): Add n10.
11766 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
11767 * config/nds32/pipelines.md: Include n10 settings.
11769 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
11770 Kito Cheng <kito.cheng@gmail.com>
11771 Chung-Ju Wu <jasonwucj@gmail.com>
11773 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
11774 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
11775 Add enum values for DSP extension instructions.
11776 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
11778 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
11779 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
11780 New code iterators.
11781 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
11782 * config/nds32/nds32-dspext.md: New file for DSP implementation.
11783 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
11784 * config/nds32/nds32-intrinsic.md: Likewise.
11785 * config/nds32/nds32_intrinsic.h: Likewise.
11786 * config/nds32/nds32-md-auxiliary.c: Likewise.
11787 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
11788 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
11789 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
11790 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
11791 * config/nds32/nds32-protos.h: New declarations for DSP extension.
11792 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
11793 TYPE_DMAC in switch statement.
11794 * config/nds32/nds32.c: New checking and implementation for DSP
11795 extension instructions.
11796 * config/nds32/nds32.h: Likewise.
11797 * config/nds32/nds32.md: Likewise.
11798 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
11799 * config/nds32/predicates.md: Implement new predicates for DSP
11802 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
11804 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
11805 Reformat alternatives and attributes so it is easier to identify
11806 which constraints/attributes go with which instruction.
11807 (mov<mode>_hardfloat32, FMOVE64): Likewise.
11808 (mov<mode>_softfloat32, FMOVE64): Likewise.
11809 (mov<mode>_hardfloat64, FMOVE64): Likewise.
11810 (mov<mode>_softfloat64, FMOVE64): Likewise.
11812 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
11814 * doc/extend.texi (PowerPC Built-in Functions): Rename this
11816 (Basic PowerPC Built-in Functions): The new name of the
11817 subsection previously known as "PowerPC Built-in Functions".
11818 (Basic PowerPC Built-in Functions Available on all Configurations):
11820 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
11821 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
11822 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
11823 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
11825 2018-05-11 Martin Jambor <mjambor@suse.cz>
11828 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
11831 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
11834 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
11836 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
11838 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
11839 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
11840 (ix86_handle_option): Handle -mwaitpkg.
11841 * config.gcc: New header.
11842 * config/i386/cpuid.h (bit_WAITPKG): New bit.
11843 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
11844 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
11846 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
11847 OPTION_MASK_ISA_WAITPKG.
11848 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
11849 (ix86_option_override_internal): Add PTA_WAITPKG.
11850 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
11851 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
11852 IX86_BUILTIN_TPAUSE.
11853 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
11854 __builtin_ia32_umwait and __builtin_ia32_tpause.
11855 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
11856 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
11857 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
11858 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
11859 UNSPECV_TPAUSE): New.
11860 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
11861 * config/i386/i386.opt: Add -mwaitpkg.
11862 * config/i386/waitpkgintrin.h: New file.
11863 * config/i386/x86intrin.h: New header.
11864 * doc/invoke.texi: Add -mwaitpkg.
11866 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
11869 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
11871 (cortex-m0): Use armv6s-m isa.
11872 (cortex-m0plus): Likewise.
11873 (cortex-m1): Likewise.
11874 (cortex-m0.small-multiply): Likewise.
11875 (cortex-m0plus.small-multiply): Likewise.
11876 (cortex-m1.small-multiply): Likewise.
11878 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
11879 Jakub Jelinek <jakub@redhat.com>
11881 PR tree-optimization/85692
11882 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
11883 source permute as well.
11885 2018-05-11 Martin Liska <mliska@suse.cz>
11888 * doc/extend.texi: Document LLVM style format for no_sanitize
11891 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
11893 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
11894 mode_supports_vsx_dform_quad to mode_supports_dq_form.
11895 (mode_supports_vsx_dform_quad): Likewise.
11896 (mode_supports_vmx_dform): Move these functions to be next to the
11897 other mode_supports functions.
11898 (mode_supports_dq_form): Likewise.
11899 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
11900 mode_supports_dq_form.
11901 (reg_offset_addressing_ok_p): Likewise.
11902 (offsettable_ok_by_alignment): Likewise.
11903 (rs6000_legitimate_offset_address_p): Likewise.
11904 (legitimate_lo_sum_address_p): Likewise.
11905 (rs6000_legitimize_address): Likewise.
11906 (rs6000_legitimize_reload_address): Likewise.
11907 (rs6000_secondary_reload_inner): Likewise.
11908 (rs6000_preferred_reload_class): Likewise.
11909 (rs6000_output_move_128bit): Likewise.
11911 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
11913 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
11914 Generate SImode target register for null target.
11915 <case IX86_BUILTIN_XGETBV>: Ditto.
11916 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
11917 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
11919 2018-05-10 Carl Love <cel@us.ibm.com>
11921 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
11922 dcbtt and dcbtstt if operands[2] is 0.
11924 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
11927 * config/i386/sse.md (usadv64qi): New expander.
11929 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
11931 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
11932 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
11933 -maltivec=be support.
11934 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
11935 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
11936 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
11937 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
11938 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
11939 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
11940 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
11941 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
11942 altivec_vsumsws): Adjust.
11943 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
11944 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
11945 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
11947 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
11948 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
11949 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
11950 (altivec_lve<VI_char>x): Delete expand.
11951 (*altivec_lve<VI_char>x_internal): Rename to...
11952 (altivec_lve<VI_char>x): ... this.
11953 (altivec_lvxl_<mode>): Delete expand.
11954 (*altivec_lvxl_<mode>_internal): Rename to ...
11955 (altivec_lvxl_<mode>): ... this.
11956 (altivec_stvxl_<mode>): Delete expand.
11957 (*altivec_stvxl_<mode>_internal): Rename to ...
11958 (altivec_stvxl_<mode>): ... this.
11959 (altivec_stve<VI_char>x): Delete expand.
11960 (*altivec_stve<VI_char>x_internal): Rename to ...
11961 (altivec_stve<VI_char>x): ... this.
11962 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
11963 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
11964 reduc_plus_scal_<mode>): Adjust.
11965 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
11967 (rs6000_cpu_cpp_builtins): Adjust.
11968 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
11969 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
11970 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
11971 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
11972 -maltivec=be support.
11973 (rs6000_split_vec_extract_var): Adjust.
11974 (rs6000_split_v4si_init): Adjust.
11975 (swap_selector_for_mode): Delete.
11976 (altivec_expand_lvx_be, altivec_expand_stvx_be,
11977 altivec_expand_stvex_be): Delete.
11978 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
11979 -maltivec=be support.
11980 (rs6000_gimple_fold_builtin): Ditto.
11981 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
11983 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
11984 (TARGET_DIRECT_MOVE_64BIT): Adjust.
11985 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
11986 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
11987 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
11988 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
11989 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
11990 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
11991 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
11992 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
11993 anonymous split): Adjust.
11994 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
11995 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
11997 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
11999 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
12000 when --with-gxx-include-dir is also specified.
12001 * configure: Regenerate.
12003 2018-05-09 Jim Wilson <jimw@sifive.com>
12006 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
12007 * config/riscv/t-withmultilib: New.
12008 * config/riscv/withmultilib.h: New.
12009 * doc/install.texi: Document RISC-V --with-multilib-list support.
12011 2018-05-09 Richard Biener <rguenther@suse.de>
12013 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
12015 (vect_bb_vectorization_profitable_p): Adjust. Compute
12016 actual scalar cost using the cost vector and the add_stmt_cost
12019 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
12021 PR rtl-optimization/85645
12022 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
12023 in the REG_CFA_REGISTER note for LR, don't leave it empty.
12025 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
12027 PR rtl-optimization/85645
12028 * shrink-wrap.c (spread_components): Return a boolean saying if
12029 anything was changed.
12030 (try_shrink_wrapping_separate): Iterate spread_components until
12031 nothing changes anymore.
12033 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
12035 PR rtl-optimization/85645
12036 * regrename.c (build_def_use): Also kill the chains that include the
12037 destination of a REG_CFA_REGISTER note.
12039 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
12041 PR rtl-optimization/85645
12042 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
12043 insn that has a REG_CFA_REGISTER note.
12045 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
12047 * cfgexpand.c (expand_clobber): New function.
12048 (expand_gimple_stmt_1): Use it.
12049 * tree-vect-stmts.c (vect_clobber_variable): New function,
12051 (vectorizable_simd_clone_call): ...here.
12052 (vectorizable_store): Emit a clobber either side of an
12053 IFN_STORE_LANES sequence.
12054 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
12056 2018-05-09 Tom de Vries <tom@codesourcery.com>
12059 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
12060 (define_insn "trap_if_false"): Add exit after trap.
12062 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
12064 PR rtl-optimization/85638
12065 * bb-reorder.c: Include common/common-target.h.
12066 (create_forwarder_block): New function extracted from...
12067 (fix_up_crossing_landing_pad): ...here. Rename into...
12068 (dw2_fix_up_crossing_landing_pad): ...this.
12069 (sjlj_fix_up_crossing_landing_pad): New function.
12070 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
12071 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
12072 from both partitions and exit the loop after one iteration.
12074 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
12077 * doc/extend.texi (PowerPC Built-in Functions): Rename this
12079 (Basic PowerPC Built-in Functions): The new name of the
12080 subsection previously known as "PowerPC Built-in Functions".
12081 (Basic PowerPC Built-in Functions Available on all Configurations):
12083 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
12085 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
12087 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
12089 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
12092 2018-05-08 Jim Wilson <jimw@sifive.com>
12094 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
12095 (LD_EMUL_SUFFIX): New.
12096 (LINK_SPEC): Use it.
12098 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
12100 * doc/extend.texi (PowerPC Built-in Functions): Rename this
12102 (Basic PowerPC Built-in Functions): The new name of the
12103 subsection previously known as "PowerPC Built-in Functions".
12104 (Basic PowerPC Built-in Functions Available on all Configurations):
12106 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
12108 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
12110 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
12112 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
12115 2018-05-08 Jakub Jelinek <jakub@redhat.com>
12118 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
12119 after cmpelim optimization.
12121 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
12123 * config.gcc: Support "goldmont".
12124 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
12125 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
12126 PROCESSOR_GOLDMONT.
12127 * config/i386/i386.c (m_GOLDMONT): Define.
12128 (processor_target_table): Add "goldmont".
12129 (PTA_GOLDMONT): Define.
12130 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
12131 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
12132 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
12133 (fold_builtin_cpu): Add "goldmont".
12134 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
12135 (ix86_option_override_internal): Add "goldmont".
12136 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
12137 (processor_type): Add PROCESSOR_GOLDMONT.
12138 * config/i386/i386.md: Add CPU "glm".
12139 * config/i386/glm.md: New file.
12140 * config/i386/x86-tune.def: Add m_GOLDMONT.
12141 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
12143 2018-05-08 Jakub Jelinek <jakub@redhat.com>
12146 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
12148 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
12149 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
12150 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
12153 * config/i386/i386.c (ix86_fold_builtin): Handle
12154 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
12157 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
12158 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
12160 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
12163 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
12164 (check_arch): Likewise.
12165 (check_fpu): Return the result rather than printing it.
12166 (end arch): Fix operator precedence.
12167 (end cpu): Likewise.
12168 (END): Print the result from check_fpu.
12170 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
12171 Alan Hayward <alan.hayward@arm.com>
12172 David Sherwood <david.sherwood@arm.com>
12174 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
12175 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
12176 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
12177 (*fcmuo<mode>_and): New patterns.
12179 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
12181 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
12182 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
12183 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
12184 (cmp_op, sve_imm_con): New code attributes.
12185 (SVE_COND_INT_CMP, imm_con): Delete.
12186 (cmp_op): Remove above unspecs from int attribute.
12187 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
12189 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
12190 comparison-specific unspecs.
12191 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
12192 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
12193 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
12194 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
12195 (*vec_fcm<cmp_op><mode>): Rename to...
12196 (*fcm<cmp_op><mode>): ...this and adjust likewise.
12197 (*vec_fcmuo<mode>): Rename to...
12198 (*fcmuo<mode>): ...this and adjust likewise.
12199 (*pred_fcm<cmp_op><mode>): New pattern.
12200 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
12201 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
12203 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
12205 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
12206 (aarch64_emit_sve_predicated_cond): New function.
12207 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
12208 (aarch64_emit_unspec_cond_or): Replace with...
12209 (aarch64_emit_sve_or_conds): ...this new function. Use
12210 aarch64_emit_sve_ptrue_op for the individual comparisons and
12211 aarch64_emit_binop to OR them together.
12212 (aarch64_emit_inverted_unspec_cond): Replace with...
12213 (aarch64_emit_sve_inverted_cond): ...this new function. Use
12214 aarch64_emit_sve_ptrue_op for the comparison and
12215 aarch64_emit_unop to invert the result.
12216 (aarch64_expand_sve_vec_cmp_float): Update after the above
12217 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
12219 2018-05-07 Nathan Sidwell <nathan@acm.org>
12221 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
12222 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
12223 (Backwards Compatibility): Likewise.
12225 2018-05-07 Luis Machado <luis.machado@linaro.org>
12229 2018-05-07 Luis Machado <luis.machado@linaro.org>
12231 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
12232 <prefetch_dynamic_strides>: New const bool field.
12233 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
12234 prefetch_dynamic_strides.
12235 (exynosm1_prefetch_tune): Likewise.
12236 (thunderxt88_prefetch_tune): Likewise.
12237 (thunderx_prefetch_tune): Likewise.
12238 (thunderx2t99_prefetch_tune): Likewise.
12239 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
12241 (aarch64_override_options_internal): Update to set
12242 PARAM_PREFETCH_DYNAMIC_STRIDES.
12243 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
12244 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
12245 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
12246 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
12247 prefetch-dynamic-strides setting.
12249 2018-05-07 Luis Machado <luis.machado@linaro.org>
12251 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
12252 <minimum_stride>: New const int field.
12253 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
12254 minimum_stride field.
12255 (exynosm1_prefetch_tune): Likewise.
12256 (thunderxt88_prefetch_tune): Likewise.
12257 (thunderx_prefetch_tune): Likewise.
12258 (thunderx2t99_prefetch_tune): Likewise.
12259 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
12260 (aarch64_override_options_internal): Update to set
12261 PARAM_PREFETCH_MINIMUM_STRIDE.
12262 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
12263 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
12264 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
12265 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
12266 stride is constant and is below the minimum stride threshold.
12268 2018-05-07 Luis Machado <luis.machado@linaro.org>
12270 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
12273 2018-05-07 Luis Machado <luis.machado@linaro.org>
12275 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
12276 <prefetch_dynamic_strides>: New const bool field.
12277 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
12278 prefetch_dynamic_strides.
12279 (exynosm1_prefetch_tune): Likewise.
12280 (thunderxt88_prefetch_tune): Likewise.
12281 (thunderx_prefetch_tune): Likewise.
12282 (thunderx2t99_prefetch_tune): Likewise.
12283 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
12285 (aarch64_override_options_internal): Update to set
12286 PARAM_PREFETCH_DYNAMIC_STRIDES.
12287 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
12288 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
12289 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
12290 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
12291 prefetch-dynamic-strides setting.
12293 2018-05-07 Luis Machado <luis.machado@linaro.org>
12295 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
12296 <minimum_stride>: New const int field.
12297 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
12298 minimum_stride field.
12299 (exynosm1_prefetch_tune): Likewise.
12300 (thunderxt88_prefetch_tune): Likewise.
12301 (thunderx_prefetch_tune): Likewise.
12302 (thunderx2t99_prefetch_tune): Likewise.
12303 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
12304 (aarch64_override_options_internal): Update to set
12305 PARAM_PREFETCH_MINIMUM_STRIDE.
12306 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
12307 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
12308 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
12309 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
12310 stride is constant and is below the minimum stride threshold.
12312 2018-05-06 Jakub Jelinek <jakub@redhat.com>
12315 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
12316 the type is addressable. Don't force op into register if it has
12319 2018-05-05 Roland McGrath <mcgrathr@google.com>
12322 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
12323 any section for which we don't know a specific type it should have,
12324 regardless of name. Previously this was done only for the exact
12325 names ".init_array", ".fini_array", and ".preinit_array".
12326 (default_elf_asm_named_section): Add comment about
12327 relationship with default_section_type_flags and SECTION_NOTYPE.
12328 (get_section): Don't consider it a type conflict if one side has
12329 SECTION_NOTYPE and the other doesn't, as long as neither has the
12330 SECTION_BSS et al used in the default_section_type_flags logic.
12332 2018-05-05 Tom de Vries <tom@codesourcery.com>
12335 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
12336 (workaround_barsyncs): New function.
12337 (nvptx_reorg): Use workaround_barsyncs.
12338 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
12339 (define_expand "nvptx_membar_cta"): New define_expand.
12340 (define_insn "*nvptx_membar_cta"): New insn.
12342 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
12344 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
12345 To improve optimization opportunities.
12346 * builtin-types.def: The new needed builtin types for the above.
12348 2018-05-04 Richard Biener <rguenther@suse.de>
12350 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
12351 * gimple-ssa-store-merging.c
12352 (imm_store_chain_info::output_merged_store): Remove redundant create,
12353 release split_store vector contents on failure.
12354 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
12355 scalar stmt vector on cache hit.
12357 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
12359 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
12361 * config.gcc (powerpc-xilinx-eabi*): Remove.
12362 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
12364 (fusion_addis_mem_combo_load): Ditto.
12365 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
12367 (rs6000_cpu_cpp_builtins): Ditto.
12368 * config/rs6000/rs6000-linux.c
12369 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
12370 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
12371 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
12373 (rs6000_setup_reg_addr_masks): Ditto.
12374 (rs6000_init_hard_regno_mode_ok): Ditto.
12375 (rs6000_option_override_internal): Ditto.
12376 (legitimate_lo_sum_address_p): Ditto.
12377 (rs6000_legitimize_address): Ditto.
12378 (rs6000_legitimize_reload_address): Ditto.
12379 (rs6000_legitimate_address_p): Ditto.
12380 (abi_v4_pass_in_fpr): Ditto.
12381 (setup_incoming_varargs): Ditto.
12382 (rs6000_gimplify_va_arg): Ditto.
12383 (rs6000_split_multireg_move): Ditto.
12384 (rs6000_savres_strategy): Ditto.
12385 (rs6000_emit_prologue_components): Ditto.
12386 (rs6000_emit_epilogue_components): Ditto.
12387 (rs6000_emit_prologue): Ditto.
12388 (rs6000_emit_epilogue): Ditto.
12389 (rs6000_elf_file_end): Ditto.
12390 (rs6000_function_value): Ditto.
12391 (rs6000_libcall_value): Ditto.
12392 * config/rs6000/rs6000.h: Ditto.
12393 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
12394 (TARGET_MINMAX): ... this. New.
12395 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
12396 * config/rs6000/rs6000.md: Remove Xilinx FP support.
12397 (*movsi_internal1_single): Delete.
12398 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
12399 mfpu=, mxilinx-fpu): Delete.
12400 * config/rs6000/singlefp.h: Delete.
12401 * config/rs6000/sysv4.h: Remove Xilinx FP support.
12402 * config/rs6000/t-rs6000: Ditto.
12403 * config/rs6000/t-xilinx: Delete.
12404 * config/rs6000/titan.md: Adjust for fp_type removal.
12405 * config/rs6000/vsx.md: Remove Xilinx FP support.
12406 (VStype_simple): Delete.
12407 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
12408 * config/rs6000/xfpu.h: Delete.
12409 * config/rs6000/xfpu.md: Delete.
12410 * config/rs6000/xilinx.h: Delete.
12411 * config/rs6000/xilinx.opt: Delete.
12412 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
12413 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
12415 2018-05-04 Tom de Vries <tom@codesourcery.com>
12418 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
12421 2018-05-04 Richard Biener <rguenther@suse.de>
12423 PR middle-end/85627
12424 * tree-complex.c (update_complex_assignment): We are always in SSA form.
12425 (expand_complex_div_wide): Likewise.
12426 (expand_complex_operations_1): Likewise.
12427 (expand_complex_libcall): Preserve EH info of the original stmt.
12428 (tree_lower_complex): Handle removed blocks.
12429 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
12430 on complex multiplication and division libcall builtins.
12432 2018-05-04 Richard Biener <rguenther@suse.de>
12434 PR middle-end/85574
12435 * fold-const.c (negate_expr_p): Restrict negation of operand
12436 zero of a division to when we know that can happen without
12438 (fold_negate_expr_1): Likewise.
12440 2018-05-04 Jakub Jelinek <jakub@redhat.com>
12443 * real.h (real_nextafter): Declare.
12444 * real.c (real_nextafter): New function.
12445 * fold-const-call.c (fold_const_nextafter): New function.
12446 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
12447 CASE_CFN_NEXTTOWARD.
12448 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
12449 even when arg1_mode is different from arg0_mode.
12451 2018-05-03 Nathan Sidwell <nathan@acm.org>
12453 * doc/extend.texi (Deprecated Features): Remove
12454 -ffriend-injection.
12455 (Backwards Compatibility): Likewise.
12456 * doc/invoke.texi (C++ Language Options): Likewise.
12457 (C++ Dialect Options): Likewise.
12459 2018-05-03 Jakub Jelinek <jakub@redhat.com>
12462 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
12463 _mm512_mask_mullox_epi64): New intrinsics.
12465 2018-05-03 Tom de Vries <tom@codesourcery.com>
12468 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
12469 dump files): Add offload-tree.
12471 2018-05-03 Richard Biener <rguenther@suse.de>
12473 PR tree-optimization/85615
12474 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
12475 to loops not nested in BBs loop father to avoid creating multi-entry
12478 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12480 PR tree-optimization/70291
12481 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
12482 arguments. Change return type to tree. Emit libcall as a new
12483 statement rather than replacing existing one when inplace_p is true.
12484 (expand_complex_multiplication_components): New function.
12485 (expand_complex_multiplication): Expand floating-point complex
12486 multiplication using the above.
12487 (expand_complex_division): Rename inner_type parameter to type.
12488 Update expand_complex_libcall call-site.
12489 (expand_complex_operations_1): Update expand_complex_multiplication
12490 and expand_complex_division call-sites.
12492 2018-05-02 Jakub Jelinek <jakub@redhat.com>
12495 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
12496 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
12497 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
12498 the highest significant bit of the shift count mask is clear. In
12499 check whether and[sq]i3 is needed verify that all significant bits
12500 of the shift count other than the highest are set.
12502 2018-05-02 Tom de Vries <tom@codesourcery.com>
12505 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
12506 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
12507 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
12508 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
12509 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
12510 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
12511 __builtin_goacc_parlevel_size.
12513 2018-05-02 Richard Biener <rguenther@suse.de>
12515 PR tree-optimization/85597
12516 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
12517 do not use split vect_get_vec_defs call but call vect_get_slp_defs
12520 2018-05-02 Tom de Vries <tom@codesourcery.com>
12523 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
12524 dump files): Add ltrans-tree.
12526 2018-05-02 Tom de Vries <tom@codesourcery.com>
12529 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
12530 dump files): Add wpa-ipa.
12532 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
12534 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
12535 powerpc*-*-linux*paired* target.
12536 * config/rs6000/750cl.h: Delete.
12537 * config/rs6000/paired.h: Delete.
12538 * config/rs6000/paired.md: Delete.
12539 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
12541 * config/rs6000/rs6000-builtin.def: Remove paired float support.
12542 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
12543 comment. Remove paired float support.
12544 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
12545 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
12547 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
12548 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
12550 * config/rs6000/rs6000.c: Remove paired float support.
12551 (paired_expand_vector_init, paired_expand_vector_move,
12552 paired_emit_vector_compare, paired_emit_vector_cond_expr,
12553 (paired_expand_lv_builtin, paired_expand_stv_builtin,
12554 paired_expand_builtin, paired_expand_predicate_builtin,
12555 paired_init_builtins): Delete.
12556 * config/rs6000/rs6000.h: Remove paired float support.
12557 * config/rs6000/rs6000.md: Remove paired float support.
12558 (move_from_CR_ov_bit): Delete.
12559 * config/rs6000/rs6000.opt (mpaired): Delete.
12560 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
12561 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
12563 2018-05-02 Richard Biener <rguenther@suse.de>
12565 PR middle-end/85567
12566 * gimplify.c (gimplify_save_expr): When in SSA form allow
12567 SAVE_EXPRs to compute to SSA vars.
12569 2018-05-02 Jakub Jelinek <jakub@redhat.com>
12572 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
12573 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
12574 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
12575 clobber operands[2], instead use a new pseudo. Formatting fixes.
12577 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
12579 PR tree-optimization/85586
12580 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
12581 exit early for statements in the same group if the accesses are
12584 2018-05-02 Tom de Vries <tom@codesourcery.com>
12587 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
12590 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
12592 PR tree-optimization/85143
12593 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
12595 2018-05-01 Tom de Vries <tom@codesourcery.com>
12598 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
12599 not found" error message.
12601 2018-05-01 Tom de Vries <tom@codesourcery.com>
12604 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
12605 * vec.c (test_ordered_remove_if): New function.
12606 (vec_c_tests): Call test_ordered_remove_if.
12607 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
12608 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
12609 * tree-vect-patterns.c (vect_pattern_recog_1): Use
12610 VEC_ORDERED_REMOVE_IF.
12612 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12614 PR tree-optimization/82665
12615 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
12616 pointer subtraction where arguments come from a memchr call.
12618 2018-05-01 Jakub Jelinek <jakub@redhat.com>
12620 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
12621 --push-state --as-needed and --pop-state instead of --as-needed and
12622 --no-as-needed if ld supports it.
12623 * configure: Regenerated.
12626 * doc/install.texi2html: Replace _002d with - and _002a with * in
12627 generated html files using sed.
12629 2018-04-30 David Malcolm <dmalcolm@redhat.com>
12632 * gcc-rich-location.c (blank_line_before_p): New function.
12633 (use_new_line): New function.
12634 (gcc_rich_location::add_fixit_insert_formatted): New function.
12635 * gcc-rich-location.h
12636 (gcc_rich_location::add_fixit_insert_formatted): New function.
12638 2018-04-30 David Malcolm <dmalcolm@redhat.com>
12640 * selftest.c (assert_streq): Rename "expected" and "actual" to
12641 "val1" and "val2". Extend NULL-handling to cover both inputs
12642 symmetrically, while still requiring both to be non-NULL for a pass.
12643 * selftest.h (assert_streq): Rename "expected" and "actual" to
12645 (ASSERT_EQ): Likewise.
12646 (ASSERT_EQ_AT): Likewise.
12647 (ASSERT_KNOWN_EQ): Likewise.
12648 (ASSERT_KNOWN_EQ_AT): Likewise.
12649 (ASSERT_NE): Likewise.
12650 (ASSERT_MAYBE_NE): Likewise.
12651 (ASSERT_MAYBE_NE_AT): Likewise.
12652 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
12653 the assertion to pass.
12654 (ASSERT_STREQ_AT): Likewise.
12656 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
12658 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
12659 interaction with -pie.
12661 2018-04-30 David Malcolm <dmalcolm@redhat.com>
12663 * selftest.h: Fix alphabetization of per-source-file selftest
12666 2018-04-30 Jason Merrill <jason@redhat.com>
12668 PR c++/61982 - dead stores to destroyed objects.
12669 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
12672 2018-04-30 Jason Merrill <jason@redhat.com>
12674 * tree.c (build_clobber): New.
12675 * tree.h: Declare it.
12676 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
12678 2018-04-30 David Malcolm <dmalcolm@redhat.com>
12680 * diagnostic-show-locus.c (layout::layout): Update for
12681 location_get_source_line returning a char_span.
12682 (struct char_span): Move to input.h.
12683 (struct correction): Update for fields in char_span becoming
12685 (struct source_line): Update for location_get_source_line
12686 returning a char_span.
12687 (layout::print_line): Likewise.
12688 * edit-context.c (edited_file::print_content): Likewise.
12689 (edited_file::print_diff_hunk): Likewise.
12690 (edited_file::print_run_of_changed_lines): Likewise.
12691 (edited_file::get_num_lines): Likewise.
12692 (edited_line::edited_line): Likewise.
12693 * final.c (asm_show_source): Likewise.
12694 * input.c (location_get_source_line): Convert return type
12695 from const char * to char_span, losing the final "line_len"
12697 (dump_location_info): Update for the above.
12698 (get_substring_ranges_for_loc): Likewise. Use a char_span
12699 when handling the literal within the line.
12700 (test_reading_source_line): Update for location_get_source_line
12701 returning a char_span.
12702 * input.h (class char_span): Move here from
12703 diagnostic-show-locus.c, converting from a struct to a class.
12704 Make data members private.
12705 (char_span::operator bool): New.
12706 (char_span::length): New.
12707 (char_span::get_buffer): New.
12708 (char_span::operator[]): New.
12709 (char_span::subspan): Make const.
12710 (char_span::xstrdup): New.
12711 (location_get_source_line): Convert return type from const char *
12712 to char_span, losing the final "line_size" param.
12714 2018-04-30 Jan Hubicka <jh@suse.cz>
12716 * lto-wrapper.c (ltrans_priorities): New static var.
12717 (cmp_priority): New.
12718 (run_gcc): Read priorities and if doing parallel build order
12719 the Makefile by them.
12721 2018-04-30 David Malcolm <dmalcolm@redhat.com>
12723 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
12725 2018-04-30 Richard Biener <rguenther@suse.de>
12727 * tree-cfg.c (verify_address): Remove base argument, add
12728 flag whether to check TREE_ADDRESSABLE and do that.
12729 (verify_expr): Remove.
12730 (verify_types_in_gimple_reference): Add pieces from verify_expr.
12731 (verify_gimple_assign_single): Likewise.
12732 (verify_gimple_switch): Likewise.
12733 (verify_expr_location_1): Dereference tp once. Add (disabled)
12734 piece from verify_expr.
12735 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
12737 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
12739 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
12741 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
12743 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
12744 (small_data_pattern): Likewise.
12745 (arc_rewrite_small_data): Likewise.
12746 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
12747 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
12748 (get_symbol_alignment): New function.
12749 (legitimate_small_data_address_p): Likewise.
12750 (legitimate_scaled_address): Update, call
12751 legitimate_small_data_address_p.
12752 (output_sdata): New static variable.
12753 (arc_print_operand): Update how we handle small data operands.
12754 (arc_print_operand_address): Likewise.
12755 (arc_legitimate_address_p): Update, use
12756 legitimate_small_data_address_p.
12757 (arc_rewrite_small_data_p): Remove.
12758 (arc_rewrite_small_data_1): Likewise.
12759 (arc_rewrite_small_data): Likewise.
12760 (small_data_pattern): Likewise.
12761 (compact_sda_memory_operand): Update to use
12762 legitimate_small_data_address_p and get_symbol_alignment.
12763 (prepare_move_operands): Don't rewite sdata pattern.
12764 (prepare_extend_operands): Remove.
12765 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
12767 (zero_extendqisi2): Likewise.
12768 (zero_extendhisi2): Likewise.
12769 (extendqihi2): Likewise.
12770 (extendqisi2): Likewise.
12771 (extendhisi2): Likewise.
12772 (addsi3): Likewise.
12773 (subsi3): Likewise.
12774 (andsi3): Likewise.
12775 * config/arc/constraints.md (Usd): Change it to memory constraint.
12777 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
12779 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
12780 as source of std instructions.
12781 * config/arc/arc.md (movsi_insn): Update pattern predicate to
12782 allow 6-bit constants as source for store instructions.
12783 (movdi_insn): Update instruction pattern to allow 6-bit constants
12784 as source for store instructions.
12786 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
12788 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
12790 2018-04-30 Nathan Sidwell <nathan@acm.org>
12791 Sandra Loosemore <sandra@codesourcery.com>
12793 * dumpfile.c (dump_open): Allow '-' for stdout.
12794 * doc/invoke.texi (Developer Options): Document dump filename
12795 determination early. Document stdin/stdout selection.
12797 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
12799 Microblaze Target: PIC data text relative
12801 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
12802 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
12804 * config/microblaze/microblaze.h (microblaze_constant_address_p):
12805 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
12806 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
12807 New addressing mode for data-text relative position indepenedent code.
12808 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
12809 'ADDRESS_SYMBOLIC_TXT_REL'.
12810 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
12811 (microblaze_legitimate_pic_operand): Exclude function calls from
12812 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
12813 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
12815 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
12816 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
12817 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
12818 for 'address + offset'.
12819 (microblaze_expand_prologue): Add new function prologue call for
12821 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
12822 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
12823 table in case of TARGET_PIC_DATA_TEXT_REL.
12824 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
12825 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
12826 Add new macros 'UNSPEC_TEXT',
12827 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
12828 + exclude function calls from 'UNSPEC_PLT' in case of data text
12830 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
12831 new target hook for generating address diff vector tables in case of
12833 * doc/tm.texi : Regenerate.
12834 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
12835 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
12836 of addr diff vector generation.
12837 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
12838 target hook definition.
12839 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
12840 Add default function for generate_pic_addr_diff_vec -> flag_pic.
12841 * doc/invoke.texi (Add new pic option): Add new microblaze pic
12842 option for data text relative.
12844 2018-04-30 Richard Biener <rguenther@suse.de>
12846 * tree-chrec.h (evolution_function_is_constant_p): Remove
12848 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
12850 2018-04-30 Richard Biener <rguenther@suse.de>
12853 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
12855 2018-04-30 Richard Biener <rguenther@suse.de>
12857 PR tree-optimization/28364
12858 PR tree-optimization/85275
12859 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
12860 copying first exit test.
12862 2018-04-28 Mark Wielaard <mark@klomp.org>
12864 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
12865 dwarf_version >= 5.
12866 (dwarf_AT): Handle DW_AT_addr_base.
12867 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
12869 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
12872 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
12873 (*ashl<dwi>3_doubleword_mask_1): Ditto.
12874 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
12875 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
12877 2018-04-28 Richard Biener <rguenther@suse.de>
12879 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
12880 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
12881 to reflect use. Only add interesting stmts.
12883 2018-04-27 Martin Jambor <mjambor@suse.cz>
12886 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
12887 the jump function allows for passing through aggregate values.
12889 2018-04-27 David Malcolm <dmalcolm@redhat.com>
12891 * input.h (in_system_header_at): Convert from macro to inline
12893 (from_macro_expansion_at): Likewise.
12894 (from_macro_definition_at): Likewise.
12896 2018-04-27 Jeff Law <law@redhat.com>
12898 * config.gcc: Mark tile* targets as deprecated/obsolete.
12900 2018-04-27 Richard Biener <rguenther@suse.de>
12902 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
12905 2018-04-27 Richard Biener <rguenther@suse.de>
12907 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
12909 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
12911 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
12912 with Yd constraint. Set "preferred_for_speed" attribute from
12913 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
12914 with Yd constraint.
12915 (*movdi_internal): Ditto.
12916 (movti_interunit splitters): Remove
12917 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
12918 (movdi_interunit splitters): Ditto.
12919 * config/i386/constraints.md (Ye): Remove.
12920 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
12922 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12925 * config/aarch64/constraints.md (Usg): Limit to 31.
12926 (Usj): Limit to 63.
12928 2018-04-27 Jakub Jelinek <jakub@redhat.com>
12930 PR tree-optimization/85529
12931 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
12932 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
12933 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
12934 zero extension or masking of the MSB bit.
12935 (optimize_range_tests): Add FIRST_BB argument, pass it through
12936 to optimize_range_tests_var_bound.
12937 (maybe_optimize_range_tests, reassociate_bb): Adjust
12938 optimize_range_tests callers.
12940 2018-04-26 Richard Biener <rguenther@suse.de>
12941 Jakub Jelinek <jakub@redhat.com>
12943 * cgraph.h (symbol_table): Just declare debug method here.
12944 * symtab.c (symbol_table::debug): Define.
12946 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
12948 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
12950 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
12952 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
12953 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
12954 (*movdi_internal): Substitute Yi and Yj constraint with x
12955 and Ym and Yn constraint with y constraint. Update "isa"
12956 attribute and set "preferred_for_speed" attribute from
12957 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
12958 (*movsi_internal): Ditto.
12959 (*movdf_internal): Ditto.
12960 (*movsf_internal): Ditto.
12961 (*zero_extendsidi2): Ditto.
12962 * config/i386/sse.md (vec_set<mode>_0): Ditto.
12963 (sse2_loadld): Ditto.
12964 (*vec_extract<ssevecmodelower>_0): Ditto.
12965 (*vec_extractv4si_0_zext_sse4): Ditto.
12966 (vec_concatv2di): Ditto.
12967 (*vec_dup<mode>): Ditto.
12968 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
12969 * config/i386/constraints.md (Yi): Remove.
12974 2018-04-26 Nathan Sidwell <nathan@acm.org>
12976 * dumpfile.c (dump_open): New.
12977 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
12978 (dump_finish): Detect stdio/stderr by value not name.
12980 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
12982 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
12984 2018-04-26 Tom de Vries <tom@codesourcery.com>
12987 * config/nvptx/nvptx.c (verify_neutering_jumps)
12988 (verify_neutering_labels): New function
12989 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
12991 2018-04-26 Tom de Vries <tom@codesourcery.com>
12994 * config/nvptx/nvptx.c (needs_neutering_p): New function.
12995 (nvptx_single): Use needs_neutering_p to skip over insns that do not
12998 2018-04-26 Richard Biener <rguenther@suse.de>
12999 Tom de Vries <tom@codesourcery.com>
13002 * lto-streamer-out.c (output_function): Fixup loops if required to match
13003 discovery done in the reader.
13005 2018-04-26 Richard Biener <rguenther@suse.de>
13007 PR tree-optimization/85116
13008 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
13009 have a loop exit from the single latch predecessor. Remove
13010 case of header with just condition.
13011 (ch_base::copy_headers): Exclude infinite loops from any
13013 (pass_ch::execute): Record exits.
13015 2018-04-26 Richard Biener <rguenther@suse.de>
13017 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
13018 prologue cost vector and pass it to vect_get_load_cost.
13019 (vect_get_peeling_costs_all_drs): Likewise.
13020 (vect_peeling_hash_get_lowest_cost): Likewise.
13021 (vect_enhance_data_refs_alignment): Likewise.
13023 2018-04-26 Richard Biener <rguenther@suse.de>
13025 PR middle-end/85450
13026 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
13027 checking of integer<->pointer conversions.
13028 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
13029 sign-/zero-extending pointer types.
13030 (expand_omp_for_static_chunk): Likewise.
13032 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
13033 Jean Lee <xiaoyur347@gmail.com>
13035 * config/mips/mips.c (mips_asan_shadow_offset): New function.
13036 (TARGET_ASAN_SHADOW_OFFSET): Define.
13037 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
13038 true for -fsanitize=address.
13040 2018-04-25 Mark Wielaard <mark@klomp.org>
13042 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
13045 2018-04-25 Jakub Jelinek <jakub@redhat.com>
13047 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
13048 than "alu", remove explicit "memory" and "imm_disp" attributes.
13049 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
13051 PR middle-end/85414
13052 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
13053 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
13054 gen_lowpart_no_emit.
13056 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
13059 * config/i386/i386.c (ix86_expand_builtin): Change memory
13060 operand to XI, extend p0 to Pmode.
13061 * config/i386/i386.md: Change unspec volatile and operand
13062 1 mode to XI, change operand 0 mode to P.
13064 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
13066 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
13067 GET_MODE_MASK before any checking.
13068 (nds32_can_use_bset_p): Likewise.
13069 (nds32_can_use_btgl_p): Likewise.
13071 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
13073 * config/nds32/nds32-doubleword.md: New define_split pattern for
13074 illegal register number.
13076 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
13078 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
13080 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
13082 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
13084 2018-04-25 Richard Biener <rguenther@suse.de>
13086 * lto-streamer.h (LTO_major_version): Bump to 8.
13088 2018-04-25 Jakub Jelinek <jakub@redhat.com>
13090 * BASE-VER: Set to 9.0.0.
13092 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
13094 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
13095 in __abskf2 and __powikf2.
13097 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13100 * config/aarch64/constraints.md (Usg, Usj): New constraints.
13101 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
13102 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
13103 Use the above on operand 2. Reindent.
13104 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
13106 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
13109 * common/config/i386/i386-common.c (ix86_handle_option): Don't
13111 * config/i386/i386.opt (mcet): Removed.
13112 * doc/install.texi: Remove -mcet documentation.
13113 * doc/invoke.texi: Likewise.
13115 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
13118 * doc/install.texi: Remove -mcet from bootstrap-cet.
13120 2018-04-24 Jakub Jelinek <jakub@redhat.com>
13123 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
13124 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
13128 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
13129 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
13130 containing a CONST_VECTOR.
13132 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
13134 * doc/install.texi: Update newlib dependency for nvptx.
13136 2018-04-24 Jakub Jelinek <jakub@redhat.com>
13139 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
13140 instead of INTVAL when shifting x left.
13142 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
13144 PR tree-optimization/85478
13145 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
13146 vect_grouped_store_supported for single element vectors.
13148 2018-04-24 Richard Biener <rguenther@suse.de>
13151 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
13152 load cost increase to the case of non-constant step.
13154 2018-04-24 Jakub Jelinek <jakub@redhat.com>
13157 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
13158 destination if any_malformed_asm.
13160 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
13162 PR middle-end/85496
13163 * expr.c (store_field): In the bitfield case, if the value comes from
13164 a function call and is returned in registers by means of a PARALLEL,
13165 do not change the mode of the temporary unless BLKmode and VOIDmode.
13167 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
13169 PR rtl-optimization/85423
13170 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
13171 dependencies to debug insns when the previous insn is non-debug.
13173 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
13175 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
13176 enums into a single definition.
13177 (fls): Fix predicates and printing.
13180 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
13182 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
13183 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
13184 and short u6 immediate.
13185 (check_if_valid_sleep_operand): Remove.
13186 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
13188 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
13190 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
13191 flag_always_save_lp condition.
13192 * config/nds32/nds32.opt (malways-save-lp): New option.
13194 2018-04-22 Shiva Chen <shiva0217@gmail.com>
13196 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
13197 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
13198 * config/nds32/nds32.h
13199 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
13200 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
13202 2018-04-22 Shiva Chen <shiva0217@gmail.com>
13204 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
13205 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
13207 2018-04-22 Shiva Chen <shiva0217@gmail.com>
13208 Chung-Ju Wu <jasonwucj@gmail.com>
13210 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
13212 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
13213 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
13215 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
13217 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
13219 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
13221 * config/nds32/nds32-protos.h (nds32_data_alignment,
13222 nds32_local_alignment): Declare.
13223 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
13224 nds32_local_alignment): New functions.
13225 (TARGET_CONSTANT_ALIGNMENT): Define.
13226 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
13228 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
13230 * config/nds32/nds32.c
13231 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
13232 (TARGET_MODES_TIEABLE_P): Likewise.
13234 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
13236 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
13237 level Ofast and Og.
13239 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
13240 Chung-Ju Wu <jasonwucj@gmail.com>
13242 * config/nds32/constants.md (unspec_volatile_element): Add enum values
13243 for unaligned access.
13244 * config/nds32/nds32-intrinsic.c: Implementation of expanding
13246 * config/nds32/nds32-intrinsic.md: Likewise.
13247 * config/nds32/nds32_intrinsic.h: Likewise.
13248 * config/nds32/nds32.h (nds32_builtins): Likewise.
13249 * config/nds32/nds32.opt (munaligned-access): New option.
13250 * config/nds32/nds32.c (nds32_asm_file_start): Display
13251 flag_unaligned_access status.
13253 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
13255 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
13256 -mno-relax is present.
13257 * config/riscv/linux.h (LINK_SPEC): Ditto.
13259 2018-04-20 Martin Sebor <msebor@redhat.com>
13262 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
13264 (gimple_fold_builtin_stxcpy_chk): Same.
13265 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
13267 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
13270 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
13271 __powikf2 when long double is IEEE 128-bit.
13273 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
13275 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
13276 step to make sure stack always aligned.
13278 2018-04-20 Carl Love <cel@us.ibm.com>
13281 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
13282 size check for arg0.
13284 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
13285 Tom de Vries <tom@codesourcery.com>
13288 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
13289 Emit insns for calls too.
13290 (nvptx_find_par): Always look for worker-level predecessor insn.
13291 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
13293 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
13294 (nvptx_process_pars): Propagate frames for calls.
13296 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
13299 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
13301 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
13302 (ix86_handle_option): Don't handle OPT_mibt.
13303 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
13305 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
13307 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
13308 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
13309 (ix86_target_macros): Define __CET__ with flag_cf_protection
13310 for -fcf-protection.
13311 * config/i386/i386.c (isa2_opts): Remove -mibt.
13312 * config/i386/i386.h (TARGET_IBT): Removed.
13313 (TARGET_IBT_P): Likewise.
13314 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
13315 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
13316 * config/i386/i386.opt (mcet): Update help message.
13317 (mshstk): Likewise.
13319 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
13320 -mcet as an alias for -mshstk.
13322 2018-04-20 Richard Biener <rguenther@suse.de>
13324 PR middle-end/85475
13325 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
13326 complexity by forcing a single use of the multiply operand.
13328 2018-04-20 Martin Jambor <mjambor@suse.cz>
13331 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
13332 recursion dependency to only apply to non-clones.
13334 2018-04-20 Martin Jambor <mjambor@suse.cz>
13337 * ipa-cp.c (create_specialized_node): Check that clones of
13338 self-recursive edges exist during IPA-CP.
13340 2018-04-19 Toon Moene <toon@moene.org>
13342 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
13345 2018-04-19 Jakub Jelinek <jakub@redhat.com>
13347 PR tree-optimization/85467
13348 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
13349 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
13350 VECTOR_CST element to type.
13352 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
13355 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
13356 * config/i386/i386.md (builtin_setjmp_setup): Removed.
13357 (builtin_longjmp): Likewise.
13358 (save_stack_nonlocal): New pattern.
13359 (restore_stack_nonlocal): Likewise.
13361 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
13364 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
13365 Replace ASM_OUTPUT_LABEL with fprintf.
13367 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
13370 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
13371 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
13372 * config/i386/i386-c.c (ix86_target_macros_internal): Also
13373 define __IBT__ and __SHSTK__ for -fcf-protection.
13374 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
13376 (ix86_trampoline_init): Likewise.
13377 (x86_output_mi_thunk): Likewise.
13378 (ix86_notrack_prefixed_insn_p): Likewise.
13379 (ix86_option_override_internal): Don't disallow -fcf-protection.
13380 * config/i386/i386.md (rdssp<mode>): Also enable for
13382 (incssp<mode>): Likewise.
13383 (nop_endbr): Likewise.
13384 * config/i386/i386.opt (mcet): Change help message to built-in
13387 (mshstk): Likewise.
13388 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
13389 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
13390 enable CET built-in functions.
13392 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
13394 * common/config/i386/i386-common.c
13395 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
13396 OPTION_MASK_ISA_MOVDIRI_UNSET,
13397 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
13398 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
13399 * config.gcc (movdirintrin.h): New header.
13400 * config/i386/cpuid.h (bit_MOVDIRI,
13401 bit_MOVDIR64B): New bits.
13402 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
13404 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
13405 (VOID, PVOID, PCVOID)): New function types.
13406 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
13407 __builtin_ia32_directstoreu_u64,
13408 __builtin_ia32_movdir64b): New builtins.
13409 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
13410 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
13412 (ix86_valid_target_attribute_inner_p): Ditto.
13413 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
13414 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
13415 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
13416 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
13417 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
13418 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
13419 (movdiri<mode>, movdir64b_<mode>): New.
13420 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
13421 * config/i386/immintrin.h: Include movdirintrin.h.
13422 * config/i386/movdirintrin.h: New file.
13423 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
13425 2018-04-19 Richard Biener <rguenther@suse.de>
13427 PR middle-end/85455
13428 * cfg.c (clear_bb_flags): When loop state says we have
13429 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
13431 2018-04-19 Richard Biener <rguenther@suse.de>
13433 PR tree-optimization/84737
13434 * tree-vect-data-refs.c (vect_copy_ref_info): New function
13435 copying restrict info.
13436 (vect_setup_realignment): Use it.
13437 * tree-vectorizer.h (vect_copy_ref_info): Declare.
13438 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
13439 the first DR to all generated stores.
13440 (vectorizable_load): Likewise for loads.
13442 2018-04-19 Jakub Jelinek <jakub@redhat.com>
13444 PR tree-optimization/85446
13445 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
13446 the integral and pointer types to have the same precision.
13448 * doc/install.texi: Document --disable-cet being the default and
13451 2018-04-18 Martin Liska <mliska@suse.cz>
13453 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
13456 2018-04-18 Martin Liska <mliska@suse.cz>
13459 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
13462 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
13463 arguments if they are comparable.
13465 2018-04-18 Martin Liska <mliska@suse.cz>
13468 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
13471 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
13474 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
13477 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
13478 ENDBR after calling __morestack.
13480 2018-04-18 David Malcolm <dmalcolm@redhat.com>
13483 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
13484 by using gcc_base_ver to generate a gcc_driver_version, and use
13485 it when generating GCC_DRIVER_NAME.
13486 * configure: Regenerate.
13488 2018-04-18 Jakub Jelinek <jakub@redhat.com>
13491 * config.gcc: Obsolete powerpc*-*-*spe*.
13493 2018-04-17 Jakub Jelinek <jakub@redhat.com>
13496 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
13497 (stabstr_D): Change type of unum from unsigned int to
13498 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
13501 2018-04-17 Jim Wilson <jimw@sifive.com>
13504 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
13505 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
13506 Set arg_pointer_offset after using pretend_args_size.
13508 2018-04-17 Jakub Jelinek <jakub@redhat.com>
13510 PR rtl-optimization/85431
13511 * dse.c (record_store): Ignore zero width stores.
13514 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
13515 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
13516 __builtin_stack_restore rather than after it.
13517 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
13518 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
13519 argument instead of virtual_dynamic_stack_rtx.
13521 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
13523 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
13525 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13526 Add note to error message to explain internal mapping of overloaded
13527 built-in function name to non-overloaded built-in function name.
13528 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
13531 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
13534 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
13535 where the inputs overlap with the output.
13537 2018-04-17 Jakub Jelinek <jakub@redhat.com>
13539 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
13540 (=v, v) alternative and explicit "memory" attribute.
13541 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
13542 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
13544 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
13545 "sselog1" type instead of "sselog".
13546 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
13547 "sselog". Remove explicit "memory" attribute.
13548 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
13549 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
13551 (vec_extract_hi_v32hi): Merge all alternatives into one, use
13552 "sselog1" type instead of "sselog". Remove explicit "memory"
13554 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
13555 use "sselog1" type instead of "sselog". Remove explicit "memory"
13557 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
13558 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
13560 (vec_extract_hi_v64qi): Merge all alternatives into one, use
13561 "sselog1" type instead of "sselog". Remove explicit "memory"
13563 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
13564 use "sselog1" type instead of "sselog". Remove explicit "memory"
13568 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
13570 PR middle-end/85414
13571 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
13574 2018-04-17 Martin Jambor <mjambor@suse.cz>
13577 * ipa-cp.c (create_specialized_node): Call
13578 expand_all_artificial_thunks if necessary.
13580 2018-04-17 Martin Liska <mliska@suse.cz>
13583 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
13584 in message, remote space in between '_G' and '('.
13586 2018-04-17 Jakub Jelinek <jakub@redhat.com>
13589 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
13590 avx512f_vmcmp<mode>3<round_saeonly_name>,
13591 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
13592 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
13593 avx512f_rndscale<mode><round_saeonly_name>,
13594 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
13595 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
13596 Use %<iptr>2 instead of %2 for -masm=intel.
13597 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
13598 avx512f_vcvttss2usi<round_saeonly_name>,
13599 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
13601 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
13602 avx512f_vcvttsd2usi<round_saeonly_name>,
13603 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
13604 Use %q1 instead of %1 for -masm=intel.
13605 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
13606 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
13607 of %3 for -masm=intel.
13608 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
13610 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
13612 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
13614 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
13615 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
13617 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
13619 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
13620 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
13621 %g1 and one with %0 and %1.
13622 (avx512er_vmrcp28<mode><round_saeonly_name>,
13623 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
13624 %1 for -masm=intel.
13625 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
13626 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
13627 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
13628 of %0 and %{%4%} for -masm=intel.
13629 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
13630 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
13631 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
13632 order of %0 and %{%5%}%{z%} for -masm=intel.
13634 2018-04-17 Jan Hubicka <jh@suse.cz>
13637 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
13639 2018-04-17 Martin Liska <mliska@suse.cz>
13642 * multiple_target.c (create_dispatcher_calls): Set apostrophes
13643 for target_clone error message. Make default implementation
13644 clone to be a local declaration.
13645 (separate_attrs): Add new argument and check for an empty
13647 (expand_target_clones): Handle it.
13648 (ipa_target_clone): Make redirection just for target_clones
13651 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
13652 Tom de Vries <tom@codesourcery.com>
13654 PR middle-end/84955
13655 * omp-expand.c (expand_oacc_for): Add dummy false branch for
13656 tiled basic blocks without omp continue statements.
13658 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
13661 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
13662 vec_extract expression as having side effects to make sure it gets
13665 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
13668 * config/i386/i386.c (get_builtin_code_for_version): Check
13671 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
13674 * config.gcc: Support "skylake".
13675 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13677 * config/i386/i386.c (m_SKYLAKE): Define.
13678 (processor_target_table): Add "skylake".
13679 (ix86_option_override_internal): Add "skylake".
13680 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
13681 PROCESSOR_CANNONLAKE.
13682 (get_builtin_code_for_version): Fix priority for
13683 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
13684 PROCESSOR_SKYLAKE-AVX512.
13685 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
13686 (processor_type): Add PROCESSOR_SKYLAKE.
13688 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
13689 Jason Merrill <jason@redhat.com>
13692 * convert.c (convert_to_integer_1): Use direct recursion for
13693 enumeral types and types with a precision less than the number
13694 of bits in their mode.
13696 2018-04-16 Julia Koval <julia.koval@intel.com>
13699 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
13700 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
13702 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
13705 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
13706 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
13707 and -mno-direct-move.
13709 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
13712 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
13713 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
13714 Ensure negative shifts result in {0}.
13716 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
13718 PR rtl-optimization/79916
13719 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
13720 regs (if any) to define how to gnerate SD moves when LRA is in
13723 2018-04-13 Jakub Jelinek <jakub@redhat.com>
13725 PR rtl-optimization/85393
13726 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
13727 * except.c (expand_dw2_landing_pad_for_region): Make static.
13728 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
13729 a label and unconditional jump to old_bb, rather than
13730 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
13733 PR rtl-optimization/85376
13734 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
13735 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
13736 instead of a specific value.
13738 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
13739 Bin Cheng <bin.cheng@arm.com>
13741 PR tree-optimization/82965
13742 PR tree-optimization/83991
13743 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
13744 by_profile_only parameter.
13745 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
13746 information if the loop was predicted to iterate too many times.
13747 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
13749 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
13752 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
13755 2018-04-13 Martin Liska <mliska@suse.cz>
13756 Jakub Jelinek <jakub@redhat.com>
13758 PR middle-end/81657
13759 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
13760 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
13761 * builtins.c (expand_builtin_memory_copy_args): Use
13762 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
13763 handle dest_addr == pc_rtx.
13765 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
13768 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
13769 asked to not generate direct moves.
13770 (fix_trunc<mode>si2_stfiwx): Similar.
13771 (fix_trunc<mode>si2_internal): Similar.
13773 2018-04-12 Jakub Jelinek <jakub@redhat.com>
13776 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
13777 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
13778 lookup if dest in some wider mode is known to be const0_rtx and
13779 if so, record permanent equivalence for it to be ZERO_EXTEND of
13780 the narrower mode destination.
13782 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
13784 * lto-streamer-out.c (output_function): Revert 259346.
13785 * omp-expand.c (expand_oacc_for): Likewise.
13787 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
13789 PR rtl-optimization/85354
13790 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
13791 * sel-sched.c (sel_global_init): ... here.
13793 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
13796 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
13797 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
13798 mode for PE-COFF targets.
13799 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
13800 (i386_pe_asm_lto_end): Likewise.
13801 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
13802 (TARGET_ASM_LTO_END): Likewise.
13803 * config/i386/winnt.c (saved_debug_info_level): New static variable.
13804 (i386_pe_asm_lto_start): New function.
13805 (i386_pe_asm_lto_end): Likewise.
13807 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
13808 Richard Biener <rguenther@suse.de>
13810 PR middle-end/84955
13811 * lto-streamer-out.c (output_function): Fix CFG loop state before
13813 * omp-expand.c (expand_oacc_for): Handle calls to internal
13814 functions like regular functions.
13816 2018-04-12 Richard Biener <rguenther@suse.de>
13819 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
13820 for the early LTO debug to properly generate references to it
13821 during DIE emission. Do not re-use that for the skeleton for
13823 (dwarf2out_early_finish): Likewise.
13825 2018-04-12 Jakub Jelinek <jakub@redhat.com>
13828 * config/i386/sse.md
13829 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
13830 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
13831 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
13832 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
13833 and output is a reg, avoid creating invalid lowpart subreg, but
13834 instead split into a 512-bit move. Don't split if not AVX512VL,
13835 input is xmm16+ reg and output is a mem.
13836 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
13837 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
13838 xmm16+ reg and output is a mem.
13840 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13842 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
13843 also for flag_dwarf2_cfi_asm.
13845 2018-04-12 Jakub Jelinek <jakub@redhat.com>
13847 PR rtl-optimization/85342
13848 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
13849 a bool scalar var inside of the loop instead. Don't try to update
13850 recog_data.operand after failed apply_change_group.
13852 2018-04-12 Tom de Vries <tom@codesourcery.com>
13855 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
13856 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
13857 array with flexible array member as array without given dimension.
13858 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
13859 argument for undefined param to true.
13861 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
13864 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
13865 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
13866 from PowerPC section.
13867 * config/rs6000/sysv4.opt (mcall-): Improve help text.
13868 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
13869 help text that is too long.
13870 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
13871 help text that is too long.
13872 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
13873 help text that is too long.
13875 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
13877 * config/alpha/alpha.md (stack_probe_internal): Rename
13878 from "probe_stack". Update all callers.
13880 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
13882 PR rtl-optimization/84566
13883 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
13884 sched_macro_fuse_insns.
13886 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
13889 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
13890 (compute_block_dependences): ... from here.
13892 2018-04-11 Jakub Jelinek <jakub@redhat.com>
13894 PR tree-optimization/85331
13895 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
13896 from int to HOST_WIDE_INT.
13898 2018-04-11 Martin Jambor <mjambor@suse.cz>
13901 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
13902 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
13903 not the same as the source val.
13904 (cgraph_edge_brings_value_p): New parameter.
13905 (gather_edges_for_value): Pass destination value to
13906 cgraph_edge_brings_value_p.
13907 (perhaps_add_new_callers): Likewise.
13908 (get_info_about_necessary_edges): Likewise and exclude values brought
13909 only by self-recursive edges.
13910 (create_specialized_node): Redirect only clones of self-calling edges.
13911 (+self_recursive_pass_through_p): New function.
13912 (find_more_scalar_values_for_callers_subset): Use it.
13913 (find_aggregate_values_for_callers_subset): Likewise.
13914 (known_aggs_to_agg_replacement_list): Removed.
13915 (decide_whether_version_node): Re-calculate known constants for all
13916 remaining context clones.
13918 2018-04-11 Richard Biener <rguenther@suse.de>
13921 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
13922 from early DWARF output.
13923 (dwarf2out_early_finish): Output line info unconditionally into
13924 early DWARF and add reference to it.
13926 2018-04-11 Jakub Jelinek <jakub@redhat.com>
13929 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
13930 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
13931 other than V2DFmode using iptr mode attribute.
13932 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
13934 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
13936 PR rtl-optimization/84659
13937 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
13939 2018-04-11 Jakub Jelinek <jakub@redhat.com>
13942 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
13944 (output_loc_list): Pass address of a dummy size variable even in the
13945 locview handling loop.
13946 (index_location_lists): Add comment on why skip_loc_list_entry can't
13949 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
13952 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
13955 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
13958 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
13959 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
13960 and -mstring-compare-inline-limit.
13962 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
13965 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
13966 for stack clash protection in a register whenever we need it to be in
13969 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
13971 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
13972 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
13974 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
13977 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
13979 (mlong-double-): Ditto.
13980 * config/rs6000/sysv4.opt (msdata=): Ditto.
13981 (mtls-size=): Ditto.
13983 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
13985 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
13986 erroneous entries for
13987 "vector int vec_ldl (int, long int *)", and
13988 "vector unsigned int vec_ldl (int, unsigned long int *)".
13989 Add comments and entries for
13990 "vector bool char vec_ldl (int, bool char *)",
13991 "vector bool short vec_ldl (int, bool short *)",
13992 "vector bool int vec_ldl (int, bool int *)",
13993 "vector bool long long vec_ldl (int, bool long long *)",
13994 "vector pixel vec_ldl (int, pixel *)",
13995 "vector long long vec_ldl (int, long long *)",
13996 "vector unsigned long long vec_ldl (int, unsigned long long *)".
13997 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
13998 type tree bool_long_long_type_node and correct definition of
13999 bool_V2DI_type_node to make reference to this new type tree.
14000 (rs6000_mangle_type): Replace erroneous reference to
14001 bool_long_type_node with bool_long_long_type_node.
14002 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
14003 comments to emphasize sign distinctions for char and int types and
14004 replace RS6000_BTI_bool_long constant with
14005 RS6000_BTI_bool_long_long constant. Also add comment to restrict
14006 use of RS6000_BTI_pixel.
14007 (bool_long_type_node): Remove this macro definition.
14008 (bool_long_long_type_node): New macro definition
14010 2018-04-10 Jakub Jelinek <jakub@redhat.com>
14012 PR rtl-optimization/85300
14013 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
14014 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
14015 simplify_unary_operation fails.
14017 2018-04-10 Martin Liska <mliska@suse.cz>
14019 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
14020 cgraph_edge and ipa_ref.
14022 2018-04-10 Jakub Jelinek <jakub@redhat.com>
14026 * config/i386/sse.md
14027 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
14028 computation of the VEC_MERGE selector from mask.
14029 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
14030 Fix decoding of the VEC_MERGE selector into mask.
14032 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
14034 PR tree-optimization/85286
14035 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
14037 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
14039 * final.c (final_1): Set insn_last_address as well as
14040 insn_current_address.
14042 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14045 * explow.c (emit_stack_probe): Call validize_mem on memory location
14046 before passing it to gen_probe_stack. Create address operand and
14047 legitimize it for the probe_stack_address case.
14049 2018-04-09 Jan Hubicka <jh@suse.cz>
14052 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
14053 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
14054 * tree.c (free_lang_data_in_type): Fix handling of binfos;
14056 (free_lang_data): Rebuild type inheritance graph.
14058 2018-04-09 Martin Sebor <msebor@redhat.com>
14060 * invoke.texi (-finline-small-functions): Mention other optimization
14062 (-findirect-inlining, -fpartial-inlining): Same.
14063 (-finline-functions-called-once): Same.
14064 (-freorder-blocks-and-partition): Same.
14066 2018-04-09 Jan Hubicka <jh@suse.cz>
14069 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
14070 jumps; choose last target that matches the criteria (i.e.
14071 no partition changes for non-crossing jumps).
14072 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
14073 support for redirecting crossing jumps to non-crossing.
14075 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
14077 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
14078 also for naked functions.
14080 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
14082 * config/arc/arc.md (add_shift): New pattern.
14083 (add_shift2): Likewise.
14084 (sub_shift): Likewise.
14085 (sub_shift_cmp0_noout): Likewise.
14086 (compare_si_ashiftsi): Likewise.
14087 (xbfu_cmp0_noout): New combine pattern.
14088 (xbfu_cmp0"): Likewise.
14089 (movsi_set_cc_insn): Place the predicable variant first.
14090 (commutative_binary_cmp0_noout): Remove clobber.
14091 (commutative_binary_cmp0): New pattern.
14092 (noncommutative_binary_cmp0): Likewise.
14093 (noncommutative_binary_cmp0_noout): Likewise.
14094 (noncommutative_binary_comparison_result_used): Removed.
14095 (rsub_cmp0): New pattern.
14096 (rsub_cmp0_noout): Likewise.
14097 (extzvsi): Changed, keep only meaningful variants.
14098 (SQH, SEZ): New iterators.
14099 (SQH_postfix): New mode attribute.
14100 (SEZ_prefix): New code attribute.
14101 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
14102 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
14103 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
14104 of numerical value.
14105 (noncommutative_operator): Check the availability of barrel
14108 2018-04-09 Richard Biener <rguenther@suse.de>
14110 PR tree-optimization/85284
14111 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
14112 Only use the niter constraining form of simple_iv when the exit
14113 is always executed.
14115 2018-04-09 Tom de Vries <tom@codesourcery.com>
14118 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
14119 (define_expand "*memory_barrier"): New define_expand.
14120 (define_insn "memory_barrier"): New insn.
14122 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
14124 PR rtl-optimization/80463
14125 PR rtl-optimization/83972
14126 PR rtl-optimization/83480
14128 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
14129 correct producer for the insn.
14130 (tidy_control_flow): Fixup seqnos in case of debug insns.
14132 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
14134 PR rtl-optimization/83913
14136 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
14137 different sched-times when merging exprs.
14139 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
14141 PR rtl-optimization/83962
14143 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
14144 tidy_fallthru_edge and tidy_control_flow.
14146 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
14148 PR rtl-optimization/83530
14150 * sel-sched.c (force_next_insn): New global variable.
14151 (remove_insn_for_debug): When force_next_insn is true, also leave only
14152 next insn in the ready list.
14153 (sel_sched_region): When the region wasn't scheduled, make another pass
14154 over it with force_next_insn set to 1.
14156 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
14158 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
14160 * config/nds32/constants.md (unspec_volatile_element): Add enum values
14161 for interrupt control.
14162 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
14163 functions for interrupt control.
14164 * config/nds32/nds32-intrinsic.md: Likewise.
14165 * config/nds32/nds32_intrinsic.h: Likewise.
14166 * config/nds32/nds32.h (nds32_builtins): Likewise.
14168 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
14170 * config/nds32/nds32.c (nds32_init_machine_status,
14171 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
14172 strict_aligned_p field.
14173 (nds32_expand_to_rtl_hook): New function.
14174 (TARGET_EXPAND_TO_RTL_HOOK): Define.
14175 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
14177 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
14178 Chung-Ju Wu <jasonwucj@gmail.com>
14180 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
14181 * config/nds32/nds32-n7.md: New file.
14182 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
14183 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
14185 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
14186 * config/nds32/nds32.md (pipeline_model): Add n7.
14187 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
14188 * config/nds32/pipelines.md: Include n7 settings.
14190 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
14191 Chung-Ju Wu <jasonwucj@gmail.com>
14193 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
14194 * config/nds32/nds32-e8.md: New file.
14195 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
14196 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
14198 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
14199 * config/nds32/nds32.md (pipeline_model): Add e8.
14200 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
14201 * config/nds32/pipelines.md: Include e8 settings.
14203 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
14204 Chung-Ju Wu <jasonwucj@gmail.com>
14206 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
14207 * config/nds32/nds32-n8.md: New file.
14208 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
14209 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
14211 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
14212 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
14213 * config/nds32/nds32.md (pipeline_model): Add n8.
14214 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
14215 * config/nds32/pipelines.md: Include n8 settings.
14217 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
14218 Chung-Ju Wu <jasonwucj@gmail.com>
14220 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
14221 * config/nds32/nds32-n9-2r1w.md: New file.
14222 * config/nds32/nds32-n9-3r2w.md: New file.
14223 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
14224 nds32_register_ports): New or modify for cpu n9.
14225 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
14227 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
14228 * config/nds32/nds32-utils.c: New file.
14229 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
14230 TARGET_MUL_SLOW): Define.
14231 * config/nds32/nds32.md (pipeline_model): New attribute.
14232 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
14233 New options that support cpu n9.
14234 * config/nds32/pipelines.md: Include n9 settings.
14235 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
14237 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
14239 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
14240 information if necessary.
14241 (output_cond_branch_compare_zero): Likewise.
14242 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
14243 (nds32_target_alignment): Refine for alignment.
14244 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
14245 (FUNCTION_BOUNDARY): Modify.
14246 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
14248 * config/nds32/nds32.opt (malways-align, malign-functions): New.
14250 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
14252 * config/nds32/constants.md (unspec_volatile_element): Add values for
14253 TLB operation and data prefetch.
14254 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
14255 functions for TLB operation and data prefetch.
14256 * config/nds32/nds32-intrinsic.md: Likewise.
14257 * config/nds32/nds32_intrinsic.h: Likewise.
14258 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
14259 (nds32_print_operand): Likewise.
14260 * config/nds32/nds32.h (nds32_builtins): Likewise.
14262 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
14263 Andrew Pinski <pinsika@gcc.gnu.org>
14265 PR middle-end/82976
14266 * match.pd: Use constant_boolean_node of correct type instead of
14267 boolean_true_node or boolean_false_node for simplifying
14268 pointer comparisons to zero.
14270 2018-04-07 Jakub Jelinek <jakub@redhat.com>
14272 PR tree-optimization/80021
14273 * tree.c (verify_type_variant): Make error call in verify_variant_match
14274 translatable and remove final full stop.
14276 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
14278 * config/nds32/constants.md (unspec_volatile_element): Add
14279 UNSPEC_VOLATILE_EH_RETURN.
14280 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
14281 nds32_output_stack_pop): Support dwarf exception handling process.
14282 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
14283 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
14284 exception handling process.
14285 (nds32_compute_stack_frame): Likewise.
14286 (nds32_return_addr_rtx): Likewise.
14287 (nds32_initial_elimination_offset): Likewise.
14288 (nds32_expand_prologue): Likewise.
14289 (nds32_expand_epilogue): Likewise.
14290 (nds32_dynamic_chain_address): New function.
14291 * config/nds32/nds32.h (machine_function): Add fields for dwarf
14292 exception handling.
14293 (DYNAMIC_CHAIN_ADDRESS): Define.
14294 (EH_RETURN_DATA_REGNO): Define.
14295 (EH_RETURN_STACKADJ_RTX): Define.
14296 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
14297 patterns for dwarf exception handling.
14299 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
14301 * config/nds32/nds32.h: Clean up obsolete macros.
14303 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
14305 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
14306 Add enum values for particular instructions.
14307 * config/nds32/nds32-intrinsic.c: Implementation of expanding
14308 particular intrinsic functions.
14309 * config/nds32/nds32-intrinsic.md: Likewise.
14310 * config/nds32/nds32_intrinsic.h: Likewise.
14311 * config/nds32/nds32.h (nds32_builtins): Likewise.
14312 * config/nds32/nds32.md (type): Add pbsad and pbsada.
14313 (btst, ave): New patterns for particular instructions.
14315 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
14317 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
14318 Add enum values for atomic load/store and memory sync.
14319 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
14321 * config/nds32/nds32-intrinsic.md: Likewise.
14322 * config/nds32/nds32_intrinsic.h: Likewise.
14323 * config/nds32/nds32.h (nds32_builtins): Likewise.
14325 2018-04-07 Jakub Jelinek <jakub@redhat.com>
14327 PR tree-optimization/85257
14328 * fold-const.c (native_encode_vector): If not all elts could fit
14329 and off is -1, return 0 rather than offset.
14330 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
14331 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
14332 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
14333 adjust buffer in native_interpret_expr call.
14335 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
14337 * config/nds32/constants.md (unspec_volatile_element): Add cache
14338 control enum values.
14339 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
14340 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
14341 * config/nds32/nds32.c (nds32_cctl_names): New.
14342 (nds32_print_operand): Handle cache control register names.
14343 * config/nds32/nds32.h (nds32_builtins): New enum values.
14344 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
14346 * config/nds32/nds32.md (type): Add mmu.
14347 * config/nds32/pipelines.md (simple_insn): Add mmu.
14349 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
14351 * config/nds32/nds32.md (type): Remove call.
14352 * config/nds32/pipelines.md (simple_insn): Likewise.
14354 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
14356 * config/nds32/constants.md (unspec_volatile_element): Add
14357 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
14358 UNSPEC_VOLATILE_FMFCFG.
14359 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
14360 description for fmfcfg and fmfcsr.
14361 (bdesc_1arg): Add fmtcsr.
14362 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
14363 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
14364 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
14365 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
14366 unspec_fmfcfg): New patterns.
14367 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
14368 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
14369 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
14370 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
14371 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
14372 __nds32__fmfcfg): Define.
14374 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
14376 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
14377 intrinsic register names.
14378 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
14379 intrinsic register enum values and macros.
14381 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
14383 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
14384 for load/store addressing form.
14385 (nds32_print_operand_address): Likewise.
14387 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
14390 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
14391 based on LABEL_REF. Remove useless assertion.
14392 (pic_address_needs_scratch): Fix formatting.
14393 (sparc_legitimize_pic_address): Minor tweaks.
14394 (sparc_delegitimize_address): Adjust assertion accordingly.
14395 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
14396 into symbolic_operand.
14397 (movsi_high_pic_label_ref): Likewise.
14398 (movsi_lo_sum_pic_label_ref): Likewise.
14399 (movdi_pic_label_ref): Likewise.
14400 (movdi_high_pic_label_ref): Likewise.
14401 (movdi_lo_sum_pic_label_ref): Likewise.
14403 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
14405 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
14406 custom LIB_SPEC setup.
14408 2018-04-06 Ruslan Bukin <br@bsdpad.com>
14409 Kito Cheng <kito.cheng@gmail.com>
14411 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
14412 * config/riscv/freebsd.h: New.
14414 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
14416 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
14417 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
14420 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
14421 Kito Cheng <kito.cheng@gmail.com>
14423 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
14424 nds32_output_call, nds32_symbol_binds_local_p): New functions.
14425 * config/nds32/nds32-protos.h (nds32_output_call,
14426 nds32_output_return): Declare.
14427 * config/nds32/nds32.md: Refine all the call and return patterns.
14429 2018-04-06 Jakub Jelinek <jakub@redhat.com>
14432 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
14433 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
14435 PR rtl-optimization/84872
14436 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
14437 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
14438 EDGE_CROSSING edge.
14440 2018-04-06 Tamar Christina <tamar.christina@arm.com>
14442 * expr.c (copy_blkmode_to_reg): Revert 254862.
14443 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
14445 2018-04-06 Richard Biener <rguenther@suse.de>
14447 PR middle-end/85244
14448 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
14449 after seeing a component reference with an adjacent field. Treat
14450 refs to arrays at struct end of external decls similar to
14451 refs to unconstrained commons.
14453 2018-04-06 Jakub Jelinek <jakub@redhat.com>
14456 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
14457 look through SAVE_EXPRs with non-side-effects argument. Adjust
14459 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
14462 2018-04-06 Richard Biener <rguenther@suse.de>
14464 PR middle-end/85180
14465 * alias.c (find_base_term): New wrapper around find_base_term
14466 unwinding CSELIB_VAL_PTR changes.
14467 (find_base_term): Do not restore CSELIB_VAL_PTR during the
14470 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14472 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
14474 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
14475 constant definitions.
14476 ("nop"): lr 0,0 -> nopr r0
14477 ("nop_lr0", "nop_lr1"): New insn definitions.
14479 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
14481 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
14482 NDS32_V3PUSH_AVAILABLE_P macro.
14484 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
14485 Chung-Ju Wu <jasonwucj@gmail.com>
14487 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
14488 (nds32*-*-*): Add float and fpu_config into supported_defaults.
14489 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
14490 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
14491 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
14492 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
14493 * config/nds32/constraints.md: New constraints and checking for hard
14494 float configuration.
14495 * config/nds32/iterators.md: New mode iterator and attribute for hard
14496 float configuration.
14497 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
14499 * config/nds32/nds32-fpu.md: New file.
14500 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
14501 deal with hard float code generation.
14502 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
14504 (abi_type, float_reg_number): New enum type.
14505 * config/nds32/nds32-predicates.c: New predicates for hard float.
14506 * config/nds32/nds32-protos.h: Declare functions for hard float.
14507 * config/nds32/nds32.c: Implementation for hard float configuration.
14508 * config/nds32/nds32.h: Definitions for hard float configuration.
14509 * config/nds32/nds32.md: Include hard float machine description and
14510 modify patterns for hard float configuration.
14511 * config/nds32/nds32.opt: New options for hard float configuration.
14512 * config/nds32/predicates.md: New predicates for hard float
14515 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
14517 * common/config/nds32/nds32-common.c
14518 (nds32_option_optimization_table): Enable -mreleax-hint by default.
14520 2018-04-05 Jakub Jelinek <jakub@redhat.com>
14522 PR middle-end/85195
14523 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
14524 CONSTRUCTOR_ELT (ctor, ...)->value.
14526 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
14529 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
14531 2018-04-05 Tom de Vries <tom@codesourcery.com>
14534 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
14537 2018-04-05 Shiva Chen <shiva0217@gmail.com>
14538 Kito Cheng <kito.cheng@gmail.com>
14540 * config/nds32/constraints.md (U33): Fine-tune checking condition.
14541 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
14542 * config/nds32/nds32.h (nds32_16bit_address_type): Add
14543 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
14545 2018-04-05 Shiva Chen <shiva0217@gmail.com>
14546 Kito Cheng <kito.cheng@gmail.com>
14548 * config/nds32/constraints.md (Ufe): New memory constraint.
14549 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
14550 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
14551 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
14553 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
14554 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
14556 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
14558 * config/nds32/nds32.md: Use optimize_size in the condition for
14559 alu-shift instructions.
14561 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
14563 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
14565 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
14567 * config/nds32/nds32.md (negsi2): Refine pattern.
14569 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
14570 Chung-Ju Wu <jasonwucj@gmail.com>
14572 * config/nds32/iterators.md (shift_rotate): New code iterator.
14573 (shift): New code attribute.
14574 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
14575 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
14576 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
14577 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
14578 bit-wise operations.
14579 (andsi3, *andsi3): Ditto.
14580 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
14581 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
14582 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
14583 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
14584 nds32_ior_operand, nds32_xor_operand): New predicates.
14586 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
14588 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
14589 (addsi3, subsi3): ... this.
14591 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
14593 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
14595 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
14597 * config/nds32/nds32.md: Adjust indention.
14599 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
14601 * config/nds32/nds32.md (feature): New attribute.
14603 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
14605 * config/nds32/nds32.md (subtype): New attribute.
14607 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
14610 * config/arm/arm-builtins.c (arm_expand_builtin): Change
14611 expansion to perform a bitwise AND of the argument followed by a
14612 boolean negation of the result.
14614 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
14616 PR rtl-optimization/84878
14617 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
14618 the basic block. Assert the use reference is not artificial and that
14619 it has an associated insn.
14621 2018-04-04 Michael Matz <matz@suse.de>
14623 * builtins.c (compute_objsize): Pass correct operand
14624 to array_at_struct_end_p.
14626 2018-04-04 Richard Biener <rguenther@suse.de>
14629 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
14630 from contexts for DINFO_LEVEL_TERSE and below.
14632 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
14634 * config/nds32/nds32-doubleword.md (move_<mode>): Require
14635 resiter_operand condition.
14636 * config/nds32/nds32.md (*move<mode>): Ditto.
14638 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
14639 Monk Chiang <sh.chiang04@gmail.com>
14641 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
14643 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
14645 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
14647 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
14648 Kito Cheng <kito.cheng@gmail.com>
14650 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
14651 nds32_cond_code_str, output_cond_branch,
14652 output_cond_branch_compare_zero, nds32_expand_cbranch,
14653 nds32_expand_cstore, nds32_expand_movcc,
14654 nds32_output_cbranchsi4_equality_zero,
14655 nds32_output_cbranchsi4_equality_reg,
14656 nds32_output_cbranchsi4_equality_reg_or_const_int,
14657 nds32_output_cbranchsi4_greater_less_zero: New functions.
14658 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
14659 nds32_expand_cstore, nds32_expand_movcc,
14660 nds32_output_cbranchsi4_equality_zero,
14661 nds32_output_cbranchsi4_equality_reg,
14662 nds32_output_cbranchsi4_equality_reg_or_const_int,
14663 nds32_output_cbranchsi4_greater_less_zero): Declare.
14664 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
14665 nds32_rimm11s_operand): New predicates.
14666 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
14667 * config/nds32/nds32.md: Rewrite all the branch and conditional move
14670 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
14672 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
14673 * config/nds32/nds32.md: Ditto.
14674 * config/nds32/pipelines.md: Ditto.
14676 2018-04-04 Richard Biener <rguenther@suse.de>
14678 PR tree-optimization/85168
14679 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
14680 propagating abnormals.
14682 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
14684 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
14686 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
14687 Kito Cheng <kito.cheng@gmail.com>
14689 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
14690 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
14691 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
14692 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
14693 * config/nds32/nds32.md (sibcall_internal): New.
14694 (sibcall_register): Remove.
14695 (sibcall_immediate): Remove.
14696 (sibcall_value_internal): New.
14697 (sibcall_value_register): Remove.
14698 (sibcall_value_immediate): Remove.
14699 * config/nds32/predicates.md (nds32_general_register_operand): New.
14700 (nds32_call_address_operand): New.
14702 2018-04-03 Jakub Jelinek <jakub@redhat.com>
14704 PR rtl-optimization/85167
14705 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
14706 bb_defs if *split_p, instead preinitialize it to NULL.
14708 PR tree-optimization/85156
14709 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
14710 evaluating the argument multiple times.
14712 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
14714 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
14716 (_mm_cvtpd_ps): Likewise.
14717 (_mm_cvttpd_epi32): Likewise.
14718 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
14719 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
14720 vector, pixel, and bool following altivec.h include.
14722 2018-04-03 Martin Sebor <msebor@redhat.com>
14724 * doc/extend.texi (Common Function Attributes): Clarify.
14725 (const attribute): Likewise.
14726 (pure attribute): Likewise.
14728 2018-04-03 Jakub Jelinek <jakub@redhat.com>
14731 * config/i386/i386.c (ix86_expand_vector_set): Use
14732 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
14734 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
14736 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
14737 instructions when changing rounding bits to preserve precision bits
14738 in the x87 control word.
14740 2018-04-03 Martin Liska <mliska@suse.cz>
14742 PR tree-optimization/82491
14743 * rtl.h (strip_offset_and_add): Replace += suboffset with
14744 poly_uint64 () + suboffset.
14746 2018-03-29 Martin Liska <mliska@suse.cz>
14747 Martin Jambor <mjambor@suse.cz>
14750 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
14751 param_type is not an integral or pointer type.
14753 2018-04-03 Richard Biener <rguenther@suse.de>
14755 * sese.h (recompute_all_dominators): Remove.
14757 2018-04-02 Martin Sebor <msebor@redhat.com>
14759 * doc/invoke.texi (-Wrestrict): Fix typos.
14761 2018-04-02 Jim Wilson <jimw@sifive.com>
14763 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
14764 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
14765 (<optab>di3, <optab>si3_extend): Likewise.
14766 (<optab>si3_mask, <optab>si3_mask_1): New.
14767 (<optab>di3_mask, <optab>di3_mask_1): New.
14768 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
14769 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
14770 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
14772 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
14774 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
14777 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
14779 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
14780 (nds32_canonicalize_comparison): New function.
14782 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
14783 Kito Cheng <kito.cheng@gmail.com>
14784 Kuan-Lin Chen <kuanlinchentw@gmail.com>
14786 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
14787 * config/nds32/constants.md (unspec_volatile_element): Add
14788 UNSPEC_VOLATILE_RELAX_GROUP.
14789 * config/nds32/nds32-relax-opt.c: New file.
14790 * config/nds32/nds32-predicates.c
14791 (nds32_symbol_load_store_p): New function.
14792 * config/nds32/nds32-protos.h
14793 (nds32_symbol_load_store_p): Declare function.
14794 (make_pass_nds32_relax_opt): Declare new rtl pass function.
14795 * config/nds32/nds32.c
14796 (nds32_register_pass): New function to register pass.
14797 (nds32_register_passes): New function to register passes.
14798 * config/nds32/nds32.md (relax_group): New pattern.
14799 * config/nds32/nds32.opt (mrelax-hint): New option.
14800 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
14802 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
14804 * config/nds32/t-nds32: Modify files dependency.
14806 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
14808 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
14809 (PROFILE_HOOK): Define its implementation.
14811 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
14813 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
14814 type and 32-bit size.
14816 2018-04-01 Jakub Jelinek <jakub@redhat.com>
14818 PR middle-end/85090
14819 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
14820 (V_128_256): New mode iterator.
14821 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
14822 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
14823 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
14825 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
14826 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
14828 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
14831 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
14832 NaN inputs correctly.
14834 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
14837 * config/rs6000/vsx.md (??r): New mode attribute.
14838 (*vsx_mov<mode>_64bit): Use it.
14839 (*vsx_mov<mode>_32bit): Likewise.
14841 2018-03-30 Martin Sebor <msebor@redhat.com>
14843 PR tree-optimization/84818
14844 * builtins.c (check_access): Use warning_n.
14846 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14849 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
14851 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
14854 2018-03-30 Julia Koval <julia.koval@intel.com>
14857 * x86-tune.def (movx, partial_reg_dependency): Enable for
14860 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
14862 PR inline-asm/84985
14863 * lra-constraints.c (process_alt_operands): Move setting
14864 this_alternative_matches below.
14866 2018-03-29 Martin Liska <mliska@suse.cz>
14869 * doc/invoke.texi: Document how LTO works with debug info.
14870 Describe auto-load support of binutils. Mention 'x86-64'
14871 as valid option value of -march option.
14873 2018-03-29 Jakub Jelinek <jakub@redhat.com>
14875 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
14878 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
14879 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
14880 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
14883 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
14886 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
14887 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
14888 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
14889 for RS6000_BTM_POWERPC64.
14890 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
14891 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
14892 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
14895 (DIVDEU): Likewise.
14897 2018-03-28 Carl Love <cel@us.ibm.com>
14900 2017-09-27 Carl Love <cel@us.ibm.com>
14902 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
14903 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
14904 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
14907 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14909 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
14910 instead of __vector bool.
14911 (_mm_max_pu8): Likewise.
14912 (_mm_min_pi16): Likewise.
14914 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
14917 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
14918 (DIVWEUO): Likewise.
14919 (DIVDEO): Likewise.
14920 (DIVDEUO): Likewise.
14921 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
14922 DIVWEUO and DIVDEUO.
14923 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
14924 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
14925 (div_extend): Likewise.
14926 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
14928 (__builtin_divweuo): Likewise.
14929 (__builtin_divdeo): Likewise.
14930 (__builtin_divdeuo): Likewise.
14932 2018-03-28 Jakub Jelinek <jakub@redhat.com>
14935 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
14936 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
14938 PR tree-optimization/82004
14939 * gimple-match-head.c (optimize_pow_to_exp): New function.
14940 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
14941 Don't fold to exp if optimize_pow_to_exp is false.
14943 2018-03-28 Martin Liska <mliska@suse.cz>
14946 * calls.c (initialize_argument_information): Fix trailing space.
14947 * common.opt: Fix typo and provide better explanation for
14948 -fsanitize-coverage option.
14949 * config/i386/i386.opt: Fix typo.
14951 2018-03-28 Jakub Jelinek <jakub@redhat.com>
14952 Martin Liska <mliska@suse.cz>
14955 * gimplify.c (asan_poison_variable): Don't do the check for
14956 gimplify_omp_ctxp here.
14957 (gimplify_decl_expr): Do it here.
14958 (gimplify_target_expr): Likewise.
14960 2018-03-28 Martin Liska <mliska@suse.cz>
14963 * config/i386/i386.c (ix86_function_arg_advance): Do not call
14964 chkp_type_bounds_count if MPX is not enabled.
14966 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
14968 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
14970 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
14973 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
14974 function to create the function decl for complex long double
14975 multiply and divide for -mabi=ieeelongdouble.
14976 (init_float128_ieee): Call it.
14978 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
14981 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
14982 -fcf-protection=branch -mibt.
14983 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
14985 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14988 * config/arm/arm.c (arm_valid_symbolic_address): Handle
14989 arm_word_relocations.
14991 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
14994 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
14995 extern array declarations.
14997 2018-03-27 Richard Biener <rguenther@suse.de>
14999 PR middle-end/84067
15000 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
15001 explicit single_use checks.
15003 2018-03-27 Richard Biener <rguenther@suse.de>
15005 PR tree-optimization/85082
15006 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
15009 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15011 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
15012 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
15013 Turn on fasynchronous-unwind-tables and funwind-tables.
15015 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
15018 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
15019 (*bmi_blsr_<mode>_ccz): Ditto.
15021 2018-03-26 Tom de Vries <tom@codesourcery.com>
15023 PR tree-optimization/85063
15024 * omp-general.c (offloading_function_p): New function. Factor out
15026 * omp-offload.c (pass_omp_target_link::gate): ... here.
15027 * omp-general.h (offloading_function_p): Declare.
15028 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
15029 with attribute omp declare target for offloading functions.
15031 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
15033 PR tree-optimization/84005
15034 * tree-data-ref.h (get_base_for_alignment): Declare.
15035 * tree-data-ref.c (get_base_for_alignment_1): New function.
15036 (get_base_for_alignment): Likewise.
15037 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
15038 get_base_for_alignment to find a suitable base object, instead
15039 of always using drb->base_address.
15041 2018-03-23 Jakub Jelinek <jakub@redhat.com>
15043 PR inline-asm/85022
15044 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
15045 known size by default.
15047 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
15049 PR inline-asm/85030
15050 * lra-constraints.c (process_alt_operands): Don't match BLKmode
15051 and non BLKmode operands.
15053 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15056 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
15057 Clean up attributes.
15059 2018-03-23 Richard Biener <rguenther@suse.de>
15062 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
15063 we are going to emit early debug for LTO.
15065 2018-03-23 Jakub Jelinek <jakub@redhat.com>
15067 PR inline-asm/85034
15068 * function.c (match_asm_constraints_1): Don't optimize if input
15069 doesn't satisfy general_operand predicate for output's mode.
15071 PR inline-asm/85022
15072 * alias.c (write_dependence_p): Don't require for x_canonicalized
15073 non-VOIDmode if x has VOIDmode.
15076 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
15077 just don't try to optimize it rather than assert it never happens.
15079 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
15081 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
15082 macro expansions for definition of ST_INTERNAL_<mode> and
15083 LD_INTERNAL_<mode> builtins.
15084 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
15086 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
15088 (altivec_expand_st_builtin): Likewise.
15089 (altivec_expand_builtin): Remove calls to deleted functions.
15090 (rs6000_address_for_altivec): Delete this function.
15091 * config/rs6000/vector.md: Remove expands for
15092 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
15094 2018-03-22 Sudakshina Das <sudi.das@arm.com>
15097 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
15098 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
15099 re-computing once computed.
15100 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
15101 (arm_init_machine_status): Initialize
15102 machine->static_chain_stack_bytes.
15104 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
15107 * doc/extend.texi: Add four new prototypes for vec_ld.
15108 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
15109 definitions for more logical presentation.
15110 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
15111 entries for V1TI variants of __builtin_altivec_ld builtin.
15112 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
15113 handling of V1TI variant of LVX icode pattern.
15114 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
15115 (rs6000_gimple_fold_builtin): Likewise.
15116 (altivec_init_builtins): Add code to define
15117 __builtin_altivec_lvx_v1ti function.
15119 2018-03-22 Jakub Jelinek <jakub@redhat.com>
15121 PR inline-asm/84941
15122 * function.c (match_asm_constraints_1): Don't do the optimization
15123 if input isn't a REG, SUBREG, MEM or constant.
15125 2018-03-22 Tom de Vries <tom@codesourcery.com>
15127 PR tree-optimization/84956
15128 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
15129 bb_has_abnormal_pred.
15131 2018-03-22 Jakub Jelinek <jakub@redhat.com>
15134 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
15135 DECL_INITIAL (decl) to decl at the end.
15136 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
15137 adjust the comment.
15139 2018-03-21 Joseph Myers <joseph@codesourcery.com>
15141 * doc/extend.texi (__builtin_tgmath): Document when complex
15142 integer types are treated as _Complex _Float64.
15144 2018-03-21 Tom de Vries <tom@codesourcery.com>
15146 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
15148 2018-03-21 Jakub Jelinek <jakub@redhat.com>
15150 PR tree-optimization/84960
15151 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
15152 if it is ENTRY block, move them into single succ of ENTRY in that case.
15154 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
15156 PR tree-optimization/84811
15157 * poly-int.h (poly_span_traits): Remove the T3 parameter and
15158 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
15159 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
15160 (known_subrange_p): Update accordingly. Cast each value involved
15161 in the size comparison, rather than casting the result of the
15164 2018-03-21 Jakub Jelinek <jakub@redhat.com>
15166 PR tree-optimization/84982
15167 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
15168 by flipping the least significant bit rather than all bits from
15169 bitpos to bitpos + bitsize - 1.
15171 2018-03-21 Nathan Sidwell <nathan@acm.org>
15173 * doc/extend.texi (Deprecated Features): Remove mention of
15174 long-deleted deprecations.
15176 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15179 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
15180 * configure: Regenerate.
15182 2018-03-21 Tom de Vries <tom@codesourcery.com>
15184 PR tree-optimization/83126
15185 * tree-parloops.c (num_phis): New function.
15186 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
15188 2018-03-21 Nathan Sidwell <nathan@acm.org>
15190 * doc/extend.texi (Deprecated Features): Update deprecated flags,
15191 mention anon-struct/union members and trailing attributes.
15193 2018-03-21 Bin Cheng <bin.cheng@arm.com>
15195 PR tree-optimization/84969
15196 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
15197 builtin memset partitions if they set different rhs values.
15199 2018-03-21 Jakub Jelinek <jakub@redhat.com>
15201 PR rtl-optimization/84989
15202 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
15203 VEC_DUPLICATE with scalar result mode.
15205 2018-03-21 Martin Liska <mliska@suse.cz>
15208 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
15209 not intended return statement.
15211 2018-03-21 Martin Liska <mliska@suse.cz>
15214 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
15215 (chkp_find_bound_slots_1): Limit number of iterations.
15217 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
15220 * Minor grammar fixes for x86 options.
15222 2018-03-20 Jakub Jelinek <jakub@redhat.com>
15225 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
15226 holding REG_CFA_RESTORE notes, instead turn them into a USE.
15228 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
15231 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
15232 (altivec_lvx_<mode>_1op): Likewise.
15233 (altivec_stvx_<mode>_2op): Likewise.
15234 (altivec_stvx_<mode>_1op): Likewise.
15235 (altivec_lvx_<VM2:mode>): New define_expand.
15236 (altivec_stvx_<VM2:mode>): Likewise.
15237 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
15238 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
15239 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
15240 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
15241 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
15242 (rs6000_gen_lvx): Likewise.
15243 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
15244 (altivec_expand_stv_builtin): Likewise.
15245 (altivec_expand_builtin): Likewise.
15246 * config/rs6000/vector.md: Likewise.
15248 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15251 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
15254 2018-03-20 Richard Biener <rguenther@suse.de>
15257 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
15258 sign-conversions as zero, fall back to standard scalar_stmt
15261 2018-03-20 Martin Liska <mliska@suse.cz>
15264 * predict.c (rebuild_frequencies): Handle case when we have
15265 PROFILE_ABSENT, but flag_guess_branch_prob is false.
15267 2018-03-20 Jakub Jelinek <jakub@redhat.com>
15270 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
15271 flag_section_anchors.
15272 * varasm.c (use_blocks_for_decl_p): Remove hack for
15273 dw2_force_const_mem.
15276 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
15278 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
15279 be created, use lowpart_subreg of operands[0] rather than operands[0]
15281 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
15282 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
15283 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
15284 and n constraint instead of aarch64_shift_imm_di and Usd.
15285 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
15286 (*aarch64_<optab>_reg_minus<mode>3): ... this.
15288 2018-03-20 Sudakshina Das <sudi.das@arm.com>
15291 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
15292 to favor GPR over NEON registers.
15293 (<shift>di3_neon): Likewise.
15295 2018-03-20 Tom de Vries <tom@codesourcery.com>
15298 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
15299 (nvptx_process_pars): Emit bar.sync asap and alap.
15301 2018-03-20 Tom de Vries <tom@codesourcery.com>
15304 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
15305 seen_label if seen_label is already set.
15307 2018-03-20 Jakub Jelinek <jakub@redhat.com>
15310 * config/i386/i386.c (fold_builtin_cpu): For features above 31
15311 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
15312 Use 1U instead of 1. Formatting fixes.
15315 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
15316 instead of TREE_TYPE (s1) for the return value.
15318 2018-03-19 Jakub Jelinek <jakub@redhat.com>
15320 PR tree-optimization/84946
15321 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
15322 bitsize + bitsize in poly_uint64 rather than poly_int64.
15325 * dwarf2asm.c: Include fold-const.c.
15326 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
15327 of decl rather than decl itself.
15329 PR rtl-optimization/84643
15330 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
15332 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
15335 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
15336 calling assemble_variable.
15338 2018-03-19 Sudakshina Das <sudi.das@arm.com>
15341 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
15342 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
15344 2018-03-19 Jim Wilson <jimw@sifive.com>
15347 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
15348 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
15349 (riscv_first_stack_step): Likewise.
15350 (riscv_option_override): Use STACK_BOUNDARY instead of
15351 MIN_STACK_BOUNDARY.
15352 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
15353 MIN_STACK_BOUNDARY.
15354 (BIGGEST_ALIGNMENT): Set to 128.
15355 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
15356 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
15359 2018-03-19 Richard Biener <rguenther@suse.de>
15361 PR tree-optimization/84933
15362 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
15363 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
15365 2018-03-19 Richard Biener <rguenther@suse.de>
15367 PR tree-optimization/84859
15368 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
15369 (cond_if_else_store_replacement): Perform sinking operation on
15370 single-store BBs regardless of MAX_STORES_TO_SINK setting.
15371 Generalize what a BB with a single eligible store is.
15373 2018-03-19 Richard Biener <rguenther@suse.de>
15375 PR tree-optimization/84929
15376 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
15377 chrec_is_positive against non-chrec arg.
15379 2018-03-19 Tamar Christina <tamar.christina@arm.com>
15382 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
15384 2018-03-18 Martin Liska <mliska@suse.cz>
15386 PR rtl-optimization/84635
15387 * regrename.c (build_def_use): Use matches_mode only when
15390 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
15392 PR tree-optimization/84913
15393 * tree-vect-loop.c (vectorizable_reduction): Don't try to
15394 vectorize chains of COND_EXPRs.
15396 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
15398 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
15400 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
15402 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
15404 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
15406 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
15408 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
15409 Kito Cheng <kito.cheng@gmail.com>
15411 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
15412 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
15413 (nds32_adjust_reg_alloc_order): New function.
15414 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
15416 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
15418 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
15419 nds32_print_operand, nds32_print_operand_address): Use
15420 HOST_WIDE_INT_PRINT_DEC instead.
15422 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
15424 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
15426 2018-03-17 Jakub Jelinek <jakub@redhat.com>
15429 * config/i386/i386.c (initial_ix86_tune_features,
15430 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
15431 unsigned long long.
15432 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
15433 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
15434 rather than 1u << ix86_tune. Formatting fix.
15435 (ix86_option_override_internal): Change ix86_arch_mask from
15436 unsigned int to unsigned HOST_WIDE_INT, initialize to
15437 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
15438 (ix86_function_specific_restore): Likewise.
15440 2018-03-16 Jakub Jelinek <jakub@redhat.com>
15443 * postreload.c (reload_combine_recognize_pattern): Perform
15444 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
15445 truncate_int_for_mode the result for the destination's mode.
15448 * hsa-gen.c (mem_type_for_type): Fix comment typo.
15449 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
15451 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
15454 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
15457 * lra-assigns.c (lra_split_hard_reg_for): Don't use
15458 regno_allocno_class_array and sorted_pseudos.
15459 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
15460 insns where regno is used.
15462 2018-03-16 Martin Liska <mliska@suse.cz>
15465 * multiple_target.c (create_dispatcher_calls): Redirect
15466 reference in the symbol table.
15468 2018-03-16 Martin Liska <mliska@suse.cz>
15471 * multiple_target.c (create_dispatcher_calls): Redirect also
15474 2018-03-16 Jakub Jelinek <jakub@redhat.com>
15478 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
15479 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
15480 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
15482 2018-03-16 Julia Koval <julia.koval@intel.com>
15484 * doc/invoke.texi (Skylake Server): Add CLWB.
15485 Cannonlake): Remove CLWB.
15487 2018-03-16 Jakub Jelinek <jakub@redhat.com>
15489 PR tree-optimization/84841
15490 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
15492 (FLOAT_ONE_CONST_TYPE): Define.
15493 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
15494 (sort_by_operand_rank): Put entries with higher constant_type last
15495 rather than first to match comments.
15497 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
15499 * config/nios2/nios2.md (movsi_internal): Fix thinko in
15502 2018-03-15 Jakub Jelinek <jakub@redhat.com>
15505 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
15506 check and use address of target always.
15508 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
15511 * config/i386/i386.c (indirect_thunk_needed): Update comments.
15512 (indirect_thunk_bnd_needed): Likewise.
15513 (indirect_thunks_used): Likewise.
15514 (indirect_thunks_bnd_used): Likewise.
15515 (indirect_return_needed): New.
15516 (indirect_return_bnd_needed): Likewise.
15517 (output_indirect_thunk_function): Add a bool argument for
15519 (output_indirect_thunk_function): Don't generate alias for
15520 function return thunk.
15521 (ix86_code_end): Call output_indirect_thunk_function to generate
15522 function return thunks.
15523 (ix86_output_function_return): Set indirect_return_bnd_needed
15524 and indirect_return_needed instead of indirect_thunk_bnd_needed
15525 and indirect_thunk_needed.
15527 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
15529 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
15530 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
15531 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
15533 2018-03-15 David Malcolm <dmalcolm@redhat.com>
15534 Paul Hua <paul.hua.gm@gmail.com>
15537 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
15539 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
15541 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
15542 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
15543 resp. SFmode cases.
15545 2018-03-15 Tamar Christina <tamar.christina@arm.com>
15548 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
15549 instead of GET_MODE_SIZE when comparing Units.
15551 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
15554 * varasm.c (hash_section): Return an unchangeble hash value
15555 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
15556 Return !aarch64_can_use_per_function_literal_pools_p ().
15558 2018-03-15 Jakub Jelinek <jakub@redhat.com>
15561 * optabs.c (emit_conditional_move): Pass address of cmode's copy
15562 rather than address of cmode as last argument to prepare_cmp_insn.
15564 2018-03-15 Julia Koval <julia.koval@intel.com>
15566 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
15567 F_AVX512VNNI, F_AVX512BITALG): New.
15569 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
15572 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
15573 insn for floating-point loads and stores.
15575 2018-03-14 Carl Love <cel@us.ibm.com>
15577 * config/rs6000/rs6000-c.c: Add macro definitions for
15578 ALTIVEC_BUILTIN_VEC_PERMXOR.
15579 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
15580 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
15581 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
15582 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
15584 * config/doc/extend.texi: Add prototypes for vec_permxor.
15586 2018-03-14 David Malcolm <dmalcolm@redhat.com>
15589 * diagnostic-show-locus.c (class layout_point): Convert m_line
15590 from int to linenum_type.
15591 (line_span::comparator): Use linenum "compare" function when
15592 comparing line numbers.
15593 (test_line_span): New function.
15594 (layout_range::contains_point): Convert param "row" from int to
15596 (layout_range::intersects_line_p): Likewise.
15597 (layout::will_show_line_p): Likewise.
15598 (layout::print_source_line): Likewise.
15599 (layout::should_print_annotation_line_p): Likewise.
15600 (layout::print_annotation_line): Likewise.
15601 (layout::print_leading_fixits): Likewise.
15602 (layout::annotation_line_showed_range_p): Likewise.
15603 (struct line_corrections): Likewise for field m_row.
15604 (line_corrections::line_corrections): Likewise for param "row".
15605 (layout::print_trailing_fixits): Likewise.
15606 (layout::get_state_at_point): Likewise.
15607 (layout::get_x_bound_for_row): Likewise.
15608 (layout::print_line): Likewise.
15609 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
15610 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
15611 * input.c (selftest::test_linenum_comparisons): New function.
15612 (selftest::input_c_tests): Call it.
15613 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
15614 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
15615 * selftest.h (ASSERT_GT): New macro.
15616 (ASSERT_GT_AT): New macro.
15617 (ASSERT_LT): New macro.
15618 (ASSERT_LT_AT): New macro.
15620 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
15622 PR rtl-optimization/84780
15623 * combine.c (distribute_links): Don't make a link based on pc_rtx.
15625 2018-03-14 Martin Liska <mliska@suse.cz>
15627 * tree.c (record_node_allocation_statistics): Use
15628 get_stats_node_kind.
15629 (get_stats_node_kind): New function extracted from
15630 record_node_allocation_statistics.
15631 (free_node): Use get_stats_node_kind.
15633 2018-03-14 Richard Biener <rguenther@suse.de>
15635 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
15636 that the value-set of ANTIC_IN doesn't grow.
15639 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
15641 (BB_VISITED_WITH_VISITED_SUCCS): New define.
15642 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
15644 2018-03-14 Julia Koval <julia.koval@intel.com>
15646 * config.gcc (icelake-client, icelake-server): New.
15648 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
15649 (initial_ix86_arch_features): Ditto.
15650 (PTA_SKYLAKE): Add SGX.
15651 (PTA_ICELAKE): Remove.
15652 (PTA_ICELAKE_CLIENT): New.
15653 (PTA_ICELAKE_SERVER): New.
15654 (ix86_option_override_internal): Split up icelake on icelake client and
15656 (get_builtin_code_for_version): Ditto.
15657 (fold_builtin_cpu): Ditto.
15658 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
15659 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
15660 * config/i386/i386.h (processor_type): Ditto.
15661 * doc/invoke.texi: Ditto.
15663 2018-03-14 Jakub Jelinek <jakub@redhat.com>
15666 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
15667 INTEGER_CST offset, add it together with bitpos / 8 and
15668 sign extend based on POINTER_SIZE.
15672 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
15675 * config/i386/constraints.md (Yc): New register constraint.
15676 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
15677 Use Yc constraint for alternative 2 of operand 0. Remove
15678 preferred_for_speed attribute.
15680 2018-03-14 Richard Biener <rguenther@suse.de>
15682 PR tree-optimization/84830
15683 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
15684 with the old one to avoid oscillations.
15686 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
15689 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
15691 (assign_by_spills): Return a flag of reload assignment failure.
15692 Do not process the reload assignment failures. Do not spill other
15693 reload pseudos if they has the same reg class. Update n if
15695 (lra_assign): Add a return arg. Set up from the result of
15696 assign_by_spills call.
15697 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
15698 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
15699 usage_insns if it is not NULL.
15700 (spill_hard_reg_in_range): New function.
15701 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
15702 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
15703 function prototypes.
15704 (lra_assign): Change prototype.
15705 * lra.c (lra): Add code to deal with fails by splitting hard reg
15708 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
15710 * config/riscv/riscv.opt (mrelax): New option.
15711 * config/riscv/riscv.c (riscv_file_start): Emit ".option
15712 "norelax" when riscv_mrelax is disabled.
15713 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
15715 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15718 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
15719 reassociation for int modes.
15721 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
15723 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
15724 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
15726 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
15727 * config/aarch64/aarch64-sve.md
15728 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
15729 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
15730 (*extend<mode><Vwide>2): Rename to...
15731 (aarch64_sve_extend<mode><Vwide>2): ...this.
15732 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
15733 renaming the old pattern to...
15734 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
15736 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
15737 define_expand, renaming the old pattern to...
15738 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
15739 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
15740 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
15741 account when deciding which SVE instruction the optab should use.
15742 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
15744 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
15746 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
15747 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
15748 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
15749 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
15750 (tlsdesc_small_<mode>): Turn a define_expand and use
15751 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
15752 (tlsdesc_small_advsimd_<mode>): ...this.
15753 (tlsdesc_small_sve_<mode>): New pattern.
15755 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
15757 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
15758 (UNSPEC_UMUL_HIGHPART): New constants.
15759 (MUL_HIGHPART): New int iteraor.
15760 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
15761 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
15763 (*<su>mul<mode>3_highpart): New define_insn.
15765 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
15768 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
15771 2018-03-13 Martin Liska <mliska@suse.cz>
15774 * (sem_item_optimizer::sem_item_optimizer): Initialize new
15776 (sem_item_optimizer::~sem_item_optimizer): Release it.
15777 (sem_item_optimizer::merge_classes): Register variable aliases.
15778 (sem_item_optimizer::fixup_pt_set): New function.
15779 (sem_item_optimizer::fixup_points_to_sets): Likewise.
15780 * ipa-icf.h: Declare new variables and functions.
15782 2018-03-13 Jakub Jelinek <jakub@redhat.com>
15784 PR middle-end/84834
15785 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
15786 integer_pow2p@2 and test integer_pow2p in condition.
15787 (A < 0 ? C : 0): Similarly for @1.
15789 PR middle-end/84831
15790 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
15791 characters starting at p contain '\0' character, don't look beyond
15795 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
15796 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
15799 * reg-stack.c (change_stack): Change update_end var from int to
15800 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
15801 also call set_block_for_insn on the newly added insns and rescan.
15804 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
15805 on the last operand.
15808 * tree.c (stabilize_reference_1): Return save_expr (e) for
15809 STATEMENT_LIST even if it doesn't have side-effects.
15811 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
15813 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
15815 2018-03-12 Renlin Li <renlin.li@arm.com>
15817 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
15818 aarch64_output_scalar_simd_mov_immediate.
15820 2018-03-12 Martin Sebor <msebor@redhat.com>
15822 PR tree-optimization/83456
15823 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
15824 for perfectly overlapping calls to memcpy.
15825 (gimple_fold_builtin_memory_chk): Same.
15826 (gimple_fold_builtin_strcpy): Handle no-warning.
15827 (gimple_fold_builtin_stxcpy_chk): Same.
15828 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
15830 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
15832 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
15833 parameter. Use it for SFmode.
15834 (rs6000_function_arg_advance_1): Adjust.
15835 (rs6000_function_arg): Adjust.
15836 (rs6000_gimplify_va_arg): Pass false for that new parameter.
15838 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
15840 PR rtl-optimization/84169
15841 PR rtl-optimization/84780
15842 * combine.c (can_combine_p): Check for a 2-insn combination whether
15843 the destination register is used between the two insns, too.
15845 2018-03-12 Richard Biener <rguenther@suse.de>
15847 PR tree-optimization/84803
15848 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
15849 for refs DR analysis didn't process.
15851 2018-03-12 Richard Biener <rguenther@suse.de>
15853 PR tree-optimization/84777
15854 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
15855 force-vectorize loops ignore whether we are optimizing for size.
15857 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
15859 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
15860 (TARGET_MD_ASM_ADJUST): Define.
15862 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
15863 Kito Cheng <kito.cheng@gmail.com>
15864 Chung-Ju Wu <jasonwucj@gmail.com>
15866 * config/nds32/nds32.c (nds32_compute_stack_frame,
15867 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
15868 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
15869 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
15870 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
15871 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
15872 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
15873 * config/nds32/nds32.md (prologue, epilogue): Use macro
15874 NDS32_V3PUSH_AVAILABLE_P to do checking.
15876 2018-03-11 Jakub Jelinek <jakub@redhat.com>
15879 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
15880 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
15881 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
15882 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
15883 addition of most attributes on !orig_type_die or the attribute not
15884 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
15886 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
15887 Chung-Ju Wu <jasonwucj@gmail.com>
15889 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
15890 __NDS32_VH__ macro.
15891 * config/nds32/nds32.opt (mvh): New option.
15893 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
15894 Chung-Ju Wu <jasonwucj@gmail.com>
15896 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
15898 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
15899 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
15902 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
15903 Chung-Ju Wu <jasonwucj@gmail.com>
15905 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
15907 * config/nds32/nds32-multiple.md (strlensi): New pattern.
15908 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
15910 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
15911 Kito Cheng <kito.cheng@gmail.com>
15912 Chung-Ju Wu <jasonwucj@gmail.com>
15914 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
15915 UNSPEC_FFMISM and UNSPEC_FLMISM.
15916 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
15917 for ffb, ffmism and flmism.
15918 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
15919 (unspec_ffmism): Ditto.
15920 (unspec_flmism): Ditto.
15921 (nds32_expand_builtin_impl): Check if string extension is available.
15922 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
15923 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
15925 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
15928 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
15931 * lra-assigns.c (assign_by_spills): Return a flag of reload
15932 assignment failure. Do not process the reload assignment
15933 failures. Do not spill other reload pseudos if they has the same
15935 (lra_assign): Add a return arg. Set up from the result of
15936 assign_by_spills call.
15937 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
15938 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
15939 usage_insns if it is not NULL.
15940 (spill_hard_reg_in_range): New function.
15941 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
15942 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
15943 function prototypes.
15944 (lra_assign): Change prototype.
15945 * lra.c (lra): Add code to deal with fails by splitting hard reg
15948 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
15951 * config/i386/i386.opt: Replace Enforcment with Enforcement.
15953 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
15956 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
15957 (dw_val_node): Add val_symbolic_view.
15958 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
15959 (symview_upper_bound): New.
15960 (new_line_info_table): Initialize symviews_since_reset.
15961 (dwarf2out_source_line): Count symviews_since_reset and set
15962 symview_upper_bound.
15963 (dw_val_equal_p): Handle symview.
15964 (add_AT_symview): New.
15965 (print_dw_val): Handle symview.
15966 (attr_checksum, attr_checksum_ordered): Likewise.
15967 (same_dw_val_p, size_of_die): Likewise.
15968 (value_format, output_die): Likewise.
15969 (add_high_low_attributes): Use add_AT_symview for entry_view.
15970 (dwarf2out_finish): Reset symview_upper_bound, clear
15973 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
15976 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
15977 Add strict argument and use it.
15978 (rs6000_split_multireg_move): Update for new strict argument.
15979 (mem_operand_gpr): Disallow all non-offsettable addresses.
15980 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
15982 2018-03-09 Jakub Jelinek <jakub@redhat.com>
15985 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
15986 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
15987 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
15990 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
15991 decl, use remap_type if we want to use the type.
15993 2018-03-09 Martin Sebor <msebor@redhat.com>
15995 PR tree-optimization/84526
15996 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
15998 (builtin_access::generic_overlap): Be prepared to handle non-array
16001 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
16003 PR rtl-optimization/84682
16004 * lra-constraints.c (process_address_1): Check is_address flag
16005 for address constraints.
16006 (process_alt_operands): Likewise.
16007 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
16008 preprocess_constraints.
16009 * recog.h (preprocess_constraints): Add oploc parameter.
16011 * recog.c (preprocess_constraints): Test address_operand for
16012 CT_ADDRESS constraints.
16014 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
16017 * lra-assigns.c (assign_by_spills): Return a flag of reload
16018 assignment failure. Do not process the reload assignment
16019 failures. Do not spill other reload pseudos if they has the same
16021 (lra_assign): Add a return arg. Set up from the result of
16022 assign_by_spills call.
16023 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
16024 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
16025 usage_insns if it is not NULL.
16026 (spill_hard_reg_in_range): New function.
16027 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
16028 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
16029 function prototypes.
16030 (lra_assign): Change prototype.
16031 * lra.c (lra): Add code to deal with fails by splitting hard reg
16034 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16037 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
16038 Accept complain bool parameter. Only emit errors if it is true.
16039 (arm_parse_cpu_option_name): Likewise.
16040 (arm_target_thumb_only): Adjust callers of the above.
16041 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
16042 prototype to take a default true bool parameter.
16043 (arm_parse_arch_option_name): Likewise.
16045 2018-03-09 David Malcolm <dmalcolm@redhat.com>
16046 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
16050 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
16051 * configure: Regenerate.
16052 * configure.ac ("linker --version-script option"): New.
16053 ("linker soname option"): New.
16055 2018-03-09 Richard Biener <rguenther@suse.de>
16057 PR tree-optimization/84775
16058 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
16059 immediate uses of predicate stmts and mark them modified.
16062 PR tree-optimization/84178
16063 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
16065 (version_loop_for_if_conversion): Delay update_ssa call.
16066 (tree_if_conversion): Delay update_ssa until after predicate
16069 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
16072 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
16073 when the function accesses prior frames.
16075 2018-03-08 Jakub Jelinek <jakub@redhat.com>
16078 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
16079 gen_llsym, otherwise call maybe_gen_llsym.
16081 PR inline-asm/84742
16082 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
16083 has ',' character inside of it.
16085 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16088 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
16089 as clobbering CC_REGNUM.
16091 2018-03-08 Richard Biener <rguenther@suse.de>
16093 PR middle-end/84552
16094 * tree-scalar-evolution.c: Include tree-into-ssa.h.
16095 (follow_copies_to_constant): Do not follow SSA names registered
16098 2018-03-08 Richard Biener <rguenther@suse.de>
16100 PR tree-optimization/84178
16101 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
16103 (version_loop_for_if_conversion): Delay update_ssa call.
16104 (tree_if_conversion): Delay update_ssa until after predicate
16107 2018-03-08 David Malcolm <dmalcolm@redhat.com>
16109 PR tree-optimization/84178
16110 * tree-if-conv.c (release_bb_predicate): Remove the
16111 the assertion that the stmts have NULL use_ops.
16112 Discard the statements, asserting that they haven't
16113 yet been added to a BB.
16115 2018-03-08 Richard Biener <rguenther@suse.de>
16117 PR tree-optimization/84746
16118 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
16119 (phi_translate): Pass in destination ANTIC_OUT set.
16120 (phi_translate_1): Likewise. For a simplified result lookup
16121 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
16122 (phi_translate_set): Adjust.
16123 (do_pre_regular_insertion): Likewise.
16124 (do_pre_partial_partial_insertion): Likewise.
16126 2018-03-08 Martin Liska <mliska@suse.cz>
16128 PR gcov-profile/84735
16129 * doc/gcov.texi: Document usage of profile files.
16130 * gcov-io.h: Document changes in the format.
16132 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
16136 * dwarf2out.c (struct dw_line_info_table): Update comments for
16138 (FORCE_RESET_NEXT_VIEW): New.
16139 (FORCE_RESETTING_VIEW_P): New.
16140 (RESETTING_VIEW_P): Check for -1 too.
16141 (ZERO_VIEW_P): Likewise.
16142 (new_line_info_table): Force-reset next view.
16143 (dwarf2out_begin_function): Likewise.
16144 (dwarf2out_source_line): Simplify zero_view_p initialization.
16145 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
16146 view directly. Omit view when omitting .loc at line 0.
16148 2018-03-08 Jakub Jelinek <jakub@redhat.com>
16150 PR tree-optimization/84740
16151 * tree-switch-conversion.c (process_switch): Call build_constructors
16152 only if info.phi_count is non-zero.
16154 PR tree-optimization/84739
16155 * tree-tailcall.c (find_tail_calls): Check call arguments against
16156 DECL_ARGUMENTS (current_function_decl) rather than
16157 DECL_ARGUMENTS (func) when checking for tail recursion.
16159 2018-03-07 Jakub Jelinek <jakub@redhat.com>
16161 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
16162 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
16163 Volker Reichelt's entry and add entries for people that perform
16164 GCC fuzzy testing and report numerous bugs.
16166 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
16169 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
16170 readonly data in sdata, if that is disabled.
16171 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
16172 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
16173 -mreadonly-in-sdata option.
16175 2018-03-07 Martin Sebor <msebor@redhat.com>
16177 PR tree-optimization/84468
16178 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
16179 basic block when looking for nul assignment.
16181 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
16184 * except.h (output_function_exception_table): Adjust prototype.
16185 * except.c (output_function_exception_table): Remove FNNAME parameter
16186 and add SECTION parameter. Ouput one part of the table at a time.
16187 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
16188 the first part of the exception table and emit unwind directives.
16189 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
16190 (i386_pe_seh_cold_init): Likewise.
16191 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
16192 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
16193 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
16194 (ix86_output_call_insn): Emit a nop in one more case for SEH.
16195 * config/i386/winnt.c: Include except.h.
16196 (struct seh_frame_state): Add reg_offset, after_prologue and
16197 in_cold_section fields.
16198 (i386_pe_seh_end_prologue): Set seh->after_prologue.
16199 (i386_pe_seh_cold_init): New function.
16200 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
16201 to seh->in_cold_section.
16202 (seh_emit_push): Record the offset of the push.
16203 (seh_emit_save): Record the offet of the save.
16204 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
16205 Test seh->after_prologue to disregard the epilogue.
16206 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
16207 (i386_pe_end_cold_function): New function.
16209 2018-03-07 Jakub Jelinek <jakub@redhat.com>
16212 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
16213 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
16216 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
16218 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
16219 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
16221 PR middle-end/84723
16222 * multiple_target.c: Include tree-inline.h and intl.h.
16223 (expand_target_clones): Diagnose and fail if node->definition and
16224 !tree_versionable_function_p (node->decl).
16226 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
16228 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
16230 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
16231 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
16232 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
16234 2018-03-06 Jakub Jelinek <jakub@redhat.com>
16237 * combine.c (try_combine): Use reg_or_subregno instead of handling
16238 just paradoxical SUBREGs and REGs.
16240 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
16242 * config/arc/arc.c (arc_finalize_pic): Remove function.
16243 (arc_must_save_register): We use single base PIC register, remove
16244 checks to save/restore the PIC register.
16245 (arc_expand_prologue): Likewise.
16246 * config/arc/arc-protos.h (arc_set_default_type_attributes):
16248 (arc_verify_short): Likewise.
16249 (arc_attr_type): Likewise.
16250 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
16251 (walk_stores): Likewise.
16252 (arc_address_cost): Make it static.
16253 (arc_verify_short): Likewise.
16254 (branch_dest): Likewise.
16255 (arc_attr_type): Likewise.
16256 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
16257 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
16258 (arc_final_prescan_insn): Remove inserting the nops due to
16259 hardware hazards. It is done in reorg step.
16260 (insn_length_variant_t): Remove.
16261 (insn_length_parameters_t): Likewise.
16262 (arc_insn_length_parameters): Likewise.
16263 (arc_get_insn_variants): Likewise.
16264 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
16266 2018-03-06 Jakub Jelinek <jakub@redhat.com>
16268 PR inline-asm/84683
16269 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
16272 PR tree-optimization/84687
16273 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
16275 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
16277 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16279 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
16280 Rename to ppc_speculation_barrier.
16281 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
16282 __builtin_ppc_speculation_barrier.
16284 2018-03-05 Jakub Jelinek <jakub@redhat.com>
16287 * combine.c (combine_simplify_rtx): Don't try to simplify if
16288 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
16291 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
16293 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
16294 to 32 bytes when compiling for POWER9.
16296 2018-03-05 Jakub Jelinek <jakub@redhat.com>
16299 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
16300 regparm >= 3 with no arg reg available also for calls with
16301 flag_force_indirect_call. Pass decl to ix86_function_regparm.
16304 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
16306 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
16308 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
16311 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
16313 2018-03-05 Richard Biener <rguenther@suse.de>
16315 PR tree-optimization/84486
16316 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
16317 When inserting a __builtin_assume_aligned call set the LHS
16318 SSA name alignment info accordingly.
16320 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
16322 PR tree-optimization/84114
16323 * config/aarch64/aarch64.c (aarch64_reassociation_width)
16324 Avoid reassociation of FLOAT_MODE addition.
16326 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
16328 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
16329 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
16330 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
16331 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
16332 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
16333 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
16334 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
16336 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
16337 __builtin_ia32_wbinvd): New builtins.
16338 (SPECIAL_ARGS2): New.
16339 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
16340 (SPECIAL_ARGS2): New.
16341 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
16342 (ix86_valid_target_attribute_inner_p): Ditto.
16343 (ix86_init_mmx_sse_builtins): Add special_args2.
16344 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
16345 TARGET_WBNOINVD_P): New.
16346 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
16347 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
16348 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
16349 * config/i386/immintrin.h (_wbinvd): New intrinsic.
16350 * config/i386/pconfigintrin.h: New file.
16351 * config/i386/wbnoinvdintrin.h: Ditto.
16352 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
16354 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
16356 2018-03-05 Richard Biener <rguenther@suse.de>
16358 PR tree-optimization/84670
16359 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
16361 (BB_VISITED_WITH_VISITED_SUCCS): New define.
16362 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
16363 (compute_antic_aux): Only assert the number of values in ANTIC_IN
16364 doesn't grow if all successors (recursively) were visited at least
16367 2018-03-05 Richard Biener <rguenther@suse.de>
16369 PR tree-optimization/84650
16370 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
16371 if executed in the loop pipeline.
16373 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
16375 * doc/configfiles.texi (Configuration Files): Move info about
16376 conditionalizing $target-protos.h to...
16377 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
16378 differs from $target-protos.h.
16380 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
16381 Chung-Ju Wu <jasonwucj@gmail.com>
16383 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
16384 * config/nds32/nds32-multiple.md (setmemsi): Define.
16385 * config/nds32/nds32-memory-manipulation.c
16386 (nds32_gen_dup_4_byte_to_word_value): New.
16387 (emit_setmem_word_loop): New.
16388 (emit_setmem_byte_loop): New.
16389 (nds32_expand_setmem_loop): New.
16390 (nds32_expand_setmem_loop_v3m): New.
16391 (nds32_expand_setmem_unroll): New.
16392 (nds32_expand_setmem): New.
16394 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
16395 Chung-Ju Wu <jasonwucj@gmail.com>
16397 * config/nds32/nds32-memory-manipulation.c
16398 (nds32_emit_load_store): New.
16399 (nds32_emit_post_inc_load_store): New.
16400 (nds32_emit_mem_move): New.
16401 (nds32_emit_mem_move_block): New.
16402 (nds32_expand_movmemsi_loop_unknown_size): New.
16403 (nds32_expand_movmemsi_loop_known_size): New.
16404 (nds32_expand_movmemsi_loop): New.
16405 (nds32_expand_movmemsi_unroll): New.
16406 (nds32_expand_movmemqi): Rename ...
16407 (nds32_expand_movmemsi): ... to this.
16408 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
16409 (movmemsi): ... to this.
16410 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
16411 (nds32_expand_movmemsi): ... to this.
16413 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
16414 Monk Chiang <sh.chiang04@gmail.com>
16415 Chung-Ju Wu <jasonwucj@gmail.com>
16417 * config/nds32/nds32-protos.h
16418 (nds32_expand_load_multiple): New arguments.
16419 (nds32_expand_store_multiple): Ditto.
16420 (nds32_valid_multiple_load_store): Rename ...
16421 (nds32_valid_multiple_load_store_p): ... to this.
16422 * config/nds32/nds32-memory-manipulation.c
16423 (nds32_expand_load_multiple): Refine implementation.
16424 (nds32_expand_store_multiple): Ditto.
16425 * config/nds32/nds32-multiple.md
16426 (load_multiple): Update nds32_expand_load_multiple interface.
16427 (store_multiple): Update nds32_expand_store_multiple interface.
16428 * config/nds32/nds32-predicates.c
16429 (nds32_valid_multiple_load_store): Rename ...
16430 (nds32_valid_multiple_load_store_p): ... to this and refine
16432 * config/nds32/predicates.md
16433 (nds32_load_multiple_and_update_address_operation): New predicate.
16434 (nds32_store_multiple_and_update_address_operation): New predicate.
16436 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
16437 Chung-Ju Wu <jasonwucj@gmail.com>
16439 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
16440 (combo): New attribute.
16441 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
16443 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
16445 * config/nds32/nds32.opt: Change -mcmodel= default value.
16447 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
16448 Monk Chiang <sh.chiang04@gmail.com>
16449 Chung-Ju Wu <jasonwucj@gmail.com>
16451 * config/nds32/constants.md (unspec_element): New enum.
16452 * config/nds32/constraints.md (Umw): New constraint.
16453 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
16454 * config/nds32/nds32-intrinsic.md: Likewise.
16455 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
16456 (nds32_valid_smw_lwm_base_p): New.
16457 (nds32_output_smw_single_word): New.
16458 (nds32_output_lmw_single_word): New.
16459 (nds32_expand_unaligned_load): New.
16460 (nds32_expand_unaligned_store): New.
16461 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
16462 (nds32_output_smw_single_word): Declare.
16463 (nds32_output_lmw_single_word): Declare.
16464 (nds32_expand_unaligned_load): Declare.
16465 (nds32_expand_unaligned_store): Declare.
16466 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
16467 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
16468 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
16469 NDS32_BUILTIN_UASTORE_DW.
16470 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
16473 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
16474 Kito Cheng <kito.cheng@gmail.com>
16475 Chung-Ju Wu <jasonwucj@gmail.com>
16477 * config/nds32/nds32-intrinsic.c
16478 (nds32_expand_builtin_null_ftype_reg): Delete.
16479 (nds32_expand_builtin_reg_ftype_imm): Ditto.
16480 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
16481 (nds32_read_argument): New.
16482 (nds32_legitimize_target): Ditto.
16483 (nds32_legitimize_argument): Ditto.
16484 (nds32_check_constant_argument): Ditto.
16485 (nds32_expand_unop_builtin): Ditto.
16486 (nds32_expand_unopimm_builtin): Ditto.
16487 (nds32_expand_binop_builtin): Ditto.
16488 (nds32_builtin_decl_impl): Ditto.
16489 (builtin_description): Ditto.
16490 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
16491 (nds32_init_builtins_impl): Ditto.
16492 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
16493 (nds32_builtin_decl): New.
16494 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
16495 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
16497 2018-03-02 Jeff Law <law@redhat.com>
16499 * reorg.c (stop_search_p): Handle DEBUG_INSN.
16500 (redundant_insn, fill_simple_delay_slots): Likewise.
16501 (fill_slots_from_thread): Likewise.
16502 * resource.c (mark_referenced_resources): Likewise.
16503 (mark_set_resources, find_dead_or_set_registers): Likewise.
16505 2018-03-02 Jakub Jelinek <jakub@redhat.com>
16507 * substring-locations.h (format_warning_va): Formatting fix for
16508 ATTRIBUTE_GCC_DIAG.
16509 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
16511 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
16512 * substring-locations.c: Include intl.h.
16513 (format_warning_va): Turned into small wrapper around
16514 format_warning_n_va, renamed to ...
16515 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
16516 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
16518 (format_warning_at_substring_n): New function.
16519 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
16520 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
16521 format_warning_at_substring with just a shorter name instead of
16522 const function pointer.
16523 (fmtwarn_n): New function.
16524 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
16525 appropriate, get rid of all the fmtstr temporaries, move conditionals
16526 with G_() wrapped string literals directly into fmtwarn arguments,
16527 cast dir.len to (int), formatting fixes.
16529 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
16531 * doc/invoke.texi: Remove "Cilk Plus" references.
16533 2018-03-02 Jakub Jelinek <jakub@redhat.com>
16534 Richard Biener <rguenther@suse.de>
16537 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
16538 for error or warning attributes if CALL_FROM_THUNK_P is set.
16541 2018-03-02 Jakub Jelinek <jakub@redhat.com>
16544 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
16545 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
16548 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
16549 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
16551 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
16552 instead of -1U in last predictors element's probability member.
16554 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
16557 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
16558 arguments if they are comparable.
16560 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
16562 PR tree-optimization/84634
16563 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
16564 masks and masked_loop_p with a single loop_masks, making sure it's
16565 null for bb vectorization.
16567 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
16569 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
16570 (vect_analyze_data_ref_access): Use loop->safe_len rather than
16571 loop->force_vectorize to check whether there is no alias.
16573 2018-03-02 Jakub Jelinek <jakub@redhat.com>
16576 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
16578 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
16580 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
16581 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
16582 instead of a loop around prev_real_insn.
16583 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
16586 PR inline-asm/84625
16587 * config/i386/i386.c (ix86_print_operand): Use conditional
16588 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
16591 2018-03-02 Richard Biener <rguenther@suse.de>
16593 PR tree-optimization/84427
16594 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
16595 (bitmap_set_subtract_values): Rewrite to handle multiple
16598 (prune_clobbered_mems): Likewise.
16599 (phi_translate): Take edge instead of pred/phiblock.
16600 (phi_translate_1): Likewise.
16601 (phi_translate_set): Likewise. Insert all translated
16602 exprs for a value into the set, keeping possibly multiple
16603 expressions per value.
16604 (compute_antic_aux): Adjust for phi_translate changes.
16605 When intersecting union the expressions and prune those
16606 not in the final value set, keeping possibly multiple
16607 expressions per value. Do not use value-insertion
16608 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
16609 all expressions. Add verification that the value-sets
16610 only shrink during iteration.
16611 (compute_partial_antic_aux): Adjust for the phi_translate changes.
16612 (do_pre_regular_insertion): Likewise.
16613 (do_pre_partial_partial_insertion): Likewise.
16615 2018-03-02 Richard Biener <rguenther@suse.de>
16618 * config/darwin.c (saved_debug_info_level): New static global.
16619 (darwin_asm_lto_start): Disable debug info generation for LTO out.
16620 (darwin_asm_lto_end): Restore debug info generation settings.
16622 2018-03-01 Martin Liska <mliska@suse.cz>
16625 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
16626 volatile arguments.
16628 2018-03-01 Richard Biener <rguenther@suse.de>
16631 * dwarf2out.c (gen_variable_die): Properly handle late VLA
16632 type annotation with LTO when debug was disabled at compile-time.
16634 2018-03-01 Matthew Fortune <mfortune@gmail.com>
16636 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
16638 (mips_final_postscan_insn): Likewise.
16640 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
16642 PR rtl-optimization/84528
16643 * alias.c (init_alias_target): Add commentary.
16644 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
16645 a unique base value if the frame pointer is not eliminated
16646 to the stack pointer.
16648 2018-03-01 Tom de Vries <tom@codesourcery.com>
16650 PR rtl-optimization/83327
16651 * lra-int.h (hard_regs_spilled_into): Declare.
16652 * lra.c (hard_regs_spilled_into): Define.
16653 (init_reg_info): Init hard_regs_spilled_into.
16654 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
16655 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
16656 (process_bb_lives): Handle hard_regs_spilled_into.
16657 (lra_create_live_ranges_1): Before doing liveness propagation, clear
16658 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
16660 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
16662 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
16663 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
16664 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
16665 * config/rs6000/aix72.h: New file.
16667 2018-02-28 Jakub Jelinek <jakub@redhat.com>
16669 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
16670 instead of warning_at with conditional singular and plural messages
16674 * stor-layout.c (update_alignment_for_field): For
16675 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
16676 && !DECL_PACKED (field), do the alignment update, just use
16677 only desired_align instead of MAX (type_align, desired_align)
16679 (place_field): Don't do known_align < desired_align handling
16680 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
16681 is non-NULL, instead do it after rli->prev_field handling and
16682 only if not within a bitfield word. For DECL_PACKED (field)
16683 use type_align of BITS_PER_UNIT.
16685 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
16687 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
16688 superfluous parentheses and trailing spaces.
16690 2018-02-28 Richard Biener <rguenther@suse.de>
16692 PR tree-optimization/84584
16693 * graphite-scop-detection.c (scop_detection::add_scop): Discard
16694 SCoPs with fake exit edge.
16696 2018-02-28 Martin Liska <mliska@suse.cz>
16699 * timevar.c (timer::print): Fix format to properly print 100%
16702 2018-02-28 Richard Biener <rguenther@suse.de>
16704 PR middle-end/84607
16705 * genmatch.c (capture_info::walk_match): Do not mark
16706 captured expressions without operands as expr_p given
16707 they act more like predicates and should be subject to
16708 "lost tail" side-effect preserving.
16710 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
16712 PR rtl-optimization/81611
16713 * auto-inc-dec.c (attempt_change): Move dead note from
16714 mem_insn if it's the next use of regno
16715 (find_address): Take address use of reg holding
16716 non-incremented value. Add parm to limit search to the named
16718 (merge_in_block): Attempt to use a mem insn that is the next
16719 use of the original regno.
16721 2018-02-27 Martin Sebor <msebor@redhat.com>
16724 * doc/invoke.texi (-Wmissing-attributes): New option.
16725 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
16727 2018-02-27 Martin Sebor <msebor@redhat.com>
16729 PR translation/84207
16730 * diagnostic-core.h (warning_n, error_n, inform_n): Change
16731 n argument to unsigned HOST_WIDE_INT.
16732 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
16733 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
16734 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
16735 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
16737 2018-02-27 Richard Biener <rguenther@suse.de>
16739 PR tree-optimization/84512
16740 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
16741 Do not use the estimate returned from record_stmt_cost for
16742 the scalar iteration cost but sum properly using add_stmt_cost.
16744 2018-02-27 Richard Biener <rguenther@suse.de>
16746 PR tree-optimization/84466
16747 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
16748 Adjust last change to less strictly validate use operands.
16750 2018-02-27 Martin Liska <mliska@suse.cz>
16752 PR gcov-profile/84548
16753 * gcov.c (process_file): Allow partial overlap and consider it
16754 also as group functions.
16755 (output_lines): Properly calculate range of lines for a group.
16757 2018-02-27 Martin Liska <mliska@suse.cz>
16759 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
16760 'ggc' suffixes. Change first column width.
16761 (timer::print): Fix formatting of the column.
16763 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
16765 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
16766 preserve inline entry blocks for the sake of debug inline
16767 entry point markers alone.
16768 (remove_unused_locals): Suggest in comments a better place to
16769 force the preservation of inline entry blocks that are
16770 otherwise unused, but do not preserve them.
16772 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
16774 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
16776 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
16779 * config/i386/constraints.md (Bs): Replace
16780 ix86_indirect_branch_register with
16781 TARGET_INDIRECT_BRANCH_REGISTER.
16783 * config/i386/i386.md (indirect_jump): Likewise.
16784 (tablejump): Likewise.
16785 (*sibcall_memory): Likewise.
16786 (*sibcall_value_memory): Likewise.
16787 Peepholes of indirect call and jump via memory: Likewise.
16788 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
16789 (*sibcall_value_GOT_32): Likewise.
16790 * config/i386/predicates.md (indirect_branch_operand): Likewise.
16791 (GOT_memory_operand): Likewise.
16792 (call_insn_operand): Likewise.
16793 (sibcall_insn_operand): Likewise.
16794 (GOT32_symbol_operand): Likewise.
16795 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
16797 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
16799 PR rtl-optimization/83496
16800 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
16801 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
16802 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
16803 redundant insn, if any.
16804 (relax_delay_slots): Likewise.
16805 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
16807 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
16809 PR tree-optimization/83965
16810 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
16811 that grouped statements are part of a reduction chain. Return
16812 true if the statement is not marked as a reduction itself but
16813 is part of a group.
16814 (vect_recog_dot_prod_pattern): Don't check whether the statement
16815 is part of a group here.
16816 (vect_recog_sad_pattern): Likewise.
16817 (vect_recog_widen_sum_pattern): Likewise.
16819 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
16822 * final.c (rest_of_clean_state): Also look for calls inside sequences.
16824 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
16827 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
16829 (ix86_output_indirect_function_return): New prototype.
16830 (ix86_split_simple_return_pop_internal): Likewise.
16831 * config/i386/i386.c (indirect_return_via_cx): New.
16832 (indirect_return_via_cx_bnd): Likewise.
16833 (indirect_thunk_name): Handle return va CX_REG.
16834 (output_indirect_thunk_function): Create alias for
16835 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
16836 (ix86_output_indirect_jmp): Remove the bool argument.
16837 (ix86_output_indirect_function_return): New function.
16838 (ix86_split_simple_return_pop_internal): Likewise.
16839 * config/i386/i386.md (*indirect_jump): Don't pass false
16840 to ix86_output_indirect_jmp.
16841 (*tablejump_1): Likewise.
16842 (simple_return_pop_internal): Change it to define_insn_and_split.
16843 Call ix86_split_simple_return_pop_internal to split it for
16844 -mfunction-return=.
16845 (simple_return_indirect_internal): Call
16846 ix86_output_indirect_function_return instead of
16847 ix86_output_indirect_jmp.
16849 2018-02-26 Jakub Jelinek <jakub@redhat.com>
16852 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
16853 memset and value initialization afterwards.
16855 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
16857 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
16859 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16862 * common/config/aarch64/aarch64-common.c
16863 (aarch_option_optimization_table[]): Switch
16864 off fomit-frame-pointer
16866 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
16867 Chung-Ju Wu <jasonwucj@gmail.com>
16869 * config/nds32/nds32-multiple.md (load_multiple): Disallow
16871 (store_multiple): Ditto.
16873 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
16875 * config.gcc: Add --with-cpu support for nds32 target.
16876 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
16877 * config/nds32/nds32.opt: Add -mcpu= option.
16879 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
16881 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
16882 isel=yes): Warn for these deprecated options.
16884 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
16886 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
16887 ISA_2_5_MASKS_EMBEDDED.
16889 2018-02-23 Jakub Jelinek <jakub@redhat.com>
16891 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
16892 p->max as pointers rather than using iterative_hash_expr.
16894 2018-02-23 Carl Love <cel@us.ibm.com>
16896 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
16897 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
16899 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
16900 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
16901 P8V_BUILTIN_VEC_VUNSIGNED2.
16903 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
16906 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
16907 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
16909 * lra-constraints.c (curr_insn_transform): Set up
16910 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
16912 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
16913 * lra-eliminations.c (spill_pseudos): Ditto.
16914 (process_insn_for_elimination): Ditto.
16915 * lra-lives.c (reg_early_clobber_p): Use the new macros.
16916 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
16917 LRA_NON_CLOBBERED_ALT.
16919 2018-02-22 Martin Sebor <msebor@redhat.com>
16921 PR tree-optimization/84480
16922 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
16923 to maybe_diag_stxncpy_trunc. Call it.
16924 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
16925 from gimple_fold_builtin_strcpy. Print inlining stack.
16926 (handle_builtin_stxncpy): Print inlining stack.
16927 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
16929 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
16932 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
16933 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
16934 and -fcheck-pointer-bounds are used together.
16935 (indirect_thunk_prefix): New enum.
16936 (indirect_thunk_need_prefix): New function.
16937 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
16938 "_nt" instead of "_bnd" for NOTRACK prefix.
16939 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
16940 (output_indirect_thunk_function): Likewise.
16942 (ix86_code_end): Update output_indirect_thunk_function calls.
16943 (ix86_output_indirect_branch_via_reg): Replace
16944 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
16945 (ix86_output_indirect_branch_via_push): Likewise.
16946 (ix86_output_function_return): Likewise.
16947 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
16948 incompatible with -fcf-protection=branch and
16949 -fcheck-pointer-bounds.
16951 2018-02-22 Steve Ellcey <sellcey@cavium.com>
16954 * config/aarch64/aarch64.c (aarch64_print_address_internal):
16955 Change gcc_assert call to output_operand_lossage.
16957 2018-02-22 Steve Ellcey <sellcey@cavium.com>
16959 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
16961 2018-02-22 DJ Delorie <dj@redhat.com>
16962 Sebastian Perta <sebastian.perta@renesas.com>
16963 Oleg Endo <olegendo@gcc.gnu.org>
16965 * config/rx/rx.c (rx_rtx_costs): New function.
16966 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
16968 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
16970 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
16972 2018-02-22 Martin Liska <mliska@suse.cz>
16975 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
16976 Add "native" as a possible value.
16978 2018-02-22 Martin Liska <mliska@suse.cz>
16981 * config/i386/i386.c (ix86_option_override_internal):
16982 Add "native" as a possible value for -march and -mtune.
16984 2018-02-22 Jakub Jelinek <jakub@redhat.com>
16987 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
16988 to all type variants.
16990 PR tree-optimization/84503
16991 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
16992 width as info->bitpos + info->bitsize - start.
16993 (merged_store_group::merge_overlapping): Simplify width computation.
16994 (check_no_overlap): New function.
16995 (imm_store_chain_info::try_coalesce_bswap): Compute expected
16996 start + width and last_order of the group, fail if check_no_overlap
16998 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
16999 to group if check_no_overlap fails.
17001 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
17003 * config/rs6000/altivec.md: Delete contraint arguments to
17004 define_expand, define_split, and define_peephole2, and in
17005 define_insn_and_split if always unused.
17006 * config/rs6000/darwin.md: Ditto.
17007 * config/rs6000/dfp.md: Ditto.
17008 * config/rs6000/rs6000.md: Ditto.
17009 * config/rs6000/sync.md: Ditto.
17010 * config/rs6000/vector.md: Ditto.
17011 * config/rs6000/vsx.md: Ditto.
17013 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
17015 * config/rs6000/altivec.md: Write output control strings as braced
17016 blocks instead of double-quoted strings.
17017 * config/rs6000/darwin.md: Ditto.
17018 * config/rs6000/rs6000.md: Ditto.
17019 * config/rs6000/vector.md: Ditto.
17020 * config/rs6000/vsx.md: Ditto.
17022 2018-02-21 Jason Merrill <jason@redhat.com>
17024 PR c++/84314 - ICE with templates and fastcall attribute.
17025 * attribs.c (build_type_attribute_qual_variant): Remove assert.
17027 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
17029 * ipa-cp.c (determine_versionability): Fix comment typos.
17031 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
17034 * ipa-cp.c (determine_versionability): Do not version functions caling
17037 2018-02-21 Martin Liska <mliska@suse.cz>
17040 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
17041 Add "native" as a possible value.
17042 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
17043 the macro when native cpu detection is available.
17045 2018-02-21 Martin Liska <mliska@suse.cz>
17048 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
17049 Add "native" as a possible value.
17050 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
17051 when native cpu detection is available.
17053 2018-02-21 Jakub Jelinek <jakub@redhat.com>
17054 Martin Sebor <msebor@redhat.com>
17056 PR tree-optimization/84478
17057 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
17059 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
17060 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
17061 support which is conservatively correct, for 2 only stay conservative
17062 for maxlen. Formatting and comment capitalization fixes. Add STRICT
17063 argument to the 2 argument get_range_strlen, adjust 6 arg
17064 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
17066 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
17067 (gimple_fold_builtin_strlen): Pass true as last argument to
17070 2018-02-20 Martin Sebor <msebor@redhat.com>
17072 PR middle-end/84095
17073 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
17074 (builtin_memref::set_base_and_offset): Same. Handle inner references.
17075 (builtin_memref::builtin_memref): Factor out parts into
17076 set_base_and_offset and call it.
17078 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
17080 PR middle-end/84406
17081 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
17082 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
17083 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
17084 search at the associated MODE_INT.
17086 2018-02-20 Jeff Law <law@redhat.com>
17088 PR middle-end/82123
17089 PR tree-optimization/81592
17090 PR middle-end/79257
17091 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
17092 for range data rather than using global data.
17093 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
17094 range data rather than using global data.
17095 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
17096 pass it to children as needed.
17097 (struct directive::fmtresult): Similarly.
17098 (struct directive::set_width): Similarly.
17099 (struct directive::set_precision): Similarly.
17100 (format_integer, format_directive, parse_directive): Similarly.
17101 (format_none): Accept unnamed vr_values parameter.
17102 (format_percent, format_floating, format_character): Similarly.
17103 (format_string, format_plain): Similarly.
17104 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
17105 the EVRP range analyzer for range data rather than using global data.
17106 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
17107 gimple-ssa-evrp-analyze.h
17108 (class sprintf_dom_walker): Add after_dom_children member function.
17109 Add evrp_range_analyzer member.
17110 (sprintf_dom_walker::before_dom_children): Call into the EVRP
17111 range analyzer as needed.
17112 (sprintf_dom_walker::after_dom_children): New member function.
17113 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
17115 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
17116 (evrp_range_analyzer::pop_to_marker): Likewise.
17118 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
17120 PR tree-optimization/84419
17121 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
17122 with the required type if its current type is compatible but
17125 2018-02-20 Jakub Jelinek <jakub@redhat.com>
17127 PR middle-end/82004
17128 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
17129 after vectorization.
17131 2018-02-20 Martin Liska <mliska@suse.cz>
17134 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
17135 possible values if we don't have a hint.
17137 2018-02-20 Martin Liska <mliska@suse.cz>
17141 * final.c (shorten_branches): Build align_tab array with one
17143 * opts.c (finish_options): Add alignment option limit check.
17144 (MAX_CODE_ALIGN): Likewise.
17145 (MAX_CODE_ALIGN_VALUE): Likewise.
17146 * doc/invoke.texi: Document maximum allowed option value for
17147 all -falign-* options.
17149 2018-02-19 Jakub Jelinek <jakub@redhat.com>
17152 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
17153 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
17154 * var-tracking.c (emit_note_insn_var_location): Remove all references
17155 to NOTE_INSN_CALL_ARG_LOCATION.
17156 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
17157 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
17158 Use copy_rtx_if_shared.
17159 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
17160 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
17161 (dwarf2out_var_location): Remove handling of
17162 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
17164 * final.c (final_scan_insn): Remove all references to
17165 NOTE_INSN_CALL_ARG_LOCATION.
17166 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
17167 before dumping final insns.
17168 * except.c (emit_note_eh_region_end): Remove all references to
17169 NOTE_INSN_CALL_ARG_LOCATION.
17170 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
17171 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
17172 * config/arc/arc.c (hwloop_optimize): Likewise.
17173 * config/arm/arm.c (create_fix_barrier): Likewise.
17174 * config/s390/s390.c (s390_chunkify_start): Likewise.
17175 * config/sh/sh.c (find_barrier): Likewise.
17176 * config/i386/i386.c (rest_of_insert_endbranch,
17177 ix86_seh_fixup_eh_fallthru): Likewise.
17178 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
17179 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
17180 * config/frv/frv.c (frv_function_prologue): Likewise.
17181 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
17183 (note_outside_basic_block_p): Remove all references to
17184 NOTE_INSN_CALL_ARG_LOCATION.
17185 * gengtype.c (adjust_field_rtx_def): Likewise.
17186 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
17188 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
17189 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
17192 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
17195 PR tree-optimization/84452
17196 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
17197 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
17200 2018-02-19 Martin Liska <mliska@suse.cz>
17203 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
17205 2018-02-19 Martin Liska <mliska@suse.cz>
17206 Richard Sandiford <richard.sandiford@linaro.org>
17208 PR tree-optimization/82491
17209 * gimple-fold.c (get_base_constructor): Make earlier bail out
17212 2018-02-19 Carl Love <cel@us.ibm.com>
17214 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
17215 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
17217 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
17218 P8V_BUILTIN_VEC_NEG.
17220 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
17222 * config/rl78/rl78.md (movdf): New define expand.
17224 2018-02-19 Martin Liska <mliska@suse.cz>
17227 * doc/invoke.texi: Fix typo.
17228 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
17230 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
17232 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
17233 handle rs6000_single_float and rs6000_double_float specially for
17236 2018-02-16 Jeff Law <law@redhat.com>
17238 * config/rx/rx.c (add_pop_cfi_notes): New function.;
17239 (pop_regs): Use it.
17241 2018-02-16 Jakub Jelinek <jakub@redhat.com>
17244 * ipa-inline.c (inline_small_functions): Fix a typo.
17246 2018-02-16 Nathan Sidwell <nathan@acm.org>
17248 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
17250 2018-02-16 Carl Love <cel@us.ibm.com>
17252 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
17253 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
17254 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
17255 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
17256 expansion to P8V_BUILTIN_VEC_FLOAT2.
17258 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
17260 PR rtl-optimization/70023
17261 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
17262 src_regno into account.
17264 2018-02-16 Carl Love <cel@us.ibm.com>
17266 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
17267 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
17268 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
17269 * config/rs6000/rs6000.c: Remove case statements for
17270 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
17271 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
17272 and P9V_BUILTIN_VEC_VINSERT4B.
17273 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
17274 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
17275 * config/rs6000/vsx.md:
17276 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
17279 2018-02-16 Carl Love <cel@us.ibm.com>
17281 * config/rs6000/altivec.h: Add builtin names vec_extract4b
17283 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
17285 * config/rs6000/rs6000-c.c: Add the definitions for
17286 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
17287 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
17288 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
17289 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
17290 definition for insert4b and define insn *insert3b_internal.
17291 * doc/extend.texi: Add documentation for vec_extract4b.
17293 2018-02-16 Nathan Sidwell <nathan@acm.org>
17295 * doc/extend.texi (Backwards Compatibility): Mention friend
17296 injection. Note for-scope is deprecated.
17297 * doc/invoke.texi (-ffriend-injection): Deprecate.
17299 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
17301 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
17302 that moved to I2, also allow destinations that are a paradoxical
17303 subreg (instead of a normal reg).
17305 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
17308 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
17311 2018-02-16 Richard Biener <rguenther@suse.de>
17313 PR tree-optimization/84037
17314 PR tree-optimization/84016
17316 * config/i386/i386.c (ix86_builtin_vectorization_cost):
17317 Adjust vec_construct for the fact we need additional higher latency
17318 128bit inserts for AVX256 and AVX512 vector builds.
17319 (ix86_add_stmt_cost): Scale vector construction cost for
17322 2018-02-16 Richard Biener <rguenther@suse.de>
17324 PR tree-optimization/84417
17325 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
17326 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
17327 (non_rewritable_lvalue_p): Likewise, use poly-ints.
17329 2018-02-16 Martin Liska <mliska@suse.cz>
17332 * internal-fn.def (ASAN_CHECK): Set proper flags.
17333 (ASAN_MARK): Likewise.
17335 2018-02-16 Julia Koval <julia.koval@intel.com>
17337 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
17338 from PTA_CANNONLAKE.
17340 2018-02-16 Jakub Jelinek <jakub@redhat.com>
17343 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
17344 Use ++iter rather than iter++ for std::list iterators.
17345 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
17346 defer deleting them until all nodes in the forest are processed. Do
17347 free even leaf nodes. Change to_process into auto_vec.
17350 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
17351 * vec.h (vec_default_construct): Use memset instead of placement new
17352 if BROKEN_VALUE_INITIALIZATION is defined.
17353 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
17354 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
17357 PR rtl-optimization/83723
17358 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
17359 * lra.c (lra_substitute_pseudo): Likewise. If true, use
17360 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
17362 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
17364 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
17366 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
17368 PR rtl-optimization/81443
17369 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
17370 from inner REGs to paradoxical SUBREGs.
17372 2018-02-16 Richard Biener <rguenther@suse.de>
17374 PR tree-optimization/84399
17375 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
17376 For operands we can analyze at their definition make sure we can
17377 analyze them at each use as well.
17379 2018-02-16 Richard Biener <rguenther@suse.de>
17381 PR tree-optimization/84190
17382 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
17383 volatile accesses if the decl isn't volatile.
17385 2018-02-15 Jason Merrill <jason@redhat.com>
17387 PR c++/84314 - ICE with templates and fastcall attribute.
17388 * attribs.c (build_type_attribute_qual_variant): Don't clobber
17389 TYPE_CANONICAL on an existing type.
17391 2018-02-15 Jakub Jelinek <jakub@redhat.com>
17393 PR tree-optimization/84383
17394 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
17395 dstoff nor call operand_equal_p if dstbase is NULL.
17397 PR tree-optimization/84334
17398 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
17399 also a CONSTANT_CLASS_P, punt.
17401 2018-02-14 Jim Wilson <jimw@sifive.com>
17403 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
17404 first SMALL_OPERAND check. New local min_second_step. Move assert
17405 to where locals are set. Add TARGET_RVC support.
17406 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
17408 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
17410 * doc/invoke.texi: Correct -Wformat-overflow code sample.
17412 2018-02-14 Martin Sebor <msebor@redhat.com>
17414 PR tree-optimization/83698
17415 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
17416 arrays constrain the offset range to their bounds.
17417 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
17418 (builtin_access::overlap): Avoid setting the size of overlap if it's
17420 (maybe_diag_overlap): Also consider arrays when deciding what values
17421 of offsets to include in diagnostics.
17423 2018-02-14 Martin Sebor <msebor@redhat.com>
17426 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
17427 that correspond to the kind of a declaration.
17429 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
17432 * config/pa/pa.md: Load address of PIC label using the linkage table
17433 if the label is nonlocal.
17435 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
17437 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
17438 warning message if user requests -maltivec=be.
17439 * doc/invoke.texi: Document deprecation of -maltivec=be.
17441 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
17444 * config/rs6000/rs6000-c.c: Update definitions for
17445 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
17446 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
17448 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
17451 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
17452 add _get_ssp intrinsics. Remove argument from
17453 __builtin_ia32_rdssp[d|q].
17454 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
17455 * config/i386/i386-builtin.def: Remove argument from
17456 __builtin_ia32_rdssp[d|q].
17457 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
17458 ix86_expand_special_args_builtin for _rdssp[d|q].
17459 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
17460 Clear register before usage.
17461 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
17462 Add documentation for new _get_ssp and _inc_ssp intrinsics.
17464 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
17466 PR tree-optimization/84357
17467 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
17468 operand 1 of an ARRAY_REF too.
17470 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
17473 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
17474 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
17476 (set_of_reg): New struct.
17477 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
17478 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
17479 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
17481 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
17482 Split into bitclr, bitset, bitinvert patterns if appropriate.
17483 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
17484 use rx_fuse_in_memory_bitop.
17485 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
17486 to named insn, correct maximum insn length.
17488 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
17491 * machmode.def: Define a complex mode for PARTIAL_INT.
17492 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
17494 * doc/rtl.texi: Document CSPImode.
17495 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
17497 (msp430_hard_regno_nregs_with_padding): Likewise.
17499 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
17502 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
17504 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
17506 PR rtl-optimization/84169
17507 * combine.c (try_combine): New variable split_i2i3. Set it to true if
17508 we generated a parallel as new i3 and we split that to new i2 and i3
17509 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
17510 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
17511 those to i2, not i1. Partially rewrite this scan code.
17513 2018-02-13 Jakub Jelinek <jakub@redhat.com>
17516 * stor-layout.c (place_field): For variable length fields, adjust
17517 offset_align afterwards not just based on the field's alignment,
17518 but also on the size.
17520 PR middle-end/84309
17521 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
17522 of exps and logs in the use_exp2 case.
17524 2018-02-13 Jeff Law <law@redhat.com>
17526 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
17527 entry for "vector".
17529 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
17532 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
17536 * common.opt (gas-loc-support, gas-locview-support): New.
17537 (ginline-points, ginternal-reset-location-views): New.
17538 * doc/invoke.texi: Document them. Use @itemx where intended.
17539 (gvariable-location-views): Adjust.
17540 * target.def (reset_location_view): New.
17541 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
17542 (TARGET_RESET_LOCATION_VIEW): New.
17543 * doc/tm.texi: Rebuilt.
17544 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
17545 (dwarf2out_default_as_locview_support): New.
17546 (output_asm_line_debug_info): Use option variables.
17547 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
17548 (output_loc_list): Likewise.
17549 (add_high_low_attributes): Check option variables.
17550 Don't output entry view attribute in strict mode.
17551 (gen_inlined_subroutine_die): Check option variables.
17552 (dwarf2out_inline_entry): Likewise.
17553 (init_sections_and_labels): Likewise.
17554 (dwarf2out_early_finish): Likewise.
17555 (maybe_reset_location_view): New, from...
17556 (dwarf2out_var_location): ... here. Call it.
17557 * debug.h (dwarf2out_default_as_loc_support): Declare.
17558 (dwarf2out_default_as_locview_support): Declare.
17559 * hooks.c (hook_int_rtx_insn_0): New.
17560 * hooks.h (hook_int_rtx_insn_0): Declare.
17561 * toplev.c (process_options): Take -gas-loc-support and
17562 -gas-locview-support from dwarf2out. Enable
17563 -gvariable-location-views by default only with locview
17564 assembler support. Enable -ginternal-reset-location-views by
17565 default only if the target defines the corresponding hook.
17566 Enable -ginline-points by default if location views are
17567 enabled; force it disabled if statement frontiers are
17569 * tree-inline.c (expand_call_inline): Check option variables.
17570 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
17572 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
17574 PR tree-optimization/84321
17575 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
17576 handling. Also check whether the anti-range contains any values
17577 that satisfy the mask; switch to a VR_RANGE if not.
17579 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
17582 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
17584 2018-02-13 Martin Jambor <mjambor@suse.cz>
17587 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
17588 of call statements, also set location of a load to a temporary.
17590 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
17592 * config/rl78/rl78.c (add_vector_labels): New function.
17593 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
17594 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
17595 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
17596 which checks that no arguments are passed.
17597 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
17598 * doc/extend.texi: Documentation for the new attribute.
17600 2018-02-13 Andreas Schwab <schwab@suse.de>
17602 * config/riscv/linux.h (CPP_SPEC): Define.
17604 2018-02-13 Jakub Jelinek <jakub@redhat.com>
17607 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
17608 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
17609 OPTION_MASK_ISA_AES as first argument to def_builtin_const
17610 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
17611 instead of OPTION_MASK_ISA_PCLMUL as first argument to
17612 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
17613 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
17614 temporarily for AES and PCLMUL builtins.
17616 PR tree-optimization/84339
17617 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
17618 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
17621 PR middle-end/84309
17622 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
17623 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
17624 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
17626 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
17628 * omp-simd-clone.h: New file.
17629 * omp-simd-clone.c: Include omp-simd-clone.h.
17630 (expand_simd_clones): No longer static.
17631 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
17632 cgraph.h and omp-simd-clone.h.
17633 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
17634 (vect_recog_widen_shift_pattern): Formatting fix.
17635 (vect_pattern_recog_1): Don't check optab for calls.
17638 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
17639 operands[2] into a REG before using gen_lowpart on it.
17641 2018-02-12 Jeff Law <law@redhat.com>
17644 * config/sh/sh.c (find_barrier): Consider a sibling call
17647 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
17648 successfully back substituting a reg.
17650 2018-02-12 Richard Biener <rguenther@suse.de>
17652 PR tree-optimization/84037
17653 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
17654 parameter, move visited init to caller.
17655 (vect_slp_analyze_operations): Separate cost from validity
17656 check, initialize visited once for all instances.
17657 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
17659 * tree-vect-stmts.c (vect_model_simple_cost): Make early
17661 (vect_model_promotion_demotion_cost): Likewise.
17662 (vectorizable_bswap): Guard cost modeling with !slp_node
17663 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
17665 (vectorizable_call): Likewise.
17666 (vectorizable_conversion): Likewise.
17667 (vectorizable_assignment): Likewise.
17668 (vectorizable_shift): Likewise.
17669 (vectorizable_operation): Likewise.
17670 (vectorizable_store): Likewise.
17671 (vectorizable_load): Likewise.
17672 (vectorizable_condition): Likewise.
17673 (vectorizable_comparison): Likewise.
17675 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
17678 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
17679 (ASAN_MARK): Fix fnspec to account for return value, change pointer
17680 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
17682 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
17684 PR middle-end/83665
17685 * params.def (inline-min-speedup): Increase from 8 to 15.
17686 (max-inline-insns-auto): Decrease from 40 to 30.
17687 * ipa-split.c (consider_split): Add some buffer for function to
17688 be considered inlining candidate.
17689 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
17692 2018-02-12 Richard Biener <rguenther@suse.de>
17694 PR tree-optimization/84037
17695 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
17696 matched stmts if we cannot swap the non-matched ones.
17698 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
17700 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
17701 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
17702 _mm_maskz_scalef_round_ss): New intrinsics.
17703 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
17704 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
17705 __builtin_ia32_scalefss_round): Remove.
17706 (__builtin_ia32_scalefsd_mask_round,
17707 __builtin_ia32_scalefss_mask_round): New intrinsics.
17708 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
17709 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
17710 ((match_operand:VF_128 2 "<round_nimm_predicate>"
17711 "<round_constraint>")): Changed to ...
17712 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
17713 "<round_scalar_constraint>")): ... this.
17714 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
17715 %0, %1, %2<round_op3>}"): Changed to ...
17716 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
17717 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
17718 %2<round_scalar_mask_op3>}"): ... this.
17719 * config/i386/subst.md (round_scalar_nimm_predicate): New.
17721 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
17723 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
17724 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
17725 (_mm_maskz_sqrt_round_ss): New intrinsics.
17726 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
17727 (__builtin_ia32_sqrtsd_mask_round)
17728 (__builtin_ia32_sqrtss_mask_round): New builtins.
17729 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
17730 (__builtin_ia32_sqrtss_round): Remove.
17731 (__builtin_ia32_sqrtsd_mask_round)
17732 (__builtin_ia32_sqrtss_mask_round): New builtins.
17733 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
17734 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
17735 ((match_operand:VF_128 1 "vector_operand"
17736 "xBm,<round_constraint>")): Changed to ...
17737 ((match_operand:VF_128 1 "vector_operand"
17738 "xBm,<round_scalar_constraint>")): ... this.
17739 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
17740 %0, %2, %<iptr>1<round_op3>}): Changed to ...
17741 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
17742 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
17743 %<iptr>1<round_scalar_mask_op3>}): ... this.
17744 ((set_attr "prefix" "<round_prefix>")): Changed to ...
17745 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
17747 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
17750 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
17751 Cast vec_cmpeq result to correct type.
17752 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
17753 Cast vec_cmpgt result to correct type.
17755 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
17757 * final.c (final_scan_insn_1): Renamed from...
17758 (final_scan_insn): ... this. New wrapper, to recover
17759 seen from the outermost call in recursive ones.
17760 * config/sparc/sparc.c (output_return): Drop seen from call.
17761 (output_sibcall): Likewise.
17762 * config/visium/visium.c (output_branch): Likewise.
17764 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
17766 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
17769 2018-02-10 Alan Modra <amodra@gmail.com>
17772 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
17773 Specify LR as an input.
17775 2018-02-10 Jakub Jelinek <jakub@redhat.com>
17778 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
17779 remove_member_access_dummy_vars): New functions.
17780 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
17781 lower_omp_1, execute_lower_omp): Use them.
17783 PR rtl-optimization/84308
17784 * shrink-wrap.c (spread_components): Release todo vector.
17786 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
17788 PR rtl-optimization/57193
17789 * ira-color.c (struct allocno_color_data): Add member
17790 conflict_allocno_hard_prefs.
17791 (update_conflict_allocno_hard_prefs): New.
17792 (bucket_allocno_compare_func): Add a preference based on
17793 conflict_allocno_hard_prefs.
17794 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
17795 (color_allocnos): Remove a dead code. Initiate
17796 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
17798 2018-02-09 Jakub Jelinek <jakub@redhat.com>
17801 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
17802 constraint from =wa to wa. Avoid a subreg on the output operand,
17803 instead use a pseudo and subreg it in a move.
17804 (p9_xxbrd_<mode>): Changed to ...
17805 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
17806 (p9_xxbrd_v2df): New expander.
17807 (p9_xxbrw_<mode>): Changed to ...
17808 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
17809 (p9_xxbrw_v4sf): New expander.
17811 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
17813 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
17815 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
17818 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
17819 multiply in 32-bit mode.
17820 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
17821 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
17824 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
17826 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
17827 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
17828 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
17829 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
17831 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
17834 * dwarf2out.c (is_trivial_indirect_ref): New function.
17835 (dwarf2out_late_global_decl): Do not generate a location
17836 attribute for variables that have a non-trivial DECL_VALUE_EXPR
17837 and that are not defined in the current unit.
17839 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
17841 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
17842 instead of a libcall for UNORDERED.
17844 2018-02-09 Tamar Christina <tamar.christina@arm.com>
17847 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
17848 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
17850 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17853 * config/s390/s390.c (s390_set_current_function): Invoke
17854 s390_indirect_branch_settings also if fndecl didn't change.
17856 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
17858 * config/rs6000/rs6000.md (blockage): Set length to zero.
17860 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
17862 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
17864 2018-02-09 Jakub Jelinek <jakub@redhat.com>
17867 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
17868 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
17872 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
17873 PARALLEL incoming that failed vt_get_decl_and_offset check.
17875 PR middle-end/84237
17876 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
17877 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
17879 (get_variable_section): For decls in named .bss* sections pass true as
17880 second argument to bss_initializer_p.
17882 2018-02-09 Marek Polacek <polacek@redhat.com>
17883 Jakub Jelinek <jakub@redhat.com>
17886 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
17887 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
17888 Sync some changes from cxx_fold_indirect_ref.
17890 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
17892 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
17894 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
17895 (BLOCK_INLINE_ENTRY_LABEL): New.
17896 (dwarf2out_var_location): Disregard inline entry markers.
17897 (inline_entry_data): New struct.
17898 (inline_entry_data_hasher): New hashtable type.
17899 (inline_entry_data_hasher::hash): New.
17900 (inline_entry_data_hasher::equal): New.
17901 (inline_entry_data_table): New variable.
17902 (add_high_low_attributes): Add DW_AT_entry_pc and
17903 DW_AT_GNU_entry_view attributes if a pending entry is found
17904 in inline_entry_data_table. Add old entry_pc attribute only
17905 if debug nonbinding markers are disabled.
17906 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
17907 markers are enabled.
17908 (block_within_block_p, dwarf2out_inline_entry): New.
17909 (dwarf2out_finish): Check that no entries remained in
17910 inline_entry_data_table.
17911 * final.c (reemit_insn_block_notes): Handle inline entry notes.
17912 (final_scan_insn, notice_source_line): Likewise.
17913 (rest_of_clean_state): Skip inline entry markers.
17914 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
17916 * gimple.c (gimple_build_debug_inline_entry): New.
17917 * gimple.h (enum gimple_debug_subcode): Add
17918 GIMPLE_DEBUG_INLINE_ENTRY.
17919 (gimple_build_debug_inline_entry): Declare.
17920 (gimple_debug_inline_entry_p): New.
17921 (gimple_debug_nonbind_marker_p): Adjust.
17922 * insn-notes.def (INLINE_ENTRY): New.
17923 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
17924 inline entry marker notes.
17925 (print_insn): Likewise.
17926 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
17927 (INSN_DEBUG_MARKER_KIND): Likewise.
17928 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
17929 * tree-inline.c (expand_call_inline): Build and insert
17930 debug_inline_entry stmt.
17931 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
17932 inline entry blocks early, if nonbind markers are enabled.
17933 (dump_scope_block): Dump fragment info.
17934 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
17935 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
17936 (gimple_build_debug_inline_entry): New.
17937 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
17938 Enable/disable inline entry points too.
17939 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
17940 (DEBUG_INSN): Describe inline entry markers.
17942 * common.opt (gvariable-location-views): New.
17943 (gvariable-location-views=incompat5): New.
17944 * config.in: Rebuilt.
17945 * configure: Rebuilt.
17946 * configure.ac: Test assembler for view support.
17947 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
17948 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
17949 * dwarf2out.c (var_loc_view): New typedef.
17950 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
17951 (dwarf2out_locviews_in_attribute): New.
17952 (dwarf2out_locviews_in_loclist): New.
17953 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
17954 (enum dw_line_info_opcode): Add LI_adv_address.
17955 (struct dw_line_info_table): Add view.
17956 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
17957 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
17958 (zero_view_p): New variable.
17959 (ZERO_VIEW_P): New macro.
17960 (output_asm_line_debug_info): New.
17961 (struct var_loc_node): Add view.
17962 (add_AT_view_list, AT_loc_list): New.
17963 (add_var_loc_to_decl): Add view param. Test it against last.
17964 (new_loc_list): Add view params. Record them.
17965 (AT_loc_list_ptr): Handle loc and view lists.
17966 (view_list_to_loc_list_val_node): New.
17967 (print_dw_val): Handle dw_val_class_view_list.
17968 (size_of_die): Likewise.
17969 (value_format): Likewise.
17970 (loc_list_has_views): New.
17971 (gen_llsym): Set vl_symbol too.
17972 (maybe_gen_llsym, skip_loc_list_entry): New.
17973 (dwarf2out_maybe_output_loclist_view_pair): New.
17974 (output_loc_list): Output view list or entries too.
17975 (output_view_list_offset): New.
17976 (output_die): Handle dw_val_class_view_list.
17977 (output_dwarf_version): New.
17978 (output_compilation_unit_header): Use it.
17979 (output_skeleton_debug_sections): Likewise.
17980 (output_rnglists, output_line_info): Likewise.
17981 (output_pubnames, output_aranges): Update version comments.
17982 (output_one_line_info_table): Output view numbers in asm comments.
17983 (dw_loc_list): Determine current endview, pass it to new_loc_list.
17984 Call maybe_gen_llsym.
17985 (loc_list_from_tree_1): Adjust.
17986 (add_AT_location_description): Create view list attribute if
17987 needed, check it's absent otherwise.
17988 (convert_cfa_to_fb_loc_list): Adjust.
17989 (maybe_emit_file): Call output_asm_line_debug_info for test.
17990 (dwarf2out_var_location): Reset views as needed. Precompute
17991 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
17992 attribute. Set view.
17993 (new_line_info_table): Reset next view.
17994 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
17995 (dwarf2out_source_line): Likewise. Output view resets and labels to
17996 the assembler, or select appropriate line info opcodes.
17997 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
17998 (optimize_string_length): Catch it. Adjust.
17999 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
18000 dw_val_class_view_list, and remove it if no longer needed.
18001 (hash_loc_list): Hash view numbers.
18002 (loc_list_hasher::equal): Compare them.
18003 (optimize_location_lists): Check whether a view list symbol is
18004 needed, and whether the locview attribute is present, and
18005 whether they match. Remove the locview attribute if no longer
18007 (index_location_lists): Call skip_loc_list_entry for test.
18008 (dwarf2out_finish): Call output_asm_line_debug_info for test.
18009 Use output_dwarf_version.
18010 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
18011 (struct dw_val_node): Add val_view_list.
18012 * final.c (SEEN_NEXT_VIEW): New.
18013 (set_next_view_needed): New.
18014 (clear_next_view_needed): New.
18015 (maybe_output_next_view): New.
18016 (final_start_function): Rename to...
18017 (final_start_function_1): ... this. Take pointer to FIRST,
18018 add SEEN parameter. Emit param bindings in the initial view.
18019 (final_start_function): Reintroduce SEEN-less interface.
18020 (final): Rename to...
18021 (final_1): ... this. Take SEEN parameter. Output final pending
18022 next view at the end.
18023 (final): Reintroduce seen-less interface.
18024 (final_scan_insn): Output pending next view before switching
18025 sections or ending a block. Mark the next view as needed when
18026 outputting variable locations. Notify debug backend of section
18027 changes, and of location view changes.
18028 (rest_of_handle_final): Adjust.
18029 * toplev.c (process_options): Autodetect value for debug variable
18030 location views option. Warn on incompat5 without -gdwarf-5.
18031 * doc/invoke.texi (gvariable-location-views): New.
18032 (gvariable-location-views=incompat5): New.
18033 (gno-variable-location-views): New.
18035 2018-02-08 David Malcolm <dmalcolm@redhat.com>
18037 PR tree-optimization/84136
18038 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
18039 that the result of find_edge is non-NULL.
18041 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
18044 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
18045 storing integer register in SImode. Fix cost of 256 and 512
18046 byte aligned SSE register store.
18048 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
18050 * config/i386/i386.c (ix86_multiplication_cost): Fix
18051 multiplication cost for TARGET_AVX512DQ.
18053 2018-02-08 Marek Polacek <polacek@redhat.com>
18055 PR tree-optimization/84238
18056 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
18059 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
18061 PR tree-optimization/84265
18062 * tree-vect-stmts.c (vectorizable_store): Don't treat
18063 VMAT_CONTIGUOUS accesses as grouped.
18064 (vectorizable_load): Likewise.
18066 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
18068 PR tree-optimization/81635
18069 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
18070 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
18071 (test_round_for_mask): New functions.
18072 (wide_int_cc_tests): Call test_round_for_mask.
18073 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
18074 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
18075 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
18076 range returned by get_range_info.
18078 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
18081 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
18082 * symtab.c: Include builtins.h
18083 (symtab_node::output_to_lto_symbol_table_p): Move here
18084 from lto-streamer-out.c:output_symbol_p.
18085 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
18086 (output_symbol_p): Move all logic to symtab.c
18087 (produce_symtab): Update.
18089 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18091 * config/s390/s390-opts.h (enum indirect_branch): Define.
18092 * config/s390/s390-protos.h (s390_return_addr_from_memory)
18093 (s390_indirect_branch_via_thunk)
18094 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
18095 (enum s390_indirect_branch_type): Define.
18096 * config/s390/s390.c (struct s390_frame_layout, struct
18097 machine_function): Remove.
18098 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
18099 (indirect_branch_table_label_no, indirect_branch_table_name):
18101 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
18102 (enum s390_indirect_branch_option): Define.
18103 (s390_return_addr_from_memory): New function.
18104 (s390_handle_string_attribute): New function.
18105 (s390_attribute_table): Add new attribute handler.
18106 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
18107 (s390_indirect_branch_via_thunk): New function.
18108 (s390_indirect_branch_via_inline_thunk): New function.
18109 (s390_function_ok_for_sibcall): When jumping via thunk disallow
18110 sibling call optimization for non z10 compiles.
18111 (s390_emit_call): Force indirect branch target to be a single
18112 register. Add r1 clobber for non-z10 compiles.
18113 (s390_emit_epilogue): Emit return jump via return_use expander.
18114 (s390_reorg): Handle JUMP_INSNs as execute targets.
18115 (s390_option_override_internal): Perform validity checks for the
18116 new command line options.
18117 (s390_indirect_branch_attrvalue): New function.
18118 (s390_indirect_branch_settings): New function.
18119 (s390_set_current_function): Invoke s390_indirect_branch_settings.
18120 (s390_output_indirect_thunk_function): New function.
18121 (s390_code_end): Implement target hook.
18122 (s390_case_values_threshold): Implement target hook.
18123 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
18125 * config/s390/s390.h (struct s390_frame_layout)
18126 (struct machine_function): Move here from s390.c.
18127 (TARGET_INDIRECT_BRANCH_NOBP_RET)
18128 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
18129 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
18130 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
18131 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
18132 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
18133 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
18134 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
18135 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
18136 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
18137 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
18138 (mnemonic attribute): Add values which aren't recognized
18140 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
18141 pattern for branch conversion. Fix mnemonic attribute.
18142 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
18143 indirect branch via thunk if requested.
18144 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
18145 ("*indirect_jump"): Disable for branch conversion using out of
18147 ("indirect_jump_via_thunk<mode>_z10")
18148 ("indirect_jump_via_thunk<mode>")
18149 ("indirect_jump_via_inlinethunk<mode>_z10")
18150 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
18151 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
18152 ("casesi_jump_via_inlinethunk<mode>_z10")
18153 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
18154 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
18155 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
18156 ("*indirect2_jump"): Disable for branch conversion.
18157 ("casesi_jump"): Turn into expander and expand patterns for branch
18159 ("return_use"): New expander.
18160 ("*return"): Emit return via thunk and rename it to ...
18161 ("*return<mode>"): ... this one.
18162 * config/s390/s390.opt: Add new options and and enum for the
18165 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
18167 * lra-constraints.c (match_reload): Unconditionally use
18168 gen_lowpart_SUBREG, rather than selecting between that
18169 and equivalent gen_rtx_SUBREG code.
18171 2018-02-08 Richard Biener <rguenther@suse.de>
18173 PR tree-optimization/84233
18174 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
18175 changed flag instead of boguously re-using phi_inserted.
18177 2018-02-08 Martin Jambor <mjambor@suse.cz>
18179 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
18180 static local variables.
18182 2018-02-08 Richard Biener <rguenther@suse.de>
18184 PR tree-optimization/84278
18185 * tree-vect-stmts.c (vectorizable_store): When looking for
18186 smaller vector types to perform grouped strided loads/stores
18187 make sure the mode is supported by the target.
18188 (vectorizable_load): Likewise.
18190 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
18192 * config/aarch64/aarch64.c (aarch64_components_for_bb):
18193 Increase LDP/STP opportunities by adding adjacent callee-saves.
18195 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
18197 PR rtl-optimization/84068
18198 PR rtl-optimization/83459
18199 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
18201 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
18203 PR tree-optimization/84224
18204 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
18205 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
18206 non-zero arguments.
18208 2018-02-07 Iain Sandoe <iain@codesourcery.com>
18211 * config/rs6000/altivec.md (*restore_world): Remove LR use.
18212 * config/rs6000/predicates.md (restore_world_operation): Adjust op
18213 count, remove one USE.
18215 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18217 * doc/install.texi (Configuration): Document the
18218 --with-long-double-format={ibm,ieee} PowerPC configuration
18222 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
18223 Convert from define_expand to be define_insn_and_split. Rework
18224 float/double/_Float128 conversions to QI/HI/SImode to work with
18225 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
18226 conversions to QI/HImode types did a store and then a load to
18227 truncate the value. For conversions to VSX registers, don't split
18228 the insn, instead emit the code directly. Use the code iterator
18229 any_fix to combine signed and unsigned conversions.
18230 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
18231 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
18232 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
18233 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
18234 (fix_<mode>di2_hw): Likewise.
18235 (fixuns_<mode>di2_hw): Likewise.
18236 (fix_<mode>si2_hw): Likewise.
18237 (fixuns_<mode>si2_hw): Likewise.
18238 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
18239 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
18240 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
18241 fix<uns>_trunc<SFDF:mode>si2_p8.
18242 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
18243 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
18244 (fix<uns>_<mode>_mem): Likewise.
18245 (fctiw<u>z_<mode>_mem): Likewise.
18246 (fix<uns>_<mode>_mem): Likewise.
18247 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
18248 the register allocator from doing a direct move to the GPRs to do
18249 a store, and instead use the ISA 3.0 store byte/half-word from
18250 vector register instruction. For IEEE 128-bit floating point,
18251 also optimize stores of 32-bit ints.
18252 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
18254 2018-02-07 Alan Hayward <alan.hayward@arm.com>
18256 * genextract.c (push_pathstr_operand): New function to support
18258 (walk_rtx): Call push_pathstr_operand.
18259 (print_path): Support [a-zA-Z].
18261 2018-02-07 Richard Biener <rguenther@suse.de>
18263 PR tree-optimization/84037
18264 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
18265 (cse_and_gimplify_to_preheader): Declare.
18266 (vect_get_place_in_interleaving_chain): Likewise.
18267 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
18269 (_loop_vec_info::~_loop_vec_info): Delete it.
18270 (cse_and_gimplify_to_preheader): New function.
18271 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
18272 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
18273 (vectorizable_load): Likewise. For grouped stores always base
18274 the IV on the first element.
18275 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
18276 condition before gimplifying.
18278 2018-02-07 Jakub Jelinek <jakub@redhat.com>
18280 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
18281 *DIV_EXPR and *MOD_EXPR.
18283 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
18286 * config/i386/i386.c (ix86_option_override_internal): Mask out
18287 the CF_SET bit when checking -fcf-protection.
18289 2018-02-07 Tom de Vries <tom@codesourcery.com>
18292 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
18295 2018-02-07 Richard Biener <rguenther@suse.de>
18297 PR tree-optimization/84204
18298 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
18301 PR tree-optimization/84205
18302 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
18303 special-case isl_ast_op_zdiv_r.
18305 PR tree-optimization/84223
18306 * graphite-scop-detection.c (gather_bbs::before_dom_children):
18307 Only add conditions from within the region.
18308 (gather_bbs::after_dom_children): Adjust.
18310 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
18313 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
18314 * config/avr/avr.md: Only post-reload split REG-REG moves if
18315 either register is GENERAL_REG_P.
18317 2018-02-07 Jakub Jelinek <jakub@redhat.com>
18319 PR tree-optimization/84235
18320 * tree-ssa-scopedtables.c
18321 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
18322 if the subtraction is performed in floating point type where NaNs are
18323 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
18324 build 1. Formatting fix.
18326 2018-02-06 Jakub Jelinek <jakub@redhat.com>
18329 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
18330 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
18331 and skip it regardless of bb boundaries. Use CALL_P macro,
18332 don't test INSN_P (insn) together with CALL_P or JUMP_P check
18333 unnecessarily, formatting fix.
18335 2018-02-06 Michael Collison <michael.collison@arm.com>
18337 * config/arm/thumb2.md:
18338 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
18339 (*thumb_mov_notscc): Ditto.
18341 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
18344 * config/rs6000/rs6000.md (su code attribute): Use "u" for
18345 unsigned_fix, not "s".
18347 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18349 * configure.ac (gcc_fn_eh_frame_ro): New function.
18350 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
18351 correct .eh_frame permissions.
18352 * configure: Regenerate.
18354 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
18356 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
18357 irrelevant options.
18359 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18361 * config/rs6000/rs6000.c (rs6000_option_override_internal):
18362 Display warning message for -mno-speculate-indirect-jumps.
18364 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
18366 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
18368 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
18370 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
18372 PR tree-optimization/84225
18373 * tree-eh.c (find_trapping_overflow): Only call
18374 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
18376 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
18379 * config/i386/i386.c: Reimplement the check of possible options
18380 -mibt/-mshstk conbination. Change error messages.
18381 * doc/invoke.texi: Fix a typo: remove extra '='.
18383 2018-02-06 Marek Polacek <polacek@redhat.com>
18385 PR tree-optimization/84228
18386 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
18388 2018-02-06 Tamar Christina <tamar.christina@arm.com>
18391 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
18392 emitted arch directives.
18393 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
18394 __ARM_FEATURE_COPROC before changing architectures.
18396 2018-02-06 Richard Biener <rguenther@suse.de>
18398 * config/i386/i386.c (print_reg): Fix typo.
18399 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
18401 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
18403 * configure: Regenerate.
18405 2018-02-05 Martin Sebor <msebor@redhat.com>
18407 PR tree-optimization/83369
18408 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
18411 2018-02-05 Martin Liska <mliska@suse.cz>
18413 * doc/invoke.texi: Cherry-pick upstream r323995.
18415 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
18417 * ira.c (ira_init_register_move_cost): Adjust comment.
18419 2018-02-05 Martin Liska <mliska@suse.cz>
18421 PR gcov-profile/84137
18422 * doc/gcov.texi: Fix typo in documentation.
18424 2018-02-05 Martin Liska <mliska@suse.cz>
18426 PR gcov-profile/83879
18427 * doc/gcov.texi: Document necessity of --dynamic-list-data when
18428 using dlopen functionality.
18430 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
18432 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
18433 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
18434 _mm_maskz_range_ss, _mm_mask_range_round_ss,
18435 _mm_maskz_range_round_ss): New intrinsics.
18436 (__builtin_ia32_rangesd128_round)
18437 (__builtin_ia32_rangess128_round): Remove.
18438 (__builtin_ia32_rangesd128_mask_round,
18439 __builtin_ia32_rangess128_mask_round): New builtins.
18440 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
18441 __builtin_ia32_rangess128_round): Remove.
18442 (__builtin_ia32_rangesd128_mask_round,
18443 __builtin_ia32_rangess128_mask_round): New builtins.
18444 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
18445 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
18446 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
18447 "<round_saeonly_constraint>")): Changed to ...
18448 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
18449 "<round_saeonly_scalar_constraint>")): ... this.
18450 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
18451 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
18452 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
18453 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
18454 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
18456 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
18458 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
18460 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
18461 Remove all values except native, 8540 and 8548.
18463 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
18465 * config/i386/i386.c (ix86_output_function_return): Pass
18466 INVALID_REGNUM, instead of -1, as invalid register number to
18467 indirect_thunk_name and output_indirect_thunk.
18469 2018-02-02 Julia Koval <julia.koval@intel.com>
18471 * config.gcc: Add -march=icelake.
18472 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
18473 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
18474 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
18475 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
18476 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
18477 (processor_target_table): Add icelake.
18478 (ix86_option_override_internal): Handle new PTAs.
18479 (get_builtin_code_for_version): Handle icelake.
18480 (M_INTEL_COREI7_ICELAKE): New.
18481 (fold_builtin_cpu): Handle icelake.
18482 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
18483 * doc/invoke.texi: Add -march=icelake.
18485 2018-02-02 Julia Koval <julia.koval@intel.com>
18487 * config/i386/i386.c (ix86_option_override_internal): Change flags type
18488 to wide_int_bitmask.
18489 * wide-int-bitmask.h: New.
18491 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
18494 * config/i386/i386.md: Replace Pmode with word_mode in
18495 builtin_setjmp_setup and builtin_longjmp to support x32.
18497 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
18501 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
18503 (rs6000_supported_cpu_names): New static variable.
18504 (linux_cpu_translation_table): Likewise.
18505 (elf_platform) <cpu>: Define new static variable and use it.
18506 Translate kernel AT_PLATFORM name to canonical name if needed.
18507 Error if platform name is unknown.
18509 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
18512 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
18514 2018-02-01 Jeff Law <law@redhat.com>
18517 * config/i386/i386.c (release_scratch_register_on_entry): Add new
18518 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
18519 the scratch if RELEASE_VIA_POP is false.
18520 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
18521 If we have to save a temporary register, decrement SIZE appropriately.
18522 Pass new arguments to release_scratch_register_on_entry.
18523 (ix86_adjust_stack_and_probe): Likewise.
18524 (ix86_emit_probe_stack_range): Pass new arguments to
18525 release_scratch_register_on_entry.
18527 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
18529 PR rtl-optimization/84157
18530 * combine.c (change_zero_ext): Use REG_P predicate in
18531 front of HARD_REGISTER_P predicate.
18533 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
18535 * config/avr/avr.c (avr_option_override): Move disabling of
18536 -fdelete-null-pointer-checks to...
18537 * common/config/avr/avr-common.c (avr_option_optimization_table):
18540 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
18542 PR tree-optimization/81635
18543 * tree-data-ref.c (split_constant_offset_1): For types that
18544 wrap on overflow, try to use range info to prove that wrapping
18547 2018-02-01 Renlin Li <renlin.li@arm.com>
18550 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
18551 TAILCALL_ADDR_REGS.
18552 (aarch64_register_move_cost): Likewise.
18553 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
18554 TAILCALL_ADDR_REGS.
18555 (REG_CLASS_NAMES): Likewise.
18556 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
18557 TAILCALL_ADDR_REGS. Remove IP registers.
18558 * config/aarch64/aarch64.md (Ucs): Update register constraint.
18560 2018-02-01 Richard Biener <rguenther@suse.de>
18562 * domwalk.h (dom_walker::dom_walker): Add additional constructor
18563 for specifying RPO order and allow NULL for that.
18564 * domwalk.c (dom_walker::dom_walker): Likewise.
18565 (dom_walker::walk): Handle NULL RPO order.
18566 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
18568 (rewrite_update_dom_walker): Likewise.
18569 (mark_def_dom_walker): Likewise.
18571 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
18573 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
18574 (aarch64_maybe_expand_sve_subreg_move): Declare.
18575 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
18576 * config/aarch64/predicates.md (aarch64_any_register_operand): New
18578 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
18579 that are semantically a reverse operation.
18580 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
18581 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
18582 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
18584 (aarch64_can_change_mode_class): For big-endian, forbid changes
18585 between two SVE modes if they have different element sizes.
18587 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
18589 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
18590 the TImode handling for big-endian targets.
18592 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
18594 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
18595 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
18597 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
18598 Remove BSWAP handing for big-endian targets and use the form of
18599 LD1RQ appropariate for the mode.
18601 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
18603 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
18604 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
18605 duplicated element.
18607 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
18610 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
18611 check for operands that need to go through aarch64_sve_reload_be.
18613 2018-02-01 Jakub Jelinek <jakub@redhat.com>
18615 PR tree-optimization/81661
18616 PR tree-optimization/84117
18617 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
18618 * tree-eh.c: Include gimplify.h.
18619 (find_trapping_overflow, replace_trapping_overflow,
18620 rewrite_to_non_trapping_overflow): New functions.
18621 * tree-vect-loop.c: Include tree-eh.h.
18622 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
18623 * tree-data-ref.c: Include tree-eh.h.
18624 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
18626 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
18628 PR rtl-optimization/84123
18629 * combine.c (change_zero_ext): Check if hard register satisfies
18630 can_change_dest_mode before calling gen_lowpart_SUBREG.
18632 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
18635 * ira.c (ira_init_register_move_cost): Remove assert.
18637 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
18639 PR rtl-optimization/84071
18640 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
18641 * doc/tm.texi: Regenerate.
18643 2018-01-31 Richard Biener <rguenther@suse.de>
18645 PR tree-optimization/84132
18646 * tree-data-ref.c (analyze_miv_subscript): Properly
18647 check whether evolution_function_is_affine_multivariate_p
18648 before calling gcd_of_steps_may_divide_p.
18650 2018-01-31 Julia Koval <julia.koval@intel.com>
18653 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
18654 * config/i386/i386.md (rdpid_rex64) New.
18655 (rdpid): Make 32bit only.
18657 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
18660 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
18661 an IDENTIFIER_NODE for FUNCTION_TYPE's.
18663 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
18666 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
18668 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
18670 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
18672 PR rtl-optimization/84071
18673 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
18674 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
18676 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
18678 * config/arc/arc.c (arc_handle_aux_attribute): New function.
18679 (arc_attribute_table): Add 'aux' attribute.
18680 (arc_in_small_data_p): Consider aux like variables.
18681 (arc_is_aux_reg_p): New function.
18682 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
18683 (arc_get_aux_arg): New function.
18684 (prepare_move_operands): Handle aux-register access.
18685 (arc_handle_aux_attribute): New function.
18686 * doc/extend.texi (ARC Variable attributes): Add subsection.
18688 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
18690 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
18691 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
18692 (arc_attribute_table): Add 'uncached' attribute.
18693 (arc_print_operand): Print '.di' flag for uncached memory
18695 (arc_in_small_data_p): Do not consider for small data the uncached
18697 (arc_is_uncached_mem_p): New function.
18698 * config/arc/predicates.md (compact_store_memory_operand): Check
18699 for uncached memory accesses.
18700 (nonvol_nonimm_operand): Likewise.
18701 * doc/extend.texi (ARC Type Attribute): New subsection.
18703 2018-01-31 Jakub Jelinek <jakub@redhat.com>
18706 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
18707 falign-loops=): Add Optimization flag.
18709 2018-01-30 Jeff Law <law@redhat.com>
18712 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
18713 INT_REGISTERS_SAVED. Check it prior to calling
18714 get_scratch_register_on_entry.
18715 (ix86_adjust_stack_and_probe): Similarly.
18716 (ix86_emit_probe_stack_range): Similarly.
18717 (ix86_expand_prologue): Corresponding changes.
18719 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18722 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
18723 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
18725 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
18728 * lra-constraints.c (curr_insn_transform): Process AND in the
18731 2018-01-30 Jakub Jelinek <jakub@redhat.com>
18733 PR rtl-optimization/83986
18734 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
18735 dependence against last_pending_memory_flush in addition to
18736 pending_jump_insns.
18738 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
18740 PR tree-optimization/81611
18741 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
18744 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18747 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
18750 2018-01-30 Richard Biener <rguenther@suse.de>
18751 Jakub Jelinek <jakub@redhat.com>
18753 PR tree-optimization/84111
18754 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
18755 inner loops added during recursion, as they don't have up-to-date
18758 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
18761 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
18762 (can_inline_edge_by_limits_p): ... here.
18763 (can_early_inline_edge_p, check_callers,
18764 update_caller_keys, update_callee_keys, recursive_inlining,
18765 add_new_edges_to_heap, speculation_useful_p,
18766 inline_small_functions,
18767 inline_small_functions, flatten_function,
18768 inline_to_all_callers_1): Update.
18770 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
18772 * profile-count.c (profile_count::combine_with_ipa_count): Handle
18775 2018-01-30 Richard Biener <rguenther@suse.de>
18777 PR tree-optimization/83008
18778 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
18779 invariant and constant vector uses in stmts when they need
18780 more than one stmt.
18782 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18785 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
18786 * configure: Regenerate.
18788 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
18790 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
18792 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
18793 Use gen_rtx_REG rather than gen_lowpart.
18795 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
18797 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
18798 rather than 0 when creating partial subregs.
18800 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
18802 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
18805 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
18808 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
18809 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
18810 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
18811 flags. This restores the settings used before the 2017-07-24.
18812 Turning off pre increment/decrement/modify allows IVOPTS to
18813 optimize DF/SF loops where the index is an int.
18815 2018-01-29 Richard Biener <rguenther@suse.de>
18816 Kelvin Nilsen <kelvin@gcc.gnu.org>
18819 * tree-vect-stmts.c (vectorizable_call): Don't call
18820 targetm.vectorize_builtin_md_vectorized_function if callee is
18823 2018-01-22 Carl Love <cel@us.ibm.com>
18825 * doc/extend.tex: Fix typo in second arg in
18826 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
18828 2018-01-29 Richard Biener <rguenther@suse.de>
18830 PR tree-optimization/84086
18831 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
18832 (flush_ssaname_freelist): When SSA names were released reset
18833 the SCEV hash table.
18835 2018-01-29 Richard Biener <rguenther@suse.de>
18837 PR tree-optimization/84057
18838 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
18839 removed paths when removing edges.
18841 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
18843 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
18844 -mfunction-return=@var{choice}.
18846 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
18848 PR diagnostic/84034
18849 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
18850 Handle CR like TAB.
18851 (layout::print_source_line): Likewise.
18852 (test_get_line_width_without_trailing_whitespace): Add test cases.
18854 2018-01-27 Jakub Jelinek <jakub@redhat.com>
18856 PR middle-end/84040
18857 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
18860 2018-01-26 Jim Wilson <jimw@sifive.com>
18862 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
18864 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
18867 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18869 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
18870 and CMP + SUB-immediate -> SUBS.
18872 2018-01-26 Martin Sebor <msebor@redhat.com>
18874 PR tree-optimization/83896
18875 * tree-ssa-strlen.c (get_string_len): Rename...
18876 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
18877 Avoid assuming length is constant.
18878 (handle_char_store): Use HOST_WIDE_INT for string length.
18880 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
18883 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
18884 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
18886 2018-01-26 Richard Biener <rguenther@suse.de>
18888 PR rtl-optimization/84003
18889 * dse.c (record_store): Only record redundant stores when
18890 the earlier store aliases at least all accesses the later one does.
18892 2018-01-26 Jakub Jelinek <jakub@redhat.com>
18894 PR rtl-optimization/83985
18895 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
18896 REG_CFA_RESTORE insns.
18897 (delete_unmarked_insns): Don't ignore separate shrink wrapping
18898 REG_CFA_RESTORE insns here.
18901 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
18902 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
18904 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
18906 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
18907 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
18908 (arc_init): Likewise.
18909 (arc_override_options): Likewise.
18910 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
18912 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
18914 * config/arc/arc.h (TARGET_DBNZ): Define.
18915 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
18916 properly set the tune attribute.
18917 (dbnz): Use TARGET_DBNZ guard.
18918 * config/arc/arc.opt (mtune): Add core3 option.
18920 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
18922 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
18923 recognize new pic like addresses.
18924 (arc_delegitimize_address): Clean up.
18926 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
18928 * config/arc/arc-arches.def: Option mrf16 valid for all
18930 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
18931 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
18932 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
18933 * config/arc/arc-tables.opt: Regenerate.
18934 * config/arc/arc.c (arc_conditional_register_usage): Handle
18935 reduced register file case.
18936 (arc_file_start): Set must have build attributes.
18937 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
18938 mrf16 option value.
18939 * config/arc/arc.opt (mrf16): Add new option.
18940 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
18941 * config/arc/genmultilib.awk: Handle new mrf16 option.
18942 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
18943 * config/arc/t-multilib: Regenerate.
18944 * doc/invoke.texi (ARC Options): Document mrf16 option.
18946 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
18948 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
18949 * config/arc/arc.c (arc_handle_secure_attribute): New function.
18950 (arc_attribute_table): Add 'secure_call' attribute.
18951 (arc_print_operand): Print secure call operand.
18952 (arc_function_ok_for_sibcall): Don't optimize tail calls when
18954 (arc_is_secure_call_p): New function. * config/arc/arc.md
18955 (call_i): Add support for sjli instruction.
18956 (call_value_i): Likewise.
18957 * config/arc/constraints.md (Csc): New constraint.
18959 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
18960 John Eric Martin <John.Martin@emmicro-us.com>
18962 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
18963 * config/arc/arc.c (_arc_jli_section): New struct.
18964 (arc_jli_section): New type.
18965 (rc_jli_sections): New static variable.
18966 (arc_handle_jli_attribute): New function.
18967 (arc_attribute_table): Add jli_always and jli_fixed attribute.
18968 (arc_file_end): New function.
18969 (TARGET_ASM_FILE_END): Define.
18970 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
18971 (arc_add_jli_section): New function.
18972 (jli_call_scan): Likewise.
18973 (arc_reorg): Call jli_call_scan.
18974 (arc_output_addsi): Remove 'S' from printing asm operand.
18975 (arc_is_jli_call_p): New function.
18976 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
18978 (movhi_insn): Likewise.
18979 (movsi_insn): Likewise.
18980 (movsi_set_cc_insn): Likewise.
18981 (loadqi_update): Likewise.
18982 (load_zeroextendqisi_update): Likewise.
18983 (load_signextendqisi_update): Likewise.
18984 (loadhi_update): Likewise.
18985 (load_zeroextendhisi_update): Likewise.
18986 (load_signextendhisi_update): Likewise.
18987 (loadsi_update): Likewise.
18988 (loadsf_update): Likewise.
18989 (movsicc_insn): Likewise.
18990 (bset_insn): Likewise.
18991 (bxor_insn): Likewise.
18992 (bclr_insn): Likewise.
18993 (bmsk_insn): Likewise.
18994 (bicsi3_insn): Likewise.
18995 (cmpsi_cc_c_insn): Likewise.
18996 (movsi_ne): Likewise.
18997 (movsi_cond_exec): Likewise.
18998 (clrsbsi2): Likewise.
18999 (norm_f): Likewise.
19009 (call_i): Remove 'S' asm letter, add jli instruction.
19010 (call_value_i): Likewise.
19011 * config/arc/arc.op (mjli-always): New option.
19012 * config/arc/constraints.md (Cji): New constraint.
19013 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
19015 (subsf3_fpx): Likewise.
19016 (mulsf3_fpx): Likewise.
19017 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
19019 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
19020 function attrbutes.
19021 * doc/invoke.texi (ARC): Document mjli-always option.
19023 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
19025 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
19026 avoid addition with 0 and use incw and decw where possible.
19028 2018-01-26 Richard Biener <rguenther@suse.de>
19030 PR tree-optimization/81082
19031 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
19032 association if it requires casting to unsigned.
19033 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
19034 from fold_plusminus_mult_expr to catch important cases late when
19035 range info is available.
19037 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19039 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
19040 * configure.ac (hidden_linkonce): New test.
19041 * configure: Regenerate.
19042 * config.in: Regenerate.
19044 2018-01-26 Julia Koval <julia.koval@intel.com>
19046 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
19047 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
19048 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
19049 _mm_mask_bitshuffle_epi64_mask): Fix type.
19050 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
19051 USI_FTYPE_V4DI_V4DI_USI): Remove.
19052 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
19053 __builtin_ia32_vpshufbitqmb256_mask,
19054 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
19055 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
19056 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
19058 2018-01-26 Alan Modra <amodra@gmail.com>
19061 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
19062 UNSPEC_VBPERMQ. Sort other unspecs.
19064 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
19066 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
19068 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
19070 PR middle-end/83055
19071 * predict.c (drop_profile): Do not push/pop cfun; update also
19073 (handle_missing_profiles): Fix logic looking for zero profiles.
19075 2018-01-25 Jakub Jelinek <jakub@redhat.com>
19077 PR middle-end/83977
19078 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
19079 on functions with #pragma omp declare simd or functions with simd
19081 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
19082 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
19083 Remove trailing \n from warning_at calls.
19085 2018-01-25 Tom de Vries <tom@codesourcery.com>
19088 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
19089 for neutered workers.
19091 2018-01-24 Joseph Myers <joseph@codesourcery.com>
19094 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
19095 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
19097 2018-01-24 Jeff Law <law@redhat.com>
19100 * i386.c (get_probe_interval): Move to earlier point.
19101 (ix86_compute_frame_layout): If -fstack-clash-protection and
19102 the frame is larger than the probe interval, then use pushes
19103 to save registers rather than reg->mem moves.
19104 (ix86_expand_prologue): Remove conditional for int_registers_saved
19107 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
19110 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
19111 min/max for never referenced object.
19113 2018-01-24 Jakub Jelinek <jakub@redhat.com>
19115 PR middle-end/83977
19116 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
19118 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
19119 attributes from DECL_ATTRIBUTES (decl) without affecting
19120 DECL_ATTRIBUTES (current_function_decl).
19121 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
19122 functions with non-NULL DECL_ABSTRACT_ORIGIN.
19124 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
19126 PR tree-optimization/83979
19127 * fold-const.c (fold_comparison): Use constant_boolean_node
19128 instead of boolean_{true,false}_node.
19130 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
19132 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
19135 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19137 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
19138 Simplify the clause that sets the length attribute.
19139 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
19140 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
19141 clause that sets the length attribute.
19142 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
19144 2018-01-24 Tom de Vries <tom@codesourcery.com>
19147 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
19148 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
19149 Add strict parameter.
19150 (prevent_branch_around_nothing): Insert dummy insn between branch to
19151 label and label with no ptx insn inbetween.
19152 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
19154 2018-01-24 Tom de Vries <tom@codesourcery.com>
19157 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
19158 for neutered threads in warp.
19159 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
19161 2018-01-24 Richard Biener <rguenther@suse.de>
19163 PR tree-optimization/83176
19164 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
19167 2018-01-24 Richard Biener <rguenther@suse.de>
19169 PR tree-optimization/82819
19170 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
19171 code generating pluses that are no-ops in the target precision.
19173 2018-01-24 Richard Biener <rguenther@suse.de>
19175 PR middle-end/84000
19176 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
19178 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
19180 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
19181 to merge probabilities.
19182 * predict.c (probably_never_executed): Also mark as cold functions
19183 with global 0 profile and guessed local profile.
19184 * profile-count.c (profile_probability::combine_with_count): New
19186 * profile-count.h (profile_probability::operator*,
19187 profile_probability::operator*=, profile_probability::operator/,
19188 profile_probability::operator/=): Reduce precision to adjusted
19189 and set value to guessed on contradictory divisions.
19190 (profile_probability::combine_with_freq): Remove.
19191 (profile_probability::combine_wiht_count): Declare.
19192 (profile_count::force_nonzero):: Set to adjusted.
19193 (profile_count::probability_in):: Set quality to adjusted.
19194 * tree-ssa-tail-merge.c (replace_block_by): Use
19195 combine_with_count.
19197 2018-01-23 Andrew Waterman <andrew@sifive.com>
19198 Jim Wilson <jimw@sifive.com>
19200 * config/riscv/riscv.c (riscv_stack_boundary): New.
19201 (riscv_option_override): Set riscv_stack_boundary. Handle
19202 riscv_preferred_stack_boundary_arg.
19203 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
19204 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
19205 (STACK_BOUNDARY): Set to riscv_stack_boundary.
19206 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
19207 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
19208 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
19210 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
19213 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
19214 of struct ix86_frame.
19215 (ix86_expand_epilogue): Likewise. Add a local variable for
19216 the reg_save_offset field in struct ix86_frame.
19218 2018-01-23 Bin Cheng <bin.cheng@arm.com>
19220 PR tree-optimization/82604
19221 * tree-loop-distribution.c (enum partition_kind): New enum item
19222 PKIND_PARTIAL_MEMSET.
19223 (partition_builtin_p): Support above new enum item.
19224 (generate_code_for_partition): Ditto.
19225 (compute_access_range): Differentiate cases that equality can be
19226 proven at all loops, the innermost loops or no loops.
19227 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
19228 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
19229 (finalize_partitions, distribute_loop): Don't fuse partition of
19230 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
19231 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
19232 parloop is enabled.
19234 2018-01-23 Martin Liska <mliska@suse.cz>
19236 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
19237 order to ignore the predictor.
19238 (PRED_POLYMORPHIC_CALL): Likewise.
19239 (PRED_RECURSIVE_CALL): Likewise.
19241 2018-01-23 Martin Liska <mliska@suse.cz>
19243 * tree-profile.c (tree_profiling): Print function header to
19244 aware reader which function we are working on.
19245 * value-prof.c (gimple_find_values_to_profile): Do not print
19246 not interesting value histograms.
19248 2018-01-23 Martin Liska <mliska@suse.cz>
19250 * profile-count.h (enum profile_quality): Add
19251 profile_uninitialized as the first value. Do not number values
19252 as they are zero based.
19253 (profile_count::verify): Update sanity check.
19254 (profile_probability::verify): Likewise.
19256 2018-01-23 Nathan Sidwell <nathan@acm.org>
19258 * doc/invoke.texi (ffor-scope): Deprecate.
19260 2018-01-23 David Malcolm <dmalcolm@redhat.com>
19262 PR tree-optimization/83510
19263 * domwalk.c (set_all_edges_as_executable): New function.
19264 (dom_walker::dom_walker): Convert bool param
19265 "skip_unreachable_blocks" to enum reachability. Move setup of
19266 edge flags to set_all_edges_as_executable and only do it when
19267 reachability is REACHABLE_BLOCKS.
19268 * domwalk.h (enum dom_walker::reachability): New enum.
19269 (dom_walker::dom_walker): Convert bool param
19270 "skip_unreachable_blocks" to enum reachability.
19271 (set_all_edges_as_executable): New decl.
19272 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
19273 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
19275 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
19276 but converting true to REACHABLE_BLOCKS.
19277 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
19279 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
19280 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
19281 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
19283 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
19284 if check_all_array_refs will be called.
19286 2018-01-23 David Malcolm <dmalcolm@redhat.com>
19288 * tree.c (selftest::test_location_wrappers): Add more test
19291 2018-01-23 David Malcolm <dmalcolm@redhat.com>
19293 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
19294 (selftest::test_bit_in_range): Likewise.
19296 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
19299 * doc/sourcebuild.texi (vect_float): Say that the selector
19300 only describes the situation when -funsafe-math-optimizations is on.
19301 (vect_float_strict): Document.
19303 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
19305 PR tree-optimization/83965
19306 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
19307 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
19308 instead of checking only for a reduction.
19309 (vect_recog_widen_sum_pattern): Likewise.
19311 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
19313 * predict.c (probably_never_executed): Only use precise profile info.
19314 (compute_function_frequency): Skip after inlining hack since we now
19315 have quality checking.
19317 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
19319 * profile-count.h (profile_probability::very_unlikely,
19320 profile_probability::unlikely, profile_probability::even): Set
19321 precision to guessed.
19323 2018-01-23 Richard Biener <rguenther@suse.de>
19325 PR tree-optimization/83963
19326 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
19327 Properly terminate dominator walk when crossing the exit edge not
19328 when visiting its source block.
19330 2018-01-23 Jakub Jelinek <jakub@redhat.com>
19333 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
19334 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
19336 2018-01-22 Jakub Jelinek <jakub@redhat.com>
19338 PR tree-optimization/83957
19339 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
19340 semicolon after for body surrounded by braces.
19342 PR tree-optimization/83081
19343 * profile-count.h (profile_probability::split): New method.
19344 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
19345 Use profile_probability::split.
19346 (do_compare_rtx_and_jump): Fix adjustment of probabilities
19347 when splitting a single conditional jump into 2.
19349 2018-01-22 David Malcolm <dmalcolm@redhat.com>
19351 PR tree-optimization/69452
19352 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
19355 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
19357 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
19358 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
19359 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
19361 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
19363 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
19365 * config/rl78/rl78.md (movdi): New define_expand.
19366 * config/rl78/rl78.c (rl78_split_movdi): New function.
19368 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
19371 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
19373 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
19374 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
19375 128-bit to produce an UNSPEC move to get the double word with the
19376 signbit and then a shift directly to do signbit.
19377 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
19378 implementation with a new version that just does either a direct
19379 move or a regular move. Move memory interface to separate insns.
19380 Move insns so they are next to the expander.
19381 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
19382 with signbit move. Split big and little endian case.
19383 (signbit<mode>2_dm_mem_le): Likewise.
19384 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
19385 (signbit<mode>2_dm2): Likewise.
19387 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
19389 * config/rl78/rl78.md (anddi3): New define_expand.
19391 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
19393 * config/rl78/rl78.md (umindi3): New define_expand.
19395 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
19397 * config/rl78/rl78.md (smindi3): New define_expand.
19399 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
19401 * config/rl78/rl78.md (smaxdi3): New define_expand.
19403 2018-01-22 Carl Love <cel@us.ibm.com>
19405 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
19406 LVX_V1TI): Add macro expansion.
19407 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
19408 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
19409 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
19410 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
19411 Change check to determine if the instruction is a byte reversing
19412 entry. Fix typo in comment.
19413 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
19414 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
19415 Add def_builtin calls for new builtins.
19416 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
19417 Add define_insn expansion.
19419 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
19421 * config/rl78/rl78.md (umaxdi3): New define_expand.
19423 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
19425 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
19426 for non-QImode registers.
19428 2018-01-22 Richard Biener <rguenther@suse.de>
19430 PR tree-optimization/83963
19431 * graphite-scop-detection.c (scop_detection::get_sese): Delay
19432 including the loop exit block.
19433 (scop_detection::merge_sese): Likewise.
19434 (scop_detection::add_scop): Do it here instead.
19436 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19438 * doc/sourcebuild.texi (arm_softfloat): Document.
19440 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
19443 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
19444 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
19445 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
19447 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19448 David Edelsohn <dje.gcc@gmail.com>
19451 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
19452 Change "crset eq" to "crset 2".
19453 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
19454 (*call_indirect_aix<mode>_nospec): Likewise.
19455 (*call_value_indirect_aix<mode>_nospec): Likewise.
19456 (*call_indirect_elfv2<mode>_nospec): Likewise.
19457 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
19458 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
19459 change assembly output from . to $.
19460 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
19461 (indirect_jump<mode>_nospec): Change assembly output from . to $.
19462 (*tablejump<mode>_internal1_nospec): Likewise.
19464 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
19467 * config/sh/sh_optimize_sett_clrt.cc:
19468 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
19470 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
19472 PR tree-optimization/83940
19473 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
19474 offset_dt to vect_constant_def rather than vect_unknown_def_type.
19475 (vect_check_load_store_mask): Add a mask_dt_out parameter and
19476 use it to pass back the definition type.
19477 (vect_check_store_rhs): Likewise rhs_dt_out.
19478 (vect_build_gather_load_calls): Add a mask_dt argument and use
19479 it instead of a call to vect_is_simple_use.
19480 (vectorizable_store): Update calls to vect_check_load_store_mask
19481 and vect_check_store_rhs. Use the dt returned by the latter instead
19482 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
19483 instead of calls to vect_is_simple_use. Pass the scalar rather
19484 than the vector operand to vect_is_simple_use when handling
19485 second and subsequent copies of an rhs value.
19486 (vectorizable_load): Update calls to vect_check_load_store_mask
19487 and vect_build_gather_load_calls. Use the cached mask_dt and
19488 gs_info.offset_dt instead of calls to vect_is_simple_use.
19490 2018-01-20 Jakub Jelinek <jakub@redhat.com>
19492 PR middle-end/83945
19493 * tree-emutls.c: Include gimplify.h.
19494 (lower_emutls_2): New function.
19495 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
19496 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
19497 it before further processing.
19500 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
19501 UINTVAL (trueop1) instead of INTVAL (op1).
19503 2018-01-19 Jakub Jelinek <jakub@redhat.com>
19507 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
19508 INCOMING_FRAME_SP_OFFSET if not defined.
19509 (scan_trace): Add ENTRY argument. If true and
19510 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
19511 emit a note to adjust the CFA offset.
19512 (create_cfi_notes): Adjust scan_trace callers.
19513 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
19514 INCOMING_FRAME_SP_OFFSET in the CIE.
19515 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
19516 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
19518 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
19519 * doc/tm.texi: Regenerated.
19521 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19523 PR rtl-optimization/83147
19524 * lra-constraints.c (remove_inheritance_pseudos): Use
19525 lra_substitute_pseudo_within_insn.
19527 2018-01-19 Tom de Vries <tom@codesourcery.com>
19528 Cesar Philippidis <cesar@codesourcery.com>
19531 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
19533 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
19536 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
19537 spaces for function labels.
19539 2018-01-19 Martin Liska <mliska@suse.cz>
19541 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
19542 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
19543 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
19544 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
19545 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
19546 (PRED_CONST_RETURN): Change from 69 to 65.
19547 (PRED_NULL_RETURN): Change from 91 to 71.
19548 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
19549 (PRED_LOOP_GUARD): Change from 66 to 73.
19551 2018-01-19 Martin Liska <mliska@suse.cz>
19553 * predict.c (predict_insn_def): Add new assert.
19554 (struct branch_predictor): Change type to signed integer.
19555 (test_prediction_value_range): Amend test to cover
19556 PROB_UNINITIALIZED.
19557 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
19558 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
19559 (PRED_LOOP_ITERATIONS_MAX): Likewise.
19560 (PRED_LOOP_IV_COMPARE): Likewise.
19561 * predict.h (PROB_UNINITIALIZED): Define new constant.
19563 2018-01-19 Martin Liska <mliska@suse.cz>
19565 * predict.c (dump_prediction): Add new format for
19566 analyze_brprob.py script which is enabled with -details
19568 * profile-count.h (precise_p): New function.
19570 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
19572 PR tree-optimization/83922
19573 * tree-vect-loop.c (vect_verify_full_masking): Return false if
19574 there are no statements that need masking.
19575 (vect_active_double_reduction_p): New function.
19576 (vect_analyze_loop_operations): Use it when handling phis that
19577 are not in the loop header.
19579 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
19581 PR tree-optimization/83914
19582 * tree-vect-loop.c (vectorizable_induction): Don't convert
19583 init_expr or apply the peeling adjustment for inductions
19584 that are nested within the vectorized loop.
19586 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19588 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
19591 2018-01-18 Jakub Jelinek <jakub@redhat.com>
19595 * function.h (gimplify_parameters): Add gimple_seq * argument.
19596 * function.c: Include gimple.h and options.h.
19597 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
19598 for the added local temporaries if needed.
19599 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
19600 if there are any parameter cleanups, wrap whole body into a
19601 try/finally with the cleanups.
19603 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
19606 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
19607 Use GET_MODE_CLASS for scalar floating point.
19609 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
19613 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
19614 Fix call of call_cgraph_insertion_hooks.
19616 2018-01-18 Martin Sebor <msebor@redhat.com>
19618 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
19620 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
19623 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
19626 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
19629 * common.opt: (-ffile-prefix-map): New option.
19630 * opts.c (common_handle_option): Defer it.
19631 * opts-global.c (handle_common_deferred_options): Handle it.
19632 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
19633 * file-prefix-map.h: New file.
19634 (remap_debug_filename, add_debug_prefix_map): ...here.
19635 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
19636 * final.c (debug_prefix_map, add_debug_prefix_map
19637 remap_debug_filename): Move to...
19638 * file-prefix-map.c: New file.
19639 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
19640 generalize, get rid of alloca(), use strrchr() instead of strchr().
19641 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
19642 Implement in terms of add_prefix_map().
19643 (remap_macro_filename, remap_debug_filename): Implement in term of
19645 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
19646 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
19647 * dbxout.c: Include file-prefix-map.h.
19648 * varasm.c: Likewise.
19649 * vmsdbgout.c: Likewise.
19650 * xcoffout.c: Likewise.
19651 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
19652 * doc/cppopts.texi (-fmacro-prefix-map): Document.
19653 * doc/invoke.texi (-ffile-prefix-map): Document.
19654 (-fdebug-prefix-map): Update description.
19656 2018-01-18 Martin Liska <mliska@suse.cz>
19658 * config/i386/i386.c (indirect_thunk_name): Document that also
19660 (output_indirect_thunk): Document why both instructions
19661 (pause and lfence) are generated.
19663 2018-01-18 Richard Biener <rguenther@suse.de>
19665 PR tree-optimization/83887
19666 * graphite-scop-detection.c
19667 (scop_detection::get_nearest_dom_with_single_entry): Remove.
19668 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
19669 (scop_detection::merge_sese): Re-implement with a flood-fill
19670 algorithm that properly finds a SESE region if it exists.
19672 2018-01-18 Jakub Jelinek <jakub@redhat.com>
19675 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
19676 pointer_diff optimizations use view_convert instead of convert.
19678 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19680 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
19681 Generate different code for -mno-speculate-indirect-jumps.
19682 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
19683 (*call_indirect_aix<mode>): Disable for
19684 -mno-speculate-indirect-jumps.
19685 (*call_indirect_aix<mode>_nospec): New define_insn.
19686 (*call_value_indirect_aix<mode>): Disable for
19687 -mno-speculate-indirect-jumps.
19688 (*call_value_indirect_aix<mode>_nospec): New define_insn.
19689 (*sibcall_nonlocal_sysv<mode>): Generate different code for
19690 -mno-speculate-indirect-jumps.
19691 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
19693 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
19695 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
19696 long double type, set the flags for noting the default long double
19697 type, even if we don't pass or return a long double type.
19699 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
19702 * ipa-inline.c (flatten_function): Do not overwrite final inlining
19705 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
19707 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
19708 support for merge[hl].
19709 (fold_mergehl_helper): New helper function.
19710 (tree-vector-builder.h): New #include for tree_vector_builder usage.
19711 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
19712 (altivec_vmrglw_direct): Add xxmrglw insn.
19714 2018-01-17 Andrew Waterman <andrew@sifive.com>
19716 * config/riscv/riscv.c (riscv_conditional_register_usage): If
19717 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
19719 2018-01-17 David Malcolm <dmalcolm@redhat.com>
19722 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
19723 call the lto_location_cache before reading the
19724 DECL_SOURCE_LOCATION of the types.
19726 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
19727 Richard Sandiford <richard.sandiford@linaro.org>
19729 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
19730 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
19731 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
19732 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
19733 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
19735 * config/aarch64/constraints.md (aarch64_movti_operand):
19737 * config/aarch64/predicates.md (Uti): Add new constraint.
19739 2018-01-17 Carl Love <cel@us.ibm.com>
19741 * config/rs6000/vsx.md (define_expand xl_len_r,
19742 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
19743 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
19745 (define_expand, define_insn): Move the shift left from the
19746 define_insn to the define_expand for lxvl and stxvl instructions.
19747 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
19748 and XL_LEN_R definitions to PURE.
19750 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
19752 * config/i386/i386.c (indirect_thunk_name): Declare regno
19753 as unsigned int. Compare regno with INVALID_REGNUM.
19754 (output_indirect_thunk): Ditto.
19755 (output_indirect_thunk_function): Ditto.
19756 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
19757 in the call to output_indirect_thunk_function.
19759 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
19761 PR middle-end/83884
19762 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
19763 rather than the size of inner_type to determine the stack slot size
19764 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
19766 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
19769 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
19772 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
19774 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
19775 endian Linux systems to optionally enable multilibs for selecting
19776 the long double type if the user configured an explicit type.
19777 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
19778 have no long double multilibs if not defined.
19779 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
19780 warn if the user used -mabi={ieee,ibm}longdouble and we built
19781 multilibs for long double.
19782 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
19783 appropriate multilib option.
19784 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
19786 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
19787 for building long double multilibs.
19788 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
19790 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
19792 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
19795 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
19797 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
19800 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
19803 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
19806 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
19808 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
19809 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
19811 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
19813 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
19814 different rtl trees depending on TARGET_64BIT.
19815 (rs6000_gen_lvx): Likewise.
19817 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
19819 * config/visium/visium.md (nop): Tweak comment.
19820 (hazard_nop): Likewise.
19822 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19824 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
19825 -mspeculate-indirect-jumps.
19826 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
19827 for -mno-speculate-indirect-jumps.
19828 (*call_indirect_elfv2<mode>_nospec): New define_insn.
19829 (*call_value_indirect_elfv2<mode>): Disable for
19830 -mno-speculate-indirect-jumps.
19831 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
19832 (indirect_jump): Emit different RTL for
19833 -mno-speculate-indirect-jumps.
19834 (*indirect_jump<mode>): Disable for
19835 -mno-speculate-indirect-jumps.
19836 (*indirect_jump<mode>_nospec): New define_insn.
19837 (tablejump): Emit different RTL for
19838 -mno-speculate-indirect-jumps.
19839 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
19840 (tablejumpsi_nospec): New define_expand.
19841 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
19842 (tablejumpdi_nospec): New define_expand.
19843 (*tablejump<mode>_internal1): Disable for
19844 -mno-speculate-indirect-jumps.
19845 (*tablejump<mode>_internal1_nospec): New define_insn.
19846 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
19849 2018-01-16 Artyom Skrobov tyomitch@gmail.com
19851 * caller-save.c (insert_save): Drop unnecessary parameter. All
19854 2018-01-16 Jakub Jelinek <jakub@redhat.com>
19855 Richard Biener <rguenth@suse.de>
19858 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
19859 return early, inline manually is_gimple_sizepos. Make sure if we
19860 call gimplify_expr we don't end up with a gimple constant.
19861 * tree.c (variably_modified_type_p): Don't return true for
19862 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
19863 * gimplify.h (is_gimple_sizepos): Remove.
19865 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
19867 PR tree-optimization/83857
19868 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
19869 vectorizable_live_operation for pure SLP statements.
19870 (vectorizable_live_operation): Handle PHIs.
19872 2018-01-16 Richard Biener <rguenther@suse.de>
19874 PR tree-optimization/83867
19875 * tree-vect-stmts.c (vect_transform_stmt): Precompute
19876 nested_in_vect_loop_p since the scalar stmt may get invalidated.
19878 2018-01-16 Jakub Jelinek <jakub@redhat.com>
19881 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
19882 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
19883 If off is not INTEGER_CST, issue a may not be aligned warning
19884 rather than isn't aligned. Use isn%'t rather than isn't.
19885 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
19887 <case MULT_EXPR>: Improve the case when bottom and one of the
19888 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
19889 operand, in that case check if the other operand is multiple of
19890 bottom divided by the INTEGER_CST operand.
19892 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
19895 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
19896 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
19897 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
19898 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
19899 * config/pa/pa.c (pa_function_arg_advance): Likewise.
19900 (pa_function_arg, pa_arg_partial_bytes): Likewise.
19901 (pa_function_arg_size): New function.
19903 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
19905 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
19906 in a separate statement.
19908 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
19910 PR tree-optimization/83847
19911 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
19912 group gathers and scatters.
19914 2018-01-16 Jakub Jelinek <jakub@redhat.com>
19916 PR rtl-optimization/86620
19917 * params.def (max-sched-ready-insns): Bump minimum value to 1.
19919 PR rtl-optimization/83213
19920 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
19921 to last if both are JUMP_INSNs.
19923 PR tree-optimization/83843
19924 * gimple-ssa-store-merging.c
19925 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
19926 store_immediate_info for bswap/nop orig_stores.
19928 2018-01-15 Andrew Waterman <andrew@sifive.com>
19930 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
19932 <UDIV>: Increase cost if !TARGET_DIV.
19934 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
19936 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
19937 (define_attr "cr_logical_3op"): New.
19938 (cceq_ior_compare): Adjust.
19939 (cceq_ior_compare_complement): Adjust.
19940 (*cceq_rev_compare): Adjust.
19941 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
19942 (is_cracked_insn): Adjust.
19943 (insn_must_be_first_in_group): Adjust.
19944 * config/rs6000/40x.md: Adjust.
19945 * config/rs6000/440.md: Adjust.
19946 * config/rs6000/476.md: Adjust.
19947 * config/rs6000/601.md: Adjust.
19948 * config/rs6000/603.md: Adjust.
19949 * config/rs6000/6xx.md: Adjust.
19950 * config/rs6000/7450.md: Adjust.
19951 * config/rs6000/7xx.md: Adjust.
19952 * config/rs6000/8540.md: Adjust.
19953 * config/rs6000/cell.md: Adjust.
19954 * config/rs6000/e300c2c3.md: Adjust.
19955 * config/rs6000/e500mc.md: Adjust.
19956 * config/rs6000/e500mc64.md: Adjust.
19957 * config/rs6000/e5500.md: Adjust.
19958 * config/rs6000/e6500.md: Adjust.
19959 * config/rs6000/mpc.md: Adjust.
19960 * config/rs6000/power4.md: Adjust.
19961 * config/rs6000/power5.md: Adjust.
19962 * config/rs6000/power6.md: Adjust.
19963 * config/rs6000/power7.md: Adjust.
19964 * config/rs6000/power8.md: Adjust.
19965 * config/rs6000/power9.md: Adjust.
19966 * config/rs6000/rs64.md: Adjust.
19967 * config/rs6000/titan.md: Adjust.
19969 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
19971 * config/i386/predicates.md (indirect_branch_operand): Rewrite
19972 ix86_indirect_branch_register logic.
19974 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
19976 * config/i386/constraints.md (Bs): Update
19977 ix86_indirect_branch_register check. Don't check
19978 ix86_indirect_branch_register with GOT_memory_operand.
19980 * config/i386/predicates.md (GOT_memory_operand): Don't check
19981 ix86_indirect_branch_register here.
19982 (GOT32_symbol_operand): Likewise.
19984 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
19986 * config/i386/predicates.md (constant_call_address_operand):
19987 Rewrite ix86_indirect_branch_register logic.
19988 (sibcall_insn_operand): Likewise.
19990 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
19992 * config/i386/constraints.md (Bs): Replace
19993 ix86_indirect_branch_thunk_register with
19994 ix86_indirect_branch_register.
19996 * config/i386/i386.md (indirect_jump): Likewise.
19997 (tablejump): Likewise.
19998 (*sibcall_memory): Likewise.
19999 (*sibcall_value_memory): Likewise.
20000 Peepholes of indirect call and jump via memory: Likewise.
20001 * config/i386/i386.opt: Likewise.
20002 * config/i386/predicates.md (indirect_branch_operand): Likewise.
20003 (GOT_memory_operand): Likewise.
20004 (call_insn_operand): Likewise.
20005 (sibcall_insn_operand): Likewise.
20006 (GOT32_symbol_operand): Likewise.
20008 2018-01-15 Jakub Jelinek <jakub@redhat.com>
20010 PR middle-end/83837
20011 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
20012 type rather than type addr's type points to.
20013 (expand_omp_atomic_mutex): Likewise.
20014 (expand_omp_atomic): Likewise.
20016 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
20019 * config/i386/i386.c (output_indirect_thunk_function): Use
20020 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
20021 for __x86_return_thunk.
20023 2018-01-15 Richard Biener <rguenther@suse.de>
20025 PR middle-end/83850
20026 * expmed.c (extract_bit_field_1): Fix typo.
20028 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20031 * config/arm/iterators.md (VF): New mode iterator.
20032 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
20033 Remove integer-related logic from pattern.
20034 (neon_vabd<mode>_3): Likewise.
20036 2018-01-15 Jakub Jelinek <jakub@redhat.com>
20038 PR middle-end/82694
20039 * common.opt (fstrict-overflow): No longer an alias.
20040 (fwrapv-pointer): New option.
20041 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
20042 also for pointer types based on flag_wrapv_pointer.
20043 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
20044 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
20045 opts->x_flag_wrapv got set.
20046 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
20047 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
20048 POINTER_TYPE_OVERFLOW_UNDEFINED.
20049 * match.pd: Likewise in address comparison pattern.
20050 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
20052 2018-01-15 Richard Biener <rguenther@suse.de>
20055 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
20056 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
20057 Reset type names to their identifier if their TYPE_DECL doesn't
20058 have linkage (and thus is used for ODR and devirt).
20059 (save_debug_info_for_decl): Remove.
20060 (save_debug_info_for_type): Likewise.
20061 (add_tree_to_fld_list): Adjust.
20062 * tree-pretty-print.c (dump_generic_node): Make dumping of
20063 type names more robust.
20065 2018-01-15 Richard Biener <rguenther@suse.de>
20067 * BASE-VER: Bump to 8.0.1.
20069 2018-01-14 Martin Sebor <msebor@redhat.com>
20072 * builtins.c (check_access): Avoid warning when the no-warning bit
20075 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
20077 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
20078 * ira-color (allocno_hard_regs_compare): Likewise.
20080 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
20083 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
20084 Use .pushsection/.popsection.
20086 2018-01-14 Martin Sebor <msebor@redhat.com>
20089 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
20091 2018-01-14 Jakub Jelinek <jakub@redhat.com>
20093 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
20094 entry from extra_headers.
20095 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
20096 extra_headers, make the list bitwise identical to the i?86-*-* one.
20098 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
20100 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
20101 -mcmodel=large with -mindirect-branch=thunk,
20102 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
20103 -mfunction-return=thunk-extern.
20104 * doc/invoke.texi: Document -mcmodel=large is incompatible with
20105 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
20106 -mfunction-return=thunk and -mfunction-return=thunk-extern.
20108 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
20110 * config/i386/i386.c (print_reg): Print the name of the full
20111 integer register without '%'.
20112 (ix86_print_operand): Handle 'V'.
20113 * doc/extend.texi: Document 'V' modifier.
20115 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
20117 * config/i386/constraints.md (Bs): Disallow memory operand for
20118 -mindirect-branch-register.
20120 * config/i386/predicates.md (indirect_branch_operand): Likewise.
20121 (GOT_memory_operand): Likewise.
20122 (call_insn_operand): Likewise.
20123 (sibcall_insn_operand): Likewise.
20124 (GOT32_symbol_operand): Likewise.
20125 * config/i386/i386.md (indirect_jump): Call convert_memory_address
20126 for -mindirect-branch-register.
20127 (tablejump): Likewise.
20128 (*sibcall_memory): Likewise.
20129 (*sibcall_value_memory): Likewise.
20130 Disallow peepholes of indirect call and jump via memory for
20131 -mindirect-branch-register.
20132 (*call_pop): Replace m with Bw.
20133 (*call_value_pop): Likewise.
20134 (*sibcall_pop_memory): Replace m with Bs.
20135 * config/i386/i386.opt (mindirect-branch-register): New option.
20136 * doc/invoke.texi: Document -mindirect-branch-register option.
20138 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
20140 * config/i386/i386-protos.h (ix86_output_function_return): New.
20141 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
20142 set function_return_type.
20143 (indirect_thunk_name): Add ret_p to indicate thunk for function
20145 (output_indirect_thunk_function): Pass false to
20146 indirect_thunk_name.
20147 (ix86_output_indirect_branch_via_reg): Likewise.
20148 (ix86_output_indirect_branch_via_push): Likewise.
20149 (output_indirect_thunk_function): Create alias for function
20150 return thunk if regno < 0.
20151 (ix86_output_function_return): New function.
20152 (ix86_handle_fndecl_attribute): Handle function_return.
20153 (ix86_attribute_table): Add function_return.
20154 * config/i386/i386.h (machine_function): Add
20155 function_return_type.
20156 * config/i386/i386.md (simple_return_internal): Use
20157 ix86_output_function_return.
20158 (simple_return_internal_long): Likewise.
20159 * config/i386/i386.opt (mfunction-return=): New option.
20160 (indirect_branch): Mention -mfunction-return=.
20161 * doc/extend.texi: Document function_return function attribute.
20162 * doc/invoke.texi: Document -mfunction-return= option.
20164 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
20166 * config/i386/i386-opts.h (indirect_branch): New.
20167 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
20168 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
20169 with local indirect jump when converting indirect call and jump.
20170 (ix86_set_indirect_branch_type): New.
20171 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
20172 (indirectlabelno): New.
20173 (indirect_thunk_needed): Likewise.
20174 (indirect_thunk_bnd_needed): Likewise.
20175 (indirect_thunks_used): Likewise.
20176 (indirect_thunks_bnd_used): Likewise.
20177 (INDIRECT_LABEL): Likewise.
20178 (indirect_thunk_name): Likewise.
20179 (output_indirect_thunk): Likewise.
20180 (output_indirect_thunk_function): Likewise.
20181 (ix86_output_indirect_branch_via_reg): Likewise.
20182 (ix86_output_indirect_branch_via_push): Likewise.
20183 (ix86_output_indirect_branch): Likewise.
20184 (ix86_output_indirect_jmp): Likewise.
20185 (ix86_code_end): Call output_indirect_thunk_function if needed.
20186 (ix86_output_call_insn): Call ix86_output_indirect_branch if
20188 (ix86_handle_fndecl_attribute): Handle indirect_branch.
20189 (ix86_attribute_table): Add indirect_branch.
20190 * config/i386/i386.h (machine_function): Add indirect_branch_type
20191 and has_local_indirect_jump.
20192 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
20194 (tablejump): Likewise.
20195 (*indirect_jump): Use ix86_output_indirect_jmp.
20196 (*tablejump_1): Likewise.
20197 (simple_return_indirect_internal): Likewise.
20198 * config/i386/i386.opt (mindirect-branch=): New option.
20199 (indirect_branch): New.
20202 (thunk-inline): Likewise.
20203 (thunk-extern): Likewise.
20204 * doc/extend.texi: Document indirect_branch function attribute.
20205 * doc/invoke.texi: Document -mindirect-branch= option.
20207 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
20210 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
20212 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
20214 * ipa-inline.c (want_inline_small_function_p): Return false if
20215 inlining has already failed with CIF_FINAL_ERROR.
20216 (update_caller_keys): Call want_inline_small_function_p before
20218 (update_callee_keys): Likewise.
20220 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
20222 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
20224 (rs6000_quadword_masked_address_p): Likewise.
20225 (quad_aligned_load_p): Likewise.
20226 (quad_aligned_store_p): Likewise.
20227 (const_load_sequence_p): Add comment to describe the outer-most loop.
20228 (mimic_memory_attributes_and_flags): New function.
20229 (rs6000_gen_stvx): Likewise.
20230 (replace_swapped_aligned_store): Likewise.
20231 (rs6000_gen_lvx): Likewise.
20232 (replace_swapped_aligned_load): Likewise.
20233 (replace_swapped_load_constant): Capitalize argument name in
20234 comment describing this function.
20235 (rs6000_analyze_swaps): Add a third pass to search for vector loads
20236 and stores that access quad-word aligned addresses and replace
20237 with stvx or lvx instructions when appropriate.
20238 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
20239 New function prototype.
20240 (rs6000_quadword_masked_address_p): Likewise.
20241 (rs6000_gen_lvx): Likewise.
20242 (rs6000_gen_stvx): Likewise.
20243 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
20244 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
20245 when memory address is aligned.
20246 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
20247 this split to select lvx instruction when memory address is aligned.
20248 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
20249 instruction when memory address is aligned.
20250 (*vsx_le_perm_load_v16qi): Likewise.
20251 (four unnamed splitters): Modify to select the stvx instruction
20252 when memory is aligned.
20254 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
20256 * predict.c (determine_unlikely_bbs): Handle correctly BBs
20257 which appears in the queue multiple times.
20259 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20260 Alan Hayward <alan.hayward@arm.com>
20261 David Sherwood <david.sherwood@arm.com>
20263 * tree-vectorizer.h (vec_lower_bound): New structure.
20264 (_loop_vec_info): Add check_nonzero and lower_bounds.
20265 (LOOP_VINFO_CHECK_NONZERO): New macro.
20266 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
20267 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
20268 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
20269 fields. Make seg_len the distance travelled, not including the
20271 (dr_direction_indicator): Declare.
20272 (dr_zero_step_indicator): Likewise.
20273 (dr_known_forward_stride_p): Likewise.
20274 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
20276 (runtime_alias_check_p): Allow runtime alias checks with
20278 (operator ==): Compare access_size and align.
20279 (prune_runtime_alias_test_list): Rework for new distinction between
20280 the access_size and seg_len.
20281 (create_intersect_range_checks_index): Likewise. Cope with polynomial
20283 (get_segment_min_max): New function.
20284 (create_intersect_range_checks): Use it.
20285 (dr_step_indicator): New function.
20286 (dr_direction_indicator): Likewise.
20287 (dr_zero_step_indicator): Likewise.
20288 (dr_known_forward_stride_p): Likewise.
20289 * tree-loop-distribution.c (data_ref_segment_size): Return
20290 DR_STEP * (niters - 1).
20291 (compute_alias_check_pairs): Update call to the dr_with_seg_len
20293 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
20294 (vect_preserves_scalar_order_p): New function, split out from...
20295 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
20296 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
20297 (vect_vfa_access_size): New function.
20298 (vect_vfa_align): Likewise.
20299 (vect_compile_time_alias): Take access_size_a and access_b arguments.
20300 (dump_lower_bound): New function.
20301 (vect_check_lower_bound): Likewise.
20302 (vect_small_gap_p): Likewise.
20303 (vectorizable_with_step_bound_p): Likewise.
20304 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
20305 depencies if the vectorization factor is 1. Convert the checks
20306 for nonzero steps into checks on the bounds of DR_STEP. Try using
20307 a bunds check for variable steps if the minimum required step is
20308 relatively small. Update calls to the dr_with_seg_len
20309 constructor and to vect_compile_time_alias.
20310 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
20312 (vect_loop_versioning): Call it.
20313 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
20315 (vect_estimate_min_profitable_iters): Account for any bounds checks.
20317 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20318 Alan Hayward <alan.hayward@arm.com>
20319 David Sherwood <david.sherwood@arm.com>
20321 * doc/sourcebuild.texi (vect_scatter_store): Document.
20322 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
20324 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
20326 * genopinit.c (main): Add supports_vec_scatter_store and
20327 supports_vec_scatter_store_cached to target_optabs.
20328 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
20329 IFN_MASK_SCATTER_STORE.
20330 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
20332 * internal-fn.h (internal_store_fn_p): Declare.
20333 (internal_fn_stored_value_index): Likewise.
20334 * internal-fn.c (scatter_store_direct): New macro.
20335 (expand_scatter_store_optab_fn): New function.
20336 (direct_scatter_store_optab_supported_p): New macro.
20337 (internal_store_fn_p): New function.
20338 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
20339 IFN_MASK_SCATTER_STORE.
20340 (internal_fn_mask_index): Likewise.
20341 (internal_fn_stored_value_index): New function.
20342 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
20343 for scatter stores.
20344 * optabs-query.h (supports_vec_scatter_store_p): Declare.
20345 * optabs-query.c (supports_vec_scatter_store_p): New function.
20346 * tree-vectorizer.h (vect_get_store_rhs): Declare.
20347 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
20348 true for scatter stores.
20349 (vect_gather_scatter_fn_p): Handle scatter stores too.
20350 (vect_check_gather_scatter): Consider using scatter stores if
20351 supports_vec_scatter_store_p.
20352 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
20353 scatter stores too.
20354 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
20355 internal_fn_stored_value_index.
20356 (check_load_store_masking): Handle scatter stores too.
20357 (vect_get_store_rhs): Make public.
20358 (vectorizable_call): Use internal_store_fn_p.
20359 (vectorizable_store): Handle scatter store internal functions.
20360 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
20361 when deciding whether the end of the group has been reached.
20362 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
20363 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
20364 (mask_scatter_store<mode>): New insns.
20366 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20367 Alan Hayward <alan.hayward@arm.com>
20368 David Sherwood <david.sherwood@arm.com>
20370 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
20371 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
20372 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
20374 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
20375 Use vect_truncate_gather_scatter_offset if we can't treat the
20376 operation as a normal gather load or scatter store.
20377 (get_group_load_store_type): Take the gather_scatter_info
20378 as argument. Try using a gather load or scatter store for
20379 single-element groups.
20380 (get_load_store_type): Update calls to get_group_load_store_type
20381 and vect_use_strided_gather_scatters_p.
20383 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20384 Alan Hayward <alan.hayward@arm.com>
20385 David Sherwood <david.sherwood@arm.com>
20387 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
20388 optional tree argument.
20389 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
20391 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
20392 but continue to use the current value as a fallback.
20393 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
20394 to compare the updates.
20395 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
20396 (get_load_store_type): Use it when handling a strided access.
20397 (vect_get_strided_load_store_ops): New function.
20398 (vect_get_data_ptr_increment): Likewise.
20399 (vectorizable_load): Handle strided gather loads. Always pass
20400 a step to vect_create_data_ref_ptr and bump_vector_ptr.
20402 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20403 Alan Hayward <alan.hayward@arm.com>
20404 David Sherwood <david.sherwood@arm.com>
20406 * doc/md.texi (gather_load@var{m}): Document.
20407 (mask_gather_load@var{m}): Likewise.
20408 * genopinit.c (main): Add supports_vec_gather_load and
20409 supports_vec_gather_load_cached to target_optabs.
20410 * optabs-tree.c (init_tree_optimization_optabs): Use
20411 ggc_cleared_alloc to allocate target_optabs.
20412 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
20413 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
20415 * internal-fn.h (internal_load_fn_p): Declare.
20416 (internal_gather_scatter_fn_p): Likewise.
20417 (internal_fn_mask_index): Likewise.
20418 (internal_gather_scatter_fn_supported_p): Likewise.
20419 * internal-fn.c (gather_load_direct): New macro.
20420 (expand_gather_load_optab_fn): New function.
20421 (direct_gather_load_optab_supported_p): New macro.
20422 (direct_internal_fn_optab): New function.
20423 (internal_load_fn_p): Likewise.
20424 (internal_gather_scatter_fn_p): Likewise.
20425 (internal_fn_mask_index): Likewise.
20426 (internal_gather_scatter_fn_supported_p): Likewise.
20427 * optabs-query.c (supports_at_least_one_mode_p): New function.
20428 (supports_vec_gather_load_p): Likewise.
20429 * optabs-query.h (supports_vec_gather_load_p): Declare.
20430 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
20431 and memory_type field.
20432 (NUM_PATTERNS): Bump to 15.
20433 * tree-vect-data-refs.c: Include internal-fn.h.
20434 (vect_gather_scatter_fn_p): New function.
20435 (vect_describe_gather_scatter_call): Likewise.
20436 (vect_check_gather_scatter): Try using internal functions for
20437 gather loads. Recognize existing calls to a gather load function.
20438 (vect_analyze_data_refs): Consider using gather loads if
20439 supports_vec_gather_load_p.
20440 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
20441 (vect_get_gather_scatter_offset_type): Likewise.
20442 (vect_convert_mask_for_vectype): Likewise.
20443 (vect_add_conversion_to_patterm): Likewise.
20444 (vect_try_gather_scatter_pattern): Likewise.
20445 (vect_recog_gather_scatter_pattern): New pattern recognizer.
20446 (vect_vect_recog_func_ptrs): Add it.
20447 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
20448 internal_fn_mask_index and internal_gather_scatter_fn_p.
20449 (check_load_store_masking): Take the gather_scatter_info as an
20450 argument and handle gather loads.
20451 (vect_get_gather_scatter_ops): New function.
20452 (vectorizable_call): Check internal_load_fn_p.
20453 (vectorizable_load): Likewise. Handle gather load internal
20455 (vectorizable_store): Update call to check_load_store_masking.
20456 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
20457 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
20458 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
20459 (aarch64_gather_scale_operand_d): New predicates.
20460 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
20461 (mask_gather_load<mode>): New insns.
20463 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20464 Alan Hayward <alan.hayward@arm.com>
20465 David Sherwood <david.sherwood@arm.com>
20467 * optabs.def (fold_left_plus_optab): New optab.
20468 * doc/md.texi (fold_left_plus_@var{m}): Document.
20469 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
20470 * internal-fn.c (fold_left_direct): Define.
20471 (expand_fold_left_optab_fn): Likewise.
20472 (direct_fold_left_optab_supported_p): Likewise.
20473 * fold-const-call.c (fold_const_fold_left): New function.
20474 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
20475 * tree-parloops.c (valid_reduction_p): New function.
20476 (gather_scalar_reductions): Use it.
20477 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
20478 (vect_finish_replace_stmt): Declare.
20479 * tree-vect-loop.c (fold_left_reduction_fn): New function.
20480 (needs_fold_left_reduction_p): New function, split out from...
20481 (vect_is_simple_reduction): ...here. Accept reductions that
20482 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
20483 (vect_force_simple_reduction): Also store the reduction type in
20484 the assignment's STMT_VINFO_REDUC_TYPE.
20485 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
20486 (merge_with_identity): New function.
20487 (vect_expand_fold_left): Likewise.
20488 (vectorize_fold_left_reduction): Likewise.
20489 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
20490 scalar phi in place for it. Check for target support and reject
20491 cases that would reassociate the operation. Defer the transform
20492 phase to vectorize_fold_left_reduction.
20493 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
20494 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
20495 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
20497 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20499 * tree-if-conv.c (predicate_mem_writes): Remove redundant
20500 call to ifc_temp_var.
20502 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20503 Alan Hayward <alan.hayward@arm.com>
20504 David Sherwood <david.sherwood@arm.com>
20506 * target.def (legitimize_address_displacement): Take the original
20507 offset as a poly_int.
20508 * targhooks.h (default_legitimize_address_displacement): Update
20510 * targhooks.c (default_legitimize_address_displacement): Likewise.
20511 * doc/tm.texi: Regenerate.
20512 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
20513 as an argument, moving assert of ad->disp == ad->disp_term to...
20514 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
20515 Try calling targetm.legitimize_address_displacement before expanding
20516 the address rather than afterwards, and adjust for the new interface.
20517 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
20518 Match the new hook interface. Handle SVE addresses.
20519 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
20520 new hook interface.
20522 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20524 * Makefile.in (OBJS): Add early-remat.o.
20525 * target.def (select_early_remat_modes): New hook.
20526 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
20527 * doc/tm.texi: Regenerate.
20528 * targhooks.h (default_select_early_remat_modes): Declare.
20529 * targhooks.c (default_select_early_remat_modes): New function.
20530 * timevar.def (TV_EARLY_REMAT): New timevar.
20531 * passes.def (pass_early_remat): New pass.
20532 * tree-pass.h (make_pass_early_remat): Declare.
20533 * early-remat.c: New file.
20534 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
20536 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
20538 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20539 Alan Hayward <alan.hayward@arm.com>
20540 David Sherwood <david.sherwood@arm.com>
20542 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
20543 vfm1 with a bound_epilog parameter.
20544 (vect_do_peeling): Update calls accordingly, and move the prologue
20545 call earlier in the function. Treat the base bound_epilog as 0 for
20546 fully-masked loops and retain vf - 1 for other loops. Add 1 to
20547 this base when peeling for gaps.
20548 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
20549 with fully-masked loops.
20550 (vect_estimate_min_profitable_iters): Handle the single peeled
20551 iteration in that case.
20553 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20554 Alan Hayward <alan.hayward@arm.com>
20555 David Sherwood <david.sherwood@arm.com>
20557 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
20558 single-element interleaving even if the size is not a power of 2.
20559 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
20560 accesses for single-element interleaving if the group size is
20563 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20564 Alan Hayward <alan.hayward@arm.com>
20565 David Sherwood <david.sherwood@arm.com>
20567 * doc/md.texi (fold_extract_last_@var{m}): Document.
20568 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
20569 * optabs.def (fold_extract_last_optab): New optab.
20570 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
20571 * internal-fn.c (fold_extract_direct): New macro.
20572 (expand_fold_extract_optab_fn): Likewise.
20573 (direct_fold_extract_optab_supported_p): Likewise.
20574 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
20575 * tree-vect-loop.c (vect_model_reduction_cost): Handle
20576 EXTRACT_LAST_REDUCTION.
20577 (get_initial_def_for_reduction): Do not create an initial vector
20578 for EXTRACT_LAST_REDUCTION reductions.
20579 (vectorizable_reduction): Leave the scalar phi in place for
20580 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
20581 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
20582 epilogue code for EXTRACT_LAST_REDUCTION and defer the
20583 transform phase to vectorizable_condition.
20584 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
20586 (vect_finish_stmt_generation): ...here.
20587 (vect_finish_replace_stmt): New function.
20588 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
20589 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
20591 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
20593 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20594 Alan Hayward <alan.hayward@arm.com>
20595 David Sherwood <david.sherwood@arm.com>
20597 * doc/md.texi (extract_last_@var{m}): Document.
20598 * optabs.def (extract_last_optab): New optab.
20599 * internal-fn.def (EXTRACT_LAST): New internal function.
20600 * internal-fn.c (cond_unary_direct): New macro.
20601 (expand_cond_unary_optab_fn): Likewise.
20602 (direct_cond_unary_optab_supported_p): Likewise.
20603 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
20604 loops using EXTRACT_LAST.
20605 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
20606 (extract_last_<mode>): ...this optab.
20607 (vec_extract<mode><Vel>): Update accordingly.
20609 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20610 Alan Hayward <alan.hayward@arm.com>
20611 David Sherwood <david.sherwood@arm.com>
20613 * target.def (empty_mask_is_expensive): New hook.
20614 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
20615 * doc/tm.texi: Regenerate.
20616 * targhooks.h (default_empty_mask_is_expensive): Declare.
20617 * targhooks.c (default_empty_mask_is_expensive): New function.
20618 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
20619 if the target says that empty masks are expensive.
20620 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
20622 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
20624 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20625 Alan Hayward <alan.hayward@arm.com>
20626 David Sherwood <david.sherwood@arm.com>
20628 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
20629 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
20630 (vect_use_loop_mask_for_alignment_p): New function.
20631 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
20632 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
20633 niters_skip argument. Make sure that the first niters_skip elements
20634 of the first iteration are inactive.
20635 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
20636 Update call to vect_set_loop_masks_directly.
20637 (get_misalign_in_elems): New function, split out from...
20638 (vect_gen_prolog_loop_niters): ...here.
20639 (vect_update_init_of_dr): Take a code argument that specifies whether
20640 the adjustment should be added or subtracted.
20641 (vect_update_init_of_drs): Likewise.
20642 (vect_prepare_for_masked_peels): New function.
20643 (vect_do_peeling): Skip prologue peeling if we're using a mask
20644 instead. Update call to vect_update_inits_of_drs.
20645 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
20647 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
20648 alignment. Do not include the number of peeled iterations in
20649 the minimum threshold in that case.
20650 (vectorizable_induction): Adjust the start value down by
20651 LOOP_VINFO_MASK_SKIP_NITERS iterations.
20652 (vect_transform_loop): Call vect_prepare_for_masked_peels.
20653 Take the number of skipped iterations into account when calculating
20655 * tree-vect-stmts.c (vect_gen_while_not): New function.
20657 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20658 Alan Hayward <alan.hayward@arm.com>
20659 David Sherwood <david.sherwood@arm.com>
20661 * doc/sourcebuild.texi (vect_fully_masked): Document.
20662 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
20663 default value to 0.
20664 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
20666 (vect_analyze_loop_2): ...here. Don't check the vectorization
20667 factor against the number of loop iterations if the loop is
20670 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20671 Alan Hayward <alan.hayward@arm.com>
20672 David Sherwood <david.sherwood@arm.com>
20674 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
20675 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
20676 (dump_groups): Update accordingly.
20677 (iv_use::mem_type): New member variable.
20678 (address_p): New function.
20679 (record_use): Add a mem_type argument and initialize the new
20681 (record_group_use): Add a mem_type argument. Use address_p.
20682 Remove obsolete null checks of base_object. Update call to record_use.
20683 (find_interesting_uses_op): Update call to record_group_use.
20684 (find_interesting_uses_cond): Likewise.
20685 (find_interesting_uses_address): Likewise.
20686 (get_mem_type_for_internal_fn): New function.
20687 (find_address_like_use): Likewise.
20688 (find_interesting_uses_stmt): Try find_address_like_use before
20689 calling find_interesting_uses_op.
20690 (addr_offset_valid_p): Use the iv mem_type field as the type
20691 of the addressed memory.
20692 (add_autoinc_candidates): Likewise.
20693 (get_address_cost): Likewise.
20694 (split_small_address_groups_p): Use address_p.
20695 (split_address_groups): Likewise.
20696 (add_iv_candidate_for_use): Likewise.
20697 (autoinc_possible_for_pair): Likewise.
20698 (rewrite_groups): Likewise.
20699 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
20700 (determine_group_iv_cost): Update after split of USE_ADDRESS.
20701 (get_alias_ptr_type_for_ptr_address): New function.
20702 (rewrite_use_address): Rewrite address uses in calls that were
20703 identified by find_address_like_use.
20705 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20706 Alan Hayward <alan.hayward@arm.com>
20707 David Sherwood <david.sherwood@arm.com>
20709 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
20711 * gimple-expr.h (is_gimple_addressable: Likewise.
20712 * gimple-expr.c (is_gimple_address): Likewise.
20713 * internal-fn.c (expand_call_mem_ref): New function.
20714 (expand_mask_load_optab_fn): Use it.
20715 (expand_mask_store_optab_fn): Likewise.
20717 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20718 Alan Hayward <alan.hayward@arm.com>
20719 David Sherwood <david.sherwood@arm.com>
20721 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
20722 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
20723 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
20724 (cond_umax@var{mode}): Document.
20725 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
20726 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
20727 (cond_umin_optab, cond_umax_optab): New optabs.
20728 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
20729 (COND_IOR, COND_XOR): New internal functions.
20730 * internal-fn.h (get_conditional_internal_fn): Declare.
20731 * internal-fn.c (cond_binary_direct): New macro.
20732 (expand_cond_binary_optab_fn): Likewise.
20733 (direct_cond_binary_optab_supported_p): Likewise.
20734 (get_conditional_internal_fn): New function.
20735 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
20736 Cope with reduction statements that are vectorized as calls rather
20738 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
20739 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
20740 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
20741 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
20742 (UNSPEC_COND_EOR): New unspecs.
20743 (optab): Add mappings for them.
20744 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
20745 (sve_int_op, sve_fp_op): New int attributes.
20747 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20748 Alan Hayward <alan.hayward@arm.com>
20749 David Sherwood <david.sherwood@arm.com>
20751 * optabs.def (while_ult_optab): New optab.
20752 * doc/md.texi (while_ult@var{m}@var{n}): Document.
20753 * internal-fn.def (WHILE_ULT): New internal function.
20754 * internal-fn.h (direct_internal_fn_supported_p): New override
20755 that takes two types as argument.
20756 * internal-fn.c (while_direct): New macro.
20757 (expand_while_optab_fn): New function.
20758 (convert_optab_supported_p): Likewise.
20759 (direct_while_optab_supported_p): New macro.
20760 * wide-int.h (wi::udiv_ceil): New function.
20761 * tree-vectorizer.h (rgroup_masks): New structure.
20762 (vec_loop_masks): New typedef.
20763 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
20764 and fully_masked_p.
20765 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
20766 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
20767 (vect_max_vf): New function.
20768 (slpeel_make_loop_iterate_ntimes): Delete.
20769 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
20770 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
20771 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
20772 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
20773 internal-fn.h, stor-layout.h and optabs-query.h.
20774 (vect_set_loop_mask): New function.
20775 (add_preheader_seq): Likewise.
20776 (add_header_seq): Likewise.
20777 (interleave_supported_p): Likewise.
20778 (vect_maybe_permute_loop_masks): Likewise.
20779 (vect_set_loop_masks_directly): Likewise.
20780 (vect_set_loop_condition_masked): Likewise.
20781 (vect_set_loop_condition_unmasked): New function, split out from
20782 slpeel_make_loop_iterate_ntimes.
20783 (slpeel_make_loop_iterate_ntimes): Rename to..
20784 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
20785 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
20786 (vect_do_peeling): Update call accordingly.
20787 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
20789 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
20790 mask_compare_type, can_fully_mask_p and fully_masked_p.
20791 (release_vec_loop_masks): New function.
20792 (_loop_vec_info): Use it to free the loop masks.
20793 (can_produce_all_loop_masks_p): New function.
20794 (vect_get_max_nscalars_per_iter): Likewise.
20795 (vect_verify_full_masking): Likewise.
20796 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
20797 retries, and free the mask rgroups before retrying. Check loop-wide
20798 reasons for disallowing fully-masked loops. Make the final decision
20799 about whether use a fully-masked loop or not.
20800 (vect_estimate_min_profitable_iters): Do not assume that peeling
20801 for the number of iterations will be needed for fully-masked loops.
20802 (vectorizable_reduction): Disable fully-masked loops.
20803 (vectorizable_live_operation): Likewise.
20804 (vect_halve_mask_nunits): New function.
20805 (vect_double_mask_nunits): Likewise.
20806 (vect_record_loop_mask): Likewise.
20807 (vect_get_loop_mask): Likewise.
20808 (vect_transform_loop): Handle the case in which the final loop
20809 iteration might handle a partial vector. Call vect_set_loop_condition
20810 instead of slpeel_make_loop_iterate_ntimes.
20811 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
20812 (check_load_store_masking): New function.
20813 (prepare_load_store_mask): Likewise.
20814 (vectorizable_store): Handle fully-masked loops.
20815 (vectorizable_load): Likewise.
20816 (supportable_widening_operation): Use vect_halve_mask_nunits for
20818 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
20819 (vect_gen_while): New function.
20820 * config/aarch64/aarch64.md (umax<mode>3): New expander.
20821 (aarch64_uqdec<mode>): New insn.
20823 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20824 Alan Hayward <alan.hayward@arm.com>
20825 David Sherwood <david.sherwood@arm.com>
20827 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
20828 (reduc_xor_scal_optab): New optabs.
20829 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
20830 (reduc_xor_scal_@var{m}): Document.
20831 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
20832 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
20833 internal functions.
20834 * fold-const-call.c (fold_const_call): Handle them.
20835 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
20836 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
20837 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
20838 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
20839 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
20840 (UNSPEC_XORV): New unspecs.
20841 (optab): Add entries for them.
20842 (BITWISEV): New int iterator.
20843 (bit_reduc_op): New int attributes.
20845 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20846 Alan Hayward <alan.hayward@arm.com>
20847 David Sherwood <david.sherwood@arm.com>
20849 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
20850 * internal-fn.def (VEC_SHL_INSERT): New internal function.
20851 * optabs.def (vec_shl_insert_optab): New optab.
20852 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
20853 (duplicate_and_interleave): Likewise.
20854 * tree-vect-loop.c: Include internal-fn.h.
20855 (neutral_op_for_slp_reduction): New function, split out from
20856 get_initial_defs_for_reduction.
20857 (get_initial_def_for_reduction): Handle option 2 for variable-length
20858 vectors by loading the neutral value into a vector and then shifting
20859 the initial value into element 0.
20860 (get_initial_defs_for_reduction): Replace the code argument with
20861 the neutral value calculated by neutral_op_for_slp_reduction.
20862 Use gimple_build_vector for constant-length vectors.
20863 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
20864 but the first group_size elements have a neutral value.
20865 Use duplicate_and_interleave otherwise.
20866 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
20867 Update call to get_initial_defs_for_reduction. Handle SLP
20868 reductions for variable-length vectors by creating one vector
20869 result for each scalar result, with the elements associated
20870 with other scalar results stubbed out with the neutral value.
20871 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
20872 Require IFN_VEC_SHL_INSERT for double reductions on
20873 variable-length vectors, or SLP reductions that have
20874 a neutral value. Require can_duplicate_and_interleave_p
20875 support for variable-length unchained SLP reductions if there
20876 is no neutral value, such as for MIN/MAX reductions. Also require
20877 the number of vector elements to be a multiple of the number of
20878 SLP statements when doing variable-length unchained SLP reductions.
20879 Update call to vect_create_epilog_for_reduction.
20880 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
20881 and remove initial values.
20882 (duplicate_and_interleave): Make public.
20883 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
20884 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
20886 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20887 Alan Hayward <alan.hayward@arm.com>
20888 David Sherwood <david.sherwood@arm.com>
20890 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
20891 (can_duplicate_and_interleave_p): New function.
20892 (vect_get_and_check_slp_defs): Take the vector of statements
20893 rather than just the current one. Remove excess parentheses.
20894 Restriction rejectinon of vect_constant_def and vect_external_def
20895 for variable-length vectors to boolean types, or types for which
20896 can_duplicate_and_interleave_p is false.
20897 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
20898 (duplicate_and_interleave): New function.
20899 (vect_get_constant_vectors): Use gimple_build_vector for
20900 constant-length vectors and suitable variable-length constant
20901 vectors. Use duplicate_and_interleave for other variable-length
20902 vectors. Don't defer the update when inserting new statements.
20904 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20905 Alan Hayward <alan.hayward@arm.com>
20906 David Sherwood <david.sherwood@arm.com>
20908 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
20909 min_profitable_iters doesn't go negative.
20911 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20912 Alan Hayward <alan.hayward@arm.com>
20913 David Sherwood <david.sherwood@arm.com>
20915 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
20916 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
20917 * optabs.def (vec_mask_load_lanes_optab): New optab.
20918 (vec_mask_store_lanes_optab): Likewise.
20919 * internal-fn.def (MASK_LOAD_LANES): New internal function.
20920 (MASK_STORE_LANES): Likewise.
20921 * internal-fn.c (mask_load_lanes_direct): New macro.
20922 (mask_store_lanes_direct): Likewise.
20923 (expand_mask_load_optab_fn): Handle masked operations.
20924 (expand_mask_load_lanes_optab_fn): New macro.
20925 (expand_mask_store_optab_fn): Handle masked operations.
20926 (expand_mask_store_lanes_optab_fn): New macro.
20927 (direct_mask_load_lanes_optab_supported_p): Likewise.
20928 (direct_mask_store_lanes_optab_supported_p): Likewise.
20929 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
20931 (vect_load_lanes_supported): Likewise.
20932 * tree-vect-data-refs.c (strip_conversion): New function.
20933 (can_group_stmts_p): Likewise.
20934 (vect_analyze_data_ref_accesses): Use it instead of checking
20935 for a pair of assignments.
20936 (vect_store_lanes_supported): Take a masked_p parameter.
20937 (vect_load_lanes_supported): Likewise.
20938 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
20939 vect_store_lanes_supported and vect_load_lanes_supported.
20940 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
20941 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
20942 parameter. Don't allow gaps for masked accesses.
20943 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
20944 and vect_load_lanes_supported.
20945 (get_load_store_type): Take a masked_p parameter and update
20946 call to get_group_load_store_type.
20947 (vectorizable_store): Update call to get_load_store_type.
20948 Handle IFN_MASK_STORE_LANES.
20949 (vectorizable_load): Update call to get_load_store_type.
20950 Handle IFN_MASK_LOAD_LANES.
20952 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20953 Alan Hayward <alan.hayward@arm.com>
20954 David Sherwood <david.sherwood@arm.com>
20956 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
20958 * config/aarch64/aarch64-protos.h
20959 (aarch64_sve_struct_memory_operand_p): Declare.
20960 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
20961 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
20962 (VPRED, vpred): Handle SVE structure modes.
20963 * config/aarch64/constraints.md (Utx): New constraint.
20964 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
20965 (aarch64_sve_struct_nonimmediate_operand): New predicates.
20966 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
20967 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
20968 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
20969 structure modes. Split into pieces after RA.
20970 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
20971 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
20973 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
20974 SVE structure modes.
20975 (aarch64_classify_address): Likewise.
20976 (sizetochar): Move earlier in file.
20977 (aarch64_print_operand): Handle SVE register lists.
20978 (aarch64_array_mode): New function.
20979 (aarch64_sve_struct_memory_operand_p): Likewise.
20980 (TARGET_ARRAY_MODE): Redefine.
20982 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20983 Alan Hayward <alan.hayward@arm.com>
20984 David Sherwood <david.sherwood@arm.com>
20986 * target.def (array_mode): New target hook.
20987 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
20988 * doc/tm.texi: Regenerate.
20989 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
20990 * hooks.c (hook_optmode_mode_uhwi_none): New function.
20991 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
20992 targetm.array_mode.
20993 * stor-layout.c (mode_for_array): Likewise. Support polynomial
20996 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
20997 Alan Hayward <alan.hayward@arm.com>
20998 David Sherwood <david.sherwood@arm.com>
21000 * fold-const.c (fold_binary_loc): Check the argument types
21001 rather than the result type when testing for a vector operation.
21003 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21005 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
21006 * doc/tm.texi: Regenerate.
21008 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21009 Alan Hayward <alan.hayward@arm.com>
21010 David Sherwood <david.sherwood@arm.com>
21012 * doc/invoke.texi (-msve-vector-bits=): Document new option.
21013 (sve): Document new AArch64 extension.
21014 * doc/md.texi (w): Extend the description of the AArch64
21015 constraint to include SVE vectors.
21016 (Upl, Upa): Document new AArch64 predicate constraints.
21017 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
21019 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
21020 (msve-vector-bits=): New option.
21021 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
21022 SVE when these are disabled.
21023 (sve): New extension.
21024 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
21025 modes. Adjust their number of units based on aarch64_sve_vg.
21026 (MAX_BITSIZE_MODE_ANY_MODE): Define.
21027 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
21028 aarch64_addr_query_type.
21029 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
21030 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
21031 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
21032 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
21033 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
21034 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
21035 (aarch64_simd_imm_zero_p): Delete.
21036 (aarch64_check_zero_based_sve_index_immediate): Declare.
21037 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
21038 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
21039 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
21040 (aarch64_sve_float_mul_immediate_p): Likewise.
21041 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
21042 rather than an rtx.
21043 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
21044 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
21045 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
21046 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
21047 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
21048 (aarch64_regmode_natural_size): Likewise.
21049 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
21050 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
21052 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
21053 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
21054 for VG and the SVE predicate registers.
21055 (V_ALIASES): Add a "z"-prefixed alias.
21056 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
21057 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
21058 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
21059 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
21060 (REG_CLASS_NAMES): Add entries for them.
21061 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
21062 and the predicate registers.
21063 (aarch64_sve_vg): Declare.
21064 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
21065 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
21066 (REGMODE_NATURAL_SIZE): Define.
21067 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
21069 * config/aarch64/aarch64.c: Include cfgrtl.h.
21070 (simd_immediate_info): Add a constructor for series vectors,
21071 and an associated step field.
21072 (aarch64_sve_vg): New variable.
21073 (aarch64_dbx_register_number): Handle VG and the predicate registers.
21074 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
21075 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
21076 (VEC_ANY_DATA, VEC_STRUCT): New constants.
21077 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
21078 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
21079 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
21080 (aarch64_get_mask_mode): New functions.
21081 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
21082 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
21083 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
21084 predicate modes and predicate registers. Explicitly restrict
21085 GPRs to modes of 16 bytes or smaller. Only allow FP registers
21086 to store a vector mode if it is recognized by
21087 aarch64_classify_vector_mode.
21088 (aarch64_regmode_natural_size): New function.
21089 (aarch64_hard_regno_caller_save_mode): Return the original mode
21091 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
21092 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
21093 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
21094 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
21096 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
21097 does not overlap dest if the function is frame-related. Handle
21099 (aarch64_split_add_offset): New function.
21100 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
21101 them aarch64_add_offset.
21102 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
21103 and update call to aarch64_sub_sp.
21104 (aarch64_add_cfa_expression): New function.
21105 (aarch64_expand_prologue): Pass extra temporary registers to the
21106 functions above. Handle the case in which we need to emit new
21107 DW_CFA_expressions for registers that were originally saved
21108 relative to the stack pointer, but now have to be expressed
21109 relative to the frame pointer.
21110 (aarch64_output_mi_thunk): Pass extra temporary registers to the
21112 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
21113 IP0 and IP1 values for SVE frames.
21114 (aarch64_expand_vec_series): New function.
21115 (aarch64_expand_sve_widened_duplicate): Likewise.
21116 (aarch64_expand_sve_const_vector): Likewise.
21117 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
21118 Handle SVE constants. Use emit_move_insn to move a force_const_mem
21119 into the register, rather than emitting a SET directly.
21120 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
21121 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
21122 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
21123 (offset_9bit_signed_scaled_p): New functions.
21124 (aarch64_replicate_bitmask_imm): New function.
21125 (aarch64_bitmask_imm): Use it.
21126 (aarch64_cannot_force_const_mem): Reject expressions involving
21127 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
21128 (aarch64_classify_index): Handle SVE indices, by requiring
21129 a plain register index with a scale that matches the element size.
21130 (aarch64_classify_address): Handle SVE addresses. Assert that
21131 the mode of the address is VOIDmode or an integer mode.
21132 Update call to aarch64_classify_symbol.
21133 (aarch64_classify_symbolic_expression): Update call to
21134 aarch64_classify_symbol.
21135 (aarch64_const_vec_all_in_range_p): New function.
21136 (aarch64_print_vector_float_operand): Likewise.
21137 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
21138 "vN" for FP registers with SVE modes. Handle (const ...) vectors
21139 and the FP immediates 1.0 and 0.5.
21140 (aarch64_print_address_internal): Handle SVE addresses.
21141 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
21142 (aarch64_regno_regclass): Handle predicate registers.
21143 (aarch64_secondary_reload): Handle big-endian reloads of SVE
21145 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
21146 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
21147 (aarch64_convert_sve_vector_bits): New function.
21148 (aarch64_override_options): Use it to handle -msve-vector-bits=.
21149 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
21150 rather than an rtx.
21151 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
21152 Handle SVE vector and predicate modes. Accept VL-based constants
21153 that need only one temporary register, and VL offsets that require
21154 no temporary registers.
21155 (aarch64_conditional_register_usage): Mark the predicate registers
21156 as fixed if SVE isn't available.
21157 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
21158 Return true for SVE vector and predicate modes.
21159 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
21160 rather than an unsigned int. Handle SVE modes.
21161 (aarch64_preferred_simd_mode): Update call accordingly. Handle
21163 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
21165 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
21166 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
21167 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
21168 (aarch64_sve_float_mul_immediate_p): New functions.
21169 (aarch64_sve_valid_immediate): New function.
21170 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
21171 Explicitly reject structure modes. Check for INDEX constants.
21172 Handle PTRUE and PFALSE constants.
21173 (aarch64_check_zero_based_sve_index_immediate): New function.
21174 (aarch64_simd_imm_zero_p): Delete.
21175 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
21176 vector modes. Accept constants in the range of CNT[BHWD].
21177 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
21178 ask for an Advanced SIMD mode.
21179 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
21180 (aarch64_simd_vector_alignment): Handle SVE predicates.
21181 (aarch64_vectorize_preferred_vector_alignment): New function.
21182 (aarch64_simd_vector_alignment_reachable): Use it instead of
21184 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
21185 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
21187 (MAX_VECT_LEN): Delete.
21188 (expand_vec_perm_d): Add a vec_flags field.
21189 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
21190 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
21191 (aarch64_evpc_ext): Don't apply a big-endian lane correction
21193 (aarch64_evpc_rev): Rename to...
21194 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
21195 (aarch64_evpc_rev_global): New function.
21196 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
21197 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
21199 (aarch64_evpc_sve_tbl): New function.
21200 (aarch64_expand_vec_perm_const_1): Update after rename of
21201 aarch64_evpc_rev. Handle SVE permutes too, trying
21202 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
21203 than aarch64_evpc_tbl.
21204 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
21205 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
21206 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
21207 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
21208 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
21209 (aarch64_expand_sve_vcond): New functions.
21210 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
21211 of aarch64_vector_mode_p.
21212 (aarch64_dwarf_poly_indeterminate_value): New function.
21213 (aarch64_compute_pressure_classes): Likewise.
21214 (aarch64_can_change_mode_class): Likewise.
21215 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
21216 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
21217 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
21218 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
21219 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
21220 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
21221 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
21222 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
21224 (Dn, Dl, Dr): Accept const as well as const_vector.
21225 (Dz): Likewise. Compare against CONST0_RTX.
21226 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
21227 of "vector" where appropriate.
21228 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
21229 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
21230 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
21231 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
21232 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
21233 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
21234 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
21235 (v_int_equiv): Extend to SVE modes.
21236 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
21238 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
21239 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
21240 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
21241 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
21242 (SVE_COND_FP_CMP): New int iterators.
21243 (perm_hilo): Handle the new unpack unspecs.
21244 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
21246 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
21247 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
21248 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
21249 (aarch64_equality_operator, aarch64_constant_vector_operand)
21250 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
21251 (aarch64_sve_nonimmediate_operand): Likewise.
21252 (aarch64_sve_general_operand): Likewise.
21253 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
21254 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
21255 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
21256 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
21257 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
21258 (aarch64_sve_float_arith_immediate): Likewise.
21259 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
21260 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
21261 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
21262 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
21263 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
21264 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
21265 (aarch64_sve_float_arith_operand): Likewise.
21266 (aarch64_sve_float_arith_with_sub_operand): Likewise.
21267 (aarch64_sve_float_mul_operand): Likewise.
21268 (aarch64_sve_vec_perm_operand): Likewise.
21269 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
21270 (aarch64_mov_operand): Accept const_poly_int and const_vector.
21271 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
21272 as well as const_vector.
21273 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
21274 in file. Use CONST0_RTX and CONSTM1_RTX.
21275 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
21276 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
21277 Use aarch64_simd_imm_zero.
21278 * config/aarch64/aarch64-sve.md: New file.
21279 * config/aarch64/aarch64.md: Include it.
21280 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
21281 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
21282 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
21283 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
21284 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
21285 (sve): New attribute.
21286 (enabled): Disable instructions with the sve attribute unless
21288 (movqi, movhi): Pass CONST_POLY_INT operaneds through
21289 aarch64_expand_mov_immediate.
21290 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
21291 CNT[BHSD] immediates.
21292 (movti): Split CONST_POLY_INT moves into two halves.
21293 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
21294 Split additions that need a temporary here if the destination
21295 is the stack pointer.
21296 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
21297 (*add<mode>3_poly_1): New instruction.
21298 (set_clobber_cc): New expander.
21300 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21302 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
21303 parameter and use it instead of GET_MODE_SIZE (innermode). Use
21304 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
21305 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
21306 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
21307 Change innermode from fixed_mode_size to machine_mode.
21308 (simplify_subreg): Update call accordingly. Handle a constant-sized
21309 subreg of a variable-length CONST_VECTOR.
21311 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21312 Alan Hayward <alan.hayward@arm.com>
21313 David Sherwood <david.sherwood@arm.com>
21315 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
21316 (add_offset_to_base): New function, split out from...
21317 (create_mem_ref): ...here. When handling a scale other than 1,
21318 check first whether the address is valid without the offset.
21319 Add it into the base if so, leaving the index and scale as-is.
21321 2018-01-12 Jakub Jelinek <jakub@redhat.com>
21324 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
21325 fold_for_warn before checking if arg2 is INTEGER_CST.
21327 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
21329 * config/rs6000/predicates.md (load_multiple_operation): Delete.
21330 (store_multiple_operation): Delete.
21331 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
21332 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
21333 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
21334 guarded by TARGET_STRING.
21335 (rs6000_output_load_multiple): Delete.
21336 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21337 OPTION_MASK_STRING / TARGET_STRING handling.
21338 (print_operand) <'N', 'O'>: Add comment that these are unused now.
21339 (const rs6000_opt_masks) <"string">: Change mask to 0.
21340 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
21341 (MASK_STRING): Delete.
21342 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
21344 (load_multiple): Delete.
21351 (store_multiple): Delete.
21358 (movmemsi_8reg): Delete.
21359 (corresponding unnamed define_insn): Delete.
21360 (movmemsi_6reg): Delete.
21361 (corresponding unnamed define_insn): Delete.
21362 (movmemsi_4reg): Delete.
21363 (corresponding unnamed define_insn): Delete.
21364 (movmemsi_2reg): Delete.
21365 (corresponding unnamed define_insn): Delete.
21366 (movmemsi_1reg): Delete.
21367 (corresponding unnamed define_insn): Delete.
21368 * config/rs6000/rs6000.opt (mno-string): New.
21369 (mstring): Replace by deprecation warning stub.
21370 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
21372 2018-01-12 Jakub Jelinek <jakub@redhat.com>
21374 * regrename.c (regrename_do_replace): If replacing the same
21375 reg multiple times, try to reuse last created gen_raw_REG.
21378 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
21379 main to workaround a bug in GDB.
21381 2018-01-12 Tom de Vries <tom@codesourcery.com>
21384 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
21386 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
21388 PR rtl-optimization/80481
21389 * ira-color.c (get_cap_member): New function.
21390 (allocnos_conflict_by_live_ranges_p): Use it.
21391 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
21392 (setup_slot_coalesced_allocno_live_ranges): Ditto.
21394 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
21397 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
21398 (*saddl_se_1): Ditto.
21399 (*ssubsi_1): Ditto.
21400 (*ssubl_se_1): Ditto.
21402 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
21404 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
21405 rather than wi::to_widest for DR_INITs.
21406 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
21407 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
21408 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
21410 (vect_analyze_group_access_1): Note that here.
21412 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
21414 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
21415 polynomial type sizes.
21417 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
21419 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
21420 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
21421 (gimple_add_tmp_var): Likewise.
21423 2018-01-12 Martin Liska <mliska@suse.cz>
21425 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
21426 (gimple_alloc_sizes): Likewise.
21427 (dump_gimple_statistics): Use PRIu64 in printf format.
21428 * gimple.h: Change uint64_t to int.
21430 2018-01-12 Martin Liska <mliska@suse.cz>
21432 * tree-core.h: Use uint64_t instead of int.
21433 * tree.c (tree_node_counts): Likewise.
21434 (tree_node_sizes): Likewise.
21435 (dump_tree_statistics): Use PRIu64 in printf format.
21437 2018-01-12 Martin Liska <mliska@suse.cz>
21439 * Makefile.in: As qsort_chk is implemented in vec.c, add
21440 vec.o to linkage of gencfn-macros.
21441 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
21442 passing the info to record_node_allocation_statistics.
21443 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
21445 * ggc-common.c (struct ggc_usage): Add operator== and use
21446 it in operator< and compare function.
21447 * mem-stats.h (struct mem_usage): Likewise.
21448 * vec.c (struct vec_usage): Remove operator< and compare
21449 function. Can be simply inherited.
21451 2018-01-12 Martin Jambor <mjambor@suse.cz>
21454 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
21455 * tree-ssa-math-opts.c: Include domwalk.h.
21456 (convert_mult_to_fma_1): New function.
21457 (fma_transformation_info): New type.
21458 (fma_deferring_state): Likewise.
21459 (cancel_fma_deferring): New function.
21460 (result_of_phi): Likewise.
21461 (last_fma_candidate_feeds_initial_phi): Likewise.
21462 (convert_mult_to_fma): Added deferring logic, split actual
21463 transformation to convert_mult_to_fma_1.
21464 (math_opts_dom_walker): New type.
21465 (math_opts_dom_walker::after_dom_children): New method, body moved
21466 here from pass_optimize_widening_mul::execute, added deferring logic
21468 (pass_optimize_widening_mul::execute): Moved most of code to
21469 math_opts_dom_walker::after_dom_children.
21470 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
21471 * config/i386/i386.c (ix86_option_override_internal): Added
21472 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
21474 2018-01-12 Richard Biener <rguenther@suse.de>
21477 * dwarf2out.c (gen_variable_die): Do not reset old_die for
21478 inline instance vars.
21480 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
21483 * config/rx/rx.c (rx_is_restricted_memory_address):
21484 Handle SUBREG case.
21486 2018-01-12 Richard Biener <rguenther@suse.de>
21488 PR tree-optimization/80846
21489 * target.def (split_reduction): New target hook.
21490 * targhooks.c (default_split_reduction): New function.
21491 * targhooks.h (default_split_reduction): Declare.
21492 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
21493 target requests first reduce vectors by combining low and high
21495 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
21496 (get_vectype_for_scalar_type_and_size): Export.
21497 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
21498 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
21499 * doc/tm.texi: Regenerate.
21500 * config/i386/i386.c (ix86_split_reduction): Implement
21501 TARGET_VECTORIZE_SPLIT_REDUCTION.
21503 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
21506 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
21507 in PIC mode except for TARGET_VXWORKS_RTP.
21508 * config/sparc/sparc.c: Include cfgrtl.h.
21509 (TARGET_INIT_PIC_REG): Define.
21510 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
21511 (sparc_pic_register_p): New predicate.
21512 (sparc_legitimate_address_p): Use it.
21513 (sparc_legitimize_pic_address): Likewise.
21514 (sparc_delegitimize_address): Likewise.
21515 (sparc_mode_dependent_address_p): Likewise.
21516 (gen_load_pcrel_sym): Remove 4th parameter.
21517 (load_got_register): Adjust call to above. Remove obsolete stuff.
21518 (sparc_expand_prologue): Do not call load_got_register here.
21519 (sparc_flat_expand_prologue): Likewise.
21520 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
21521 (sparc_use_pseudo_pic_reg): New function.
21522 (sparc_init_pic_reg): Likewise.
21523 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
21524 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
21526 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
21528 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21529 Add item for branch_cost.
21531 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
21533 PR rtl-optimization/83565
21534 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
21535 not extend the result to a larger mode for rotate operations.
21536 (num_sign_bit_copies1): Likewise.
21538 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21541 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
21543 Use values-Xc.o for -pedantic.
21544 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
21546 2018-01-12 Martin Liska <mliska@suse.cz>
21549 * ipa-devirt.c (final_warning_record::grow_type_warnings):
21551 (possible_polymorphic_call_targets): Use it.
21552 (ipa_devirt): Likewise.
21554 2018-01-12 Martin Liska <mliska@suse.cz>
21556 * profile-count.h (enum profile_quality): Use 0 as invalid
21557 enum value of profile_quality.
21559 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
21561 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
21562 -mext-string options.
21564 2018-01-12 Richard Biener <rguenther@suse.de>
21566 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
21567 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
21568 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
21570 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
21572 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
21574 * configure.ac (--with-long-double-format): Add support for the
21575 configuration option to change the default long double format on
21577 * config.gcc (powerpc*-linux*-*): Likewise.
21578 * configure: Regenerate.
21579 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
21580 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
21581 used without modification.
21583 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21585 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
21586 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
21587 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
21588 MISC_BUILTIN_SPEC_BARRIER.
21589 (rs6000_init_builtins): Likewise.
21590 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
21592 (speculation_barrier): New define_insn.
21593 * doc/extend.texi: Document __builtin_speculation_barrier.
21595 2018-01-11 Jakub Jelinek <jakub@redhat.com>
21598 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
21599 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
21600 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
21602 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
21603 integral modes instead of "ss" and "sd".
21604 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
21605 vectors with 32-bit and 64-bit elements.
21606 (vecdupssescalarmodesuffix): New mode attribute.
21607 (vec_dup<mode>): Use it.
21609 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
21612 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
21613 frame if argument is passed on stack.
21615 2018-01-11 Jakub Jelinek <jakub@redhat.com>
21618 * ree.c (combine_reaching_defs): Optimize also
21619 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
21620 reg2=any_extend(exp); reg1=reg2;, formatting fix.
21622 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
21624 PR middle-end/83189
21625 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
21627 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
21629 PR middle-end/83718
21630 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
21631 after they are computed.
21633 2018-01-11 Bin Cheng <bin.cheng@arm.com>
21635 PR tree-optimization/83695
21636 * gimple-loop-linterchange.cc
21637 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
21638 reset cached scev information after interchange.
21639 (pass_linterchange::execute): Remove call to scev_reset_htab.
21641 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21643 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
21644 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
21645 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
21646 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
21647 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
21648 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
21649 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
21650 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
21651 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
21652 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
21653 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
21654 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
21655 (V_lane_reg): Likewise.
21656 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
21658 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
21659 (vfmal_lane_low<mode>_intrinsic,
21660 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
21661 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
21662 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
21663 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
21664 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
21665 vfmsl_lane_high<mode>_intrinsic): New define_insns.
21667 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21669 * config/arm/arm-cpus.in (fp16fml): New feature.
21670 (ALL_SIMD): Add fp16fml.
21671 (armv8.2-a): Add fp16fml as an option.
21672 (armv8.3-a): Likewise.
21673 (armv8.4-a): Add fp16fml as part of fp16.
21674 * config/arm/arm.h (TARGET_FP16FML): Define.
21675 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
21677 * config/arm/arm-modes.def (V2HF): Define.
21678 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
21679 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
21680 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
21681 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
21682 vfmsl_low, vfmsl_high): New set of builtins.
21683 * config/arm/iterators.md (PLUSMINUS): New code iterator.
21684 (vfml_op): New code attribute.
21685 (VFMLHALVES): New int iterator.
21686 (VFML, VFMLSEL): New mode attributes.
21687 (V_reg): Define mapping for V2HF.
21688 (V_hi, V_lo): New mode attributes.
21689 (VF_constraint): Likewise.
21690 (vfml_half, vfml_half_selector): New int attributes.
21691 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
21693 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
21694 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
21696 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
21697 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
21698 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
21699 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
21701 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
21702 Document new effective target and option set.
21704 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21706 * config/arm/arm-cpus.in (armv8_4): New feature.
21707 (ARMv8_4a): New fgroup.
21708 (armv8.4-a): New arch.
21709 * config/arm/arm-tables.opt: Regenerate.
21710 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
21711 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
21712 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
21713 Add matching rules for -march=armv8.4-a and extensions.
21714 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
21716 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
21719 * config/rx/rx.md (BW): New mode attribute.
21720 (sync_lock_test_and_setsi): Add mode suffix to insn output.
21722 2018-01-11 Richard Biener <rguenther@suse.de>
21724 PR tree-optimization/83435
21725 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
21726 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
21727 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
21729 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
21730 Alan Hayward <alan.hayward@arm.com>
21731 David Sherwood <david.sherwood@arm.com>
21733 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
21735 (aarch64_classify_address): Initialize it. Track polynomial offsets.
21736 (aarch64_print_address_internal): Use it to check for a zero offset.
21738 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
21739 Alan Hayward <alan.hayward@arm.com>
21740 David Sherwood <david.sherwood@arm.com>
21742 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
21743 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
21744 Return a poly_int64 rather than a HOST_WIDE_INT.
21745 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
21746 rather than a HOST_WIDE_INT.
21747 * config/aarch64/aarch64.h (aarch64_frame): Protect with
21748 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
21749 hard_fp_offset, frame_size, initial_adjust, callee_offset and
21750 final_offset from HOST_WIDE_INT to poly_int64.
21751 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
21752 to_constant when getting the number of units in an Advanced SIMD
21754 (aarch64_builtin_vectorized_function): Check for a constant number
21756 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
21758 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
21759 attribute instead of GET_MODE_NUNITS.
21760 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
21761 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
21762 GET_MODE_SIZE for fixed-size registers.
21763 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
21764 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
21765 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
21766 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
21767 (aarch64_print_operand, aarch64_print_address_internal)
21768 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
21769 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
21770 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
21771 Handle polynomial GET_MODE_SIZE.
21772 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
21773 wider than SImode without modification.
21774 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
21775 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
21776 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
21777 passing and returning SVE modes.
21778 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
21779 rather than GEN_INT.
21780 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
21781 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
21782 (aarch64_allocate_and_probe_stack_space): Likewise.
21783 (aarch64_layout_frame): Cope with polynomial offsets.
21784 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
21785 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
21786 polynomial offsets.
21787 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
21788 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
21789 poly_int64 rather than a HOST_WIDE_INT.
21790 (aarch64_get_separate_components, aarch64_process_components)
21791 (aarch64_expand_prologue, aarch64_expand_epilogue)
21792 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
21793 (aarch64_anchor_offset): New function, split out from...
21794 (aarch64_legitimize_address): ...here.
21795 (aarch64_builtin_vectorization_cost): Handle polynomial
21796 TYPE_VECTOR_SUBPARTS.
21797 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
21799 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
21800 number of elements from the PARALLEL rather than the mode.
21801 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
21802 rather than GET_MODE_BITSIZE.
21803 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
21804 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
21805 (aarch64_expand_vec_perm_const_1): Handle polynomial
21806 d->perm.length () and d->perm elements.
21807 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
21808 Apply to_constant to d->perm elements.
21809 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
21810 polynomial CONST_VECTOR_NUNITS.
21811 (aarch64_move_pointer): Take amount as a poly_int64 rather
21813 (aarch64_progress_pointer): Avoid temporary variable.
21814 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
21815 the mode attribute instead of GET_MODE.
21817 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
21818 Alan Hayward <alan.hayward@arm.com>
21819 David Sherwood <david.sherwood@arm.com>
21821 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
21822 x exists before using it.
21823 (aarch64_add_constant_internal): Rename to...
21824 (aarch64_add_offset_1): ...this. Replace regnum with separate
21825 src and dest rtxes. Handle the case in which they're different,
21826 including when the offset is zero. Replace scratchreg with an rtx.
21827 Use 2 additions if there is no spare register into which we can
21828 move a 16-bit constant.
21829 (aarch64_add_constant): Delete.
21830 (aarch64_add_offset): Replace reg with separate src and dest
21831 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
21832 Use aarch64_add_offset_1.
21833 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
21834 an rtx rather than an int. Take the delta as a poly_int64
21835 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
21836 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
21837 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
21838 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
21839 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
21840 and aarch64_add_sp.
21841 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
21842 aarch64_add_constant.
21844 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
21846 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
21847 Use scalar_float_mode.
21849 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
21851 * config/aarch64/aarch64-simd.md
21852 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
21853 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
21854 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
21855 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
21856 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
21857 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
21858 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
21859 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
21860 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
21861 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
21863 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21866 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
21867 targ_options->x_arm_arch_string is non NULL.
21869 2018-01-11 Tamar Christina <tamar.christina@arm.com>
21871 * config/aarch64/aarch64.h
21872 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
21874 2018-01-11 Sudakshina Das <sudi.das@arm.com>
21877 * expmed.c (emit_store_flag_force): Swap if const op0
21878 and change VOIDmode to mode of op0.
21880 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
21882 PR rtl-optimization/83761
21883 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
21884 than bytes to mode_for_size.
21886 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
21888 PR middle-end/83189
21889 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
21890 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
21893 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
21895 PR middle-end/83575
21896 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
21897 when in layout mode.
21898 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
21899 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
21902 2018-01-10 Michael Collison <michael.collison@arm.com>
21904 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
21905 * config/aarch64/aarch64-option-extension.def: Add
21906 AARCH64_OPT_EXTENSION of 'fp16fml'.
21907 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
21908 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
21909 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
21910 * config/aarch64/constraints.md (Ui7): New constraint.
21911 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
21912 (VFMLA_SEL_W): Ditto.
21915 (VFMLA16_LOW): New int iterator.
21916 (VFMLA16_HIGH): Ditto.
21917 (UNSPEC_FMLAL): New unspec.
21918 (UNSPEC_FMLSL): Ditto.
21919 (UNSPEC_FMLAL2): Ditto.
21920 (UNSPEC_FMLSL2): Ditto.
21921 (f16mac): New code attribute.
21922 * config/aarch64/aarch64-simd-builtins.def
21923 (aarch64_fmlal_lowv2sf): Ditto.
21924 (aarch64_fmlsl_lowv2sf): Ditto.
21925 (aarch64_fmlalq_lowv4sf): Ditto.
21926 (aarch64_fmlslq_lowv4sf): Ditto.
21927 (aarch64_fmlal_highv2sf): Ditto.
21928 (aarch64_fmlsl_highv2sf): Ditto.
21929 (aarch64_fmlalq_highv4sf): Ditto.
21930 (aarch64_fmlslq_highv4sf): Ditto.
21931 (aarch64_fmlal_lane_lowv2sf): Ditto.
21932 (aarch64_fmlsl_lane_lowv2sf): Ditto.
21933 (aarch64_fmlal_laneq_lowv2sf): Ditto.
21934 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
21935 (aarch64_fmlalq_lane_lowv4sf): Ditto.
21936 (aarch64_fmlsl_lane_lowv4sf): Ditto.
21937 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
21938 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
21939 (aarch64_fmlal_lane_highv2sf): Ditto.
21940 (aarch64_fmlsl_lane_highv2sf): Ditto.
21941 (aarch64_fmlal_laneq_highv2sf): Ditto.
21942 (aarch64_fmlsl_laneq_highv2sf): Ditto.
21943 (aarch64_fmlalq_lane_highv4sf): Ditto.
21944 (aarch64_fmlsl_lane_highv4sf): Ditto.
21945 (aarch64_fmlalq_laneq_highv4sf): Ditto.
21946 (aarch64_fmlsl_laneq_highv4sf): Ditto.
21947 * config/aarch64/aarch64-simd.md:
21948 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
21949 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
21950 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
21951 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
21952 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
21953 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
21954 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
21955 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
21956 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
21957 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
21958 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
21959 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
21960 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
21961 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
21962 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
21963 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
21964 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
21965 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
21966 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
21967 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
21968 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
21969 (vfmlsl_low_u32): Ditto.
21970 (vfmlalq_low_u32): Ditto.
21971 (vfmlslq_low_u32): Ditto.
21972 (vfmlal_high_u32): Ditto.
21973 (vfmlsl_high_u32): Ditto.
21974 (vfmlalq_high_u32): Ditto.
21975 (vfmlslq_high_u32): Ditto.
21976 (vfmlal_lane_low_u32): Ditto.
21977 (vfmlsl_lane_low_u32): Ditto.
21978 (vfmlal_laneq_low_u32): Ditto.
21979 (vfmlsl_laneq_low_u32): Ditto.
21980 (vfmlalq_lane_low_u32): Ditto.
21981 (vfmlslq_lane_low_u32): Ditto.
21982 (vfmlalq_laneq_low_u32): Ditto.
21983 (vfmlslq_laneq_low_u32): Ditto.
21984 (vfmlal_lane_high_u32): Ditto.
21985 (vfmlsl_lane_high_u32): Ditto.
21986 (vfmlal_laneq_high_u32): Ditto.
21987 (vfmlsl_laneq_high_u32): Ditto.
21988 (vfmlalq_lane_high_u32): Ditto.
21989 (vfmlslq_lane_high_u32): Ditto.
21990 (vfmlalq_laneq_high_u32): Ditto.
21991 (vfmlslq_laneq_high_u32): Ditto.
21992 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
21993 (AARCH64_FL_FOR_ARCH8_4): New.
21994 (AARCH64_ISA_F16FML): New ISA flag.
21995 (TARGET_F16FML): New feature flag for fp16fml.
21996 (doc/invoke.texi): Document new fp16fml option.
21998 2018-01-10 Michael Collison <michael.collison@arm.com>
22000 * config/aarch64/aarch64-builtins.c:
22001 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
22002 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
22003 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
22004 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
22005 (AARCH64_ISA_SHA3): New ISA flag.
22006 (TARGET_SHA3): New feature flag for sha3.
22007 * config/aarch64/iterators.md (sha512_op): New int attribute.
22008 (CRYPTO_SHA512): New int iterator.
22009 (UNSPEC_SHA512H): New unspec.
22010 (UNSPEC_SHA512H2): Ditto.
22011 (UNSPEC_SHA512SU0): Ditto.
22012 (UNSPEC_SHA512SU1): Ditto.
22013 * config/aarch64/aarch64-simd-builtins.def
22014 (aarch64_crypto_sha512hqv2di): New builtin.
22015 (aarch64_crypto_sha512h2qv2di): Ditto.
22016 (aarch64_crypto_sha512su0qv2di): Ditto.
22017 (aarch64_crypto_sha512su1qv2di): Ditto.
22018 (aarch64_eor3qv8hi): Ditto.
22019 (aarch64_rax1qv2di): Ditto.
22020 (aarch64_xarqv2di): Ditto.
22021 (aarch64_bcaxqv8hi): Ditto.
22022 * config/aarch64/aarch64-simd.md:
22023 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
22024 (aarch64_crypto_sha512su0qv2di): Ditto.
22025 (aarch64_crypto_sha512su1qv2di): Ditto.
22026 (aarch64_eor3qv8hi): Ditto.
22027 (aarch64_rax1qv2di): Ditto.
22028 (aarch64_xarqv2di): Ditto.
22029 (aarch64_bcaxqv8hi): Ditto.
22030 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
22031 (vsha512h2q_u64): Ditto.
22032 (vsha512su0q_u64): Ditto.
22033 (vsha512su1q_u64): Ditto.
22034 (veor3q_u16): Ditto.
22035 (vrax1q_u64): Ditto.
22036 (vxarq_u64): Ditto.
22037 (vbcaxq_u16): Ditto.
22038 * config/arm/types.md (crypto_sha512): New type attribute.
22039 (crypto_sha3): Ditto.
22040 (doc/invoke.texi): Document new sha3 option.
22042 2018-01-10 Michael Collison <michael.collison@arm.com>
22044 * config/aarch64/aarch64-builtins.c:
22045 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
22046 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
22047 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
22048 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
22049 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
22050 (AARCH64_ISA_SM4): New ISA flag.
22051 (TARGET_SM4): New feature flag for sm4.
22052 * config/aarch64/aarch64-simd-builtins.def
22053 (aarch64_sm3ss1qv4si): Ditto.
22054 (aarch64_sm3tt1aq4si): Ditto.
22055 (aarch64_sm3tt1bq4si): Ditto.
22056 (aarch64_sm3tt2aq4si): Ditto.
22057 (aarch64_sm3tt2bq4si): Ditto.
22058 (aarch64_sm3partw1qv4si): Ditto.
22059 (aarch64_sm3partw2qv4si): Ditto.
22060 (aarch64_sm4eqv4si): Ditto.
22061 (aarch64_sm4ekeyqv4si): Ditto.
22062 * config/aarch64/aarch64-simd.md:
22063 (aarch64_sm3ss1qv4si): Ditto.
22064 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
22065 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
22066 (aarch64_sm4eqv4si): Ditto.
22067 (aarch64_sm4ekeyqv4si): Ditto.
22068 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
22069 (sm3part_op): Ditto.
22070 (CRYPTO_SM3TT): Ditto.
22071 (CRYPTO_SM3PART): Ditto.
22072 (UNSPEC_SM3SS1): New unspec.
22073 (UNSPEC_SM3TT1A): Ditto.
22074 (UNSPEC_SM3TT1B): Ditto.
22075 (UNSPEC_SM3TT2A): Ditto.
22076 (UNSPEC_SM3TT2B): Ditto.
22077 (UNSPEC_SM3PARTW1): Ditto.
22078 (UNSPEC_SM3PARTW2): Ditto.
22079 (UNSPEC_SM4E): Ditto.
22080 (UNSPEC_SM4EKEY): Ditto.
22081 * config/aarch64/constraints.md (Ui2): New constraint.
22082 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
22083 * config/arm/types.md (crypto_sm3): New type attribute.
22084 (crypto_sm4): Ditto.
22085 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
22086 (vsm3tt1aq_u32): Ditto.
22087 (vsm3tt1bq_u32): Ditto.
22088 (vsm3tt2aq_u32): Ditto.
22089 (vsm3tt2bq_u32): Ditto.
22090 (vsm3partw1q_u32): Ditto.
22091 (vsm3partw2q_u32): Ditto.
22092 (vsm4eq_u32): Ditto.
22093 (vsm4ekeyq_u32): Ditto.
22094 (doc/invoke.texi): Document new sm4 option.
22096 2018-01-10 Michael Collison <michael.collison@arm.com>
22098 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
22099 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
22100 (AARCH64_FL_FOR_ARCH8_4): New.
22101 (AARCH64_FL_V8_4): New flag.
22102 (doc/invoke.texi): Document new armv8.4-a option.
22104 2018-01-10 Michael Collison <michael.collison@arm.com>
22106 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
22107 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
22108 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
22109 * config/aarch64/aarch64-option-extension.def: Add
22110 AARCH64_OPT_EXTENSION of 'sha2'.
22111 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
22112 (crypto): Disable sha2 and aes if crypto disabled.
22113 (crypto): Enable aes and sha2 if enabled.
22114 (simd): Disable sha2 and aes if simd disabled.
22115 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
22117 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
22118 (TARGET_SHA2): New feature flag for sha2.
22119 (TARGET_AES): New feature flag for aes.
22120 * config/aarch64/aarch64-simd.md:
22121 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
22122 conditional on TARGET_AES.
22123 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
22124 (aarch64_crypto_sha1hsi): Make pattern conditional
22126 (aarch64_crypto_sha1hv4si): Ditto.
22127 (aarch64_be_crypto_sha1hv4si): Ditto.
22128 (aarch64_crypto_sha1su1v4si): Ditto.
22129 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
22130 (aarch64_crypto_sha1su0v4si): Ditto.
22131 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
22132 (aarch64_crypto_sha256su0v4si): Ditto.
22133 (aarch64_crypto_sha256su1v4si): Ditto.
22134 (doc/invoke.texi): Document new aes and sha2 options.
22136 2018-01-10 Martin Sebor <msebor@redhat.com>
22138 PR tree-optimization/83781
22139 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
22142 2018-01-11 Martin Sebor <msebor@gmail.com>
22143 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22145 PR tree-optimization/83501
22146 PR tree-optimization/81703
22148 * tree-ssa-strlen.c (get_string_cst): Rename...
22149 (get_string_len): ...to this. Handle global constants.
22150 (handle_char_store): Adjust.
22152 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
22153 Jim Wilson <jimw@sifive.com>
22155 * config/riscv/riscv-protos.h (riscv_output_return): New.
22156 * config/riscv/riscv.c (struct machine_function): New naked_p field.
22157 (riscv_attribute_table, riscv_output_return),
22158 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
22159 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
22160 (riscv_compute_frame_info): Only compute frame->mask if not a naked
22162 (riscv_expand_prologue): Add early return for naked function.
22163 (riscv_expand_epilogue): Likewise.
22164 (riscv_function_ok_for_sibcall): Return false for naked function.
22165 (riscv_set_current_function): New.
22166 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
22167 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
22168 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
22169 * doc/extend.texi (RISC-V Function Attributes): New.
22171 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
22173 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
22174 check for 128-bit long double before checking TCmode.
22175 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
22176 128-bit long doubles before checking TFmode or TCmode.
22177 (FLOAT128_IBM_P): Likewise.
22179 2018-01-10 Martin Sebor <msebor@redhat.com>
22181 PR tree-optimization/83671
22182 * builtins.c (c_strlen): Unconditionally return zero for the empty
22184 Use -Warray-bounds for warnings.
22185 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
22186 for non-constant array indices with COMPONENT_REF, arrays of
22187 arrays, and pointers to arrays.
22188 (gimple_fold_builtin_strlen): Determine and set length range for
22189 non-constant character arrays.
22191 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
22193 PR middle-end/81897
22194 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
22197 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
22199 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
22201 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
22204 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
22205 VECTOR_MEM_ALTIVEC_OR_VSX_P.
22206 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
22207 indexed_or_indirect_operand predicate.
22208 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
22209 (*vsx_le_perm_load_v8hi): Likewise.
22210 (*vsx_le_perm_load_v16qi): Likewise.
22211 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
22212 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
22213 (*vsx_le_perm_store_v8hi): Likewise.
22214 (*vsx_le_perm_store_v16qi): Likewise.
22215 (eight unnamed splitters): Likewise.
22217 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
22219 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
22220 * config/rs6000/emmintrin.h: Likewise.
22221 * config/rs6000/mmintrin.h: Likewise.
22222 * config/rs6000/xmmintrin.h: Likewise.
22224 2018-01-10 David Malcolm <dmalcolm@redhat.com>
22227 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
22229 * tree.c (tree_nop_conversion): Return true for location wrapper
22231 (maybe_wrap_with_location): New function.
22232 (selftest::check_strip_nops): New function.
22233 (selftest::test_location_wrappers): New function.
22234 (selftest::tree_c_tests): Call it.
22235 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
22236 (maybe_wrap_with_location): New decl.
22237 (EXPR_LOCATION_WRAPPER_P): New macro.
22238 (location_wrapper_p): New inline function.
22239 (tree_strip_any_location_wrapper): New inline function.
22241 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
22244 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
22245 stack_realign_offset for the largest alignment of stack slot
22247 (ix86_find_max_used_stack_alignment): New function.
22248 (ix86_finalize_stack_frame_flags): Use it. Set
22249 max_used_stack_alignment if we don't realign stack.
22250 * config/i386/i386.h (machine_function): Add
22251 max_used_stack_alignment.
22253 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
22255 * config/arm/arm.opt (-mbranch-cost): New option.
22256 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
22259 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
22262 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
22263 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
22265 2018-01-10 Richard Biener <rguenther@suse.de>
22268 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
22269 early out so it also covers the case where we have a non-NULL
22272 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
22274 PR tree-optimization/83753
22275 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
22276 for non-strided grouped accesses if the number of elements is 1.
22278 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
22281 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
22282 * i386.h (TARGET_USE_GATHER): Define.
22283 * x86-tune.def (X86_TUNE_USE_GATHER): New.
22285 2018-01-10 Martin Liska <mliska@suse.cz>
22288 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
22289 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
22291 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
22292 CLEANUP_NO_PARTITIONING is not set.
22294 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
22296 * doc/rtl.texi: Remove documentation of (const ...) wrappers
22297 for vectors, as a partial revert of r254296.
22298 * rtl.h (const_vec_p): Delete.
22299 (const_vec_duplicate_p): Don't test for vector CONSTs.
22300 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
22301 * expmed.c (make_tree): Likewise.
22304 * common.md (E, F): Use CONSTANT_P instead of checking for
22306 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
22307 checking for CONST_VECTOR.
22309 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
22311 PR middle-end/83575
22312 * predict.c (force_edge_cold): Handle in more sane way edges
22313 with no prediction.
22315 2018-01-09 Carl Love <cel@us.ibm.com>
22317 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
22319 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
22320 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
22321 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
22322 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
22323 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
22324 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
22325 * config/rs6000/rs6000-protos.h: Add extern defition for
22326 rs6000_generate_float2_double_code.
22327 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
22329 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
22330 (float2_v2df): Add define_expand.
22332 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
22335 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
22336 op_mode in the force_to_mode call.
22338 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
22340 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
22341 instead of checking each element individually.
22342 (aarch64_evpc_uzp): Likewise.
22343 (aarch64_evpc_zip): Likewise.
22344 (aarch64_evpc_ext): Likewise.
22345 (aarch64_evpc_rev): Likewise.
22346 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
22347 instead of checking each element individually. Return true without
22349 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
22350 whether all selected elements come from the same input, instead of
22351 checking each element individually. Remove calls to gen_rtx_REG,
22352 start_sequence and end_sequence and instead assert that no rtl is
22355 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
22357 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
22358 order of HIGH and CONST checks.
22360 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
22362 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
22363 if the destination isn't an SSA_NAME.
22365 2018-01-09 Richard Biener <rguenther@suse.de>
22367 PR tree-optimization/83668
22368 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
22370 (canonicalize_loop_form): ... here, renamed from ...
22371 (canonicalize_loop_closed_ssa_form): ... this and amended to
22372 swap successor edges for loop exit blocks to make us use
22373 the RPO order we need for initial schedule generation.
22375 2018-01-09 Joseph Myers <joseph@codesourcery.com>
22377 PR tree-optimization/64811
22378 * match.pd: When optimizing comparisons with Inf, avoid
22379 introducing or losing exceptions from comparisons with NaN.
22381 2018-01-09 Martin Liska <mliska@suse.cz>
22384 * asan.c (shadow_mem_size): Add gcc_assert.
22386 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
22388 Don't save registers in main().
22391 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
22392 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
22393 * config/avr/avr.c (avr_set_current_function): Don't error if
22394 naked, OS_task or OS_main are specified at the same time.
22395 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
22397 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
22399 * common/config/avr/avr-common.c (avr_option_optimization_table):
22400 Switch on -mmain-is-OS_task for optimizing compilations.
22402 2018-01-09 Richard Biener <rguenther@suse.de>
22404 PR tree-optimization/83572
22405 * graphite.c: Include cfganal.h.
22406 (graphite_transform_loops): Connect infinite loops to exit
22407 and remove fake edges at the end.
22409 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
22411 * ipa-inline.c (edge_badness): Revert accidental checkin.
22413 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
22416 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
22417 symbols; not inline clones.
22419 2018-01-09 Jakub Jelinek <jakub@redhat.com>
22422 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
22423 hard registers. Formatting fixes.
22425 PR preprocessor/83722
22426 * gcc.c (try_generate_repro): Pass
22427 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
22428 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
22431 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
22432 Kito Cheng <kito.cheng@gmail.com>
22434 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
22435 (riscv_leaf_function_p): Delete.
22436 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
22438 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22440 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
22442 (do_ifelse): New function.
22443 (do_isel): New function.
22444 (do_sub3): New function.
22445 (do_add3): New function.
22446 (do_load_mask_compare): New function.
22447 (do_overlap_load_compare): New function.
22448 (expand_compare_loop): New function.
22449 (expand_block_compare): Call expand_compare_loop() when appropriate.
22450 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
22451 option description.
22452 (-mblock-compare-inline-loop-limit): New option.
22454 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22457 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
22458 Reverse order of second and third operands in first alternative.
22459 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
22460 of first and second elements in UNSPEC_VPERMR vector.
22461 (altivec_expand_vec_perm_le): Likewise.
22463 2018-01-08 Jeff Law <law@redhat.com>
22465 PR rtl-optimizatin/81308
22466 * tree-switch-conversion.c (cfg_altered): New file scoped static.
22467 (process_switch): If group_case_labels makes a change, then set
22469 (pass_convert_switch::execute): If a switch is converted, then
22470 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
22472 PR rtl-optimization/81308
22473 * recog.c (split_all_insns): Conditionally cleanup the CFG after
22476 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
22478 PR target/83663 - Revert r255946
22479 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
22480 generation for cases where splatting a value is not useful.
22481 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
22482 across a vec_duplicate and a paradoxical subreg forming a vector
22483 mode to a vec_concat.
22485 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22487 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
22488 -march=armv8.3-a variants.
22489 * config/arm/t-multilib: Likewise.
22490 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
22492 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22494 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
22496 (cceq_ior_compare_complement): Give it a name so I can use it, and
22497 change boolean_or_operator predicate to boolean_operator so it can
22498 be used to generate a crand.
22499 (eqne): New code iterator.
22500 (bd/bd_neg): New code_attrs.
22501 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
22502 a single define_insn.
22503 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
22504 decrement (bdnzt/bdnzf/bdzt/bdzf).
22505 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
22506 with the new names of the branch decrement patterns, and added the
22507 names of the branch decrement conditional patterns.
22509 2018-01-08 Richard Biener <rguenther@suse.de>
22511 PR tree-optimization/83563
22512 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
22515 2018-01-08 Richard Biener <rguenther@suse.de>
22517 PR middle-end/83713
22518 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
22520 2018-01-08 Richard Biener <rguenther@suse.de>
22522 PR tree-optimization/83685
22523 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
22524 references to abnormals.
22526 2018-01-08 Richard Biener <rguenther@suse.de>
22529 * dwarf2out.c (output_indirect_strings): Handle empty
22530 skeleton_debug_str_hash.
22531 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
22533 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
22535 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
22536 (emit_store_direct): Likewise.
22537 (arc_trampoline_adjust_address): Likewise.
22538 (arc_asm_trampoline_template): New function.
22539 (arc_initialize_trampoline): Use asm_trampoline_template.
22540 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
22541 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
22542 * config/arc/arc.md (flush_icache): Delete pattern.
22544 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
22546 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
22547 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
22550 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
22553 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
22554 by not USED_FOR_TARGET.
22555 (make_pass_resolve_sw_modes): Likewise.
22557 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
22559 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
22562 2018-01-08 Richard Biener <rguenther@suse.de>
22564 PR middle-end/83580
22565 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
22567 2018-01-08 Richard Biener <rguenther@suse.de>
22569 PR middle-end/83517
22570 * match.pd ((t * 2) / 2) -> t): Add missing :c.
22572 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
22574 PR middle-end/81897
22575 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
22576 basic blocks with a small number of successors.
22577 (convert_control_dep_chain_into_preds): Improve handling of
22579 (dump_predicates): Split apart into...
22580 (dump_pred_chain): ...here...
22581 (dump_pred_info): ...and here.
22582 (can_one_predicate_be_invalidated_p): Add debugging printfs.
22583 (can_chain_union_be_invalidated_p): Improve check for invalidation
22585 (uninit_uses_cannot_happen): Avoid unnecessary if
22586 convert_control_dep_chain_into_preds yielded nothing.
22588 2018-01-06 Martin Sebor <msebor@redhat.com>
22590 PR tree-optimization/83640
22591 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
22592 subtracting negative offset from size.
22593 (builtin_access::overlap): Adjust offset bounds of the access to fall
22594 within the size of the object if possible.
22596 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
22598 PR rtl-optimization/83699
22599 * expmed.c (extract_bit_field_1): Restrict the vector usage of
22600 extract_bit_field_as_subreg to cases in which the extracted
22601 value is also a vector.
22603 * lra-constraints.c (process_alt_operands): Test for the equivalence
22604 substitutions when detecting a possible reload cycle.
22606 2018-01-06 Jakub Jelinek <jakub@redhat.com>
22609 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
22610 by default if flag_selective_schedling{,2}. Formatting fixes.
22612 PR rtl-optimization/83682
22613 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
22614 if it has non-VECTOR_MODE element mode.
22615 (vec_duplicate_p): Likewise.
22617 PR middle-end/83694
22618 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
22619 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
22621 2018-01-05 Jakub Jelinek <jakub@redhat.com>
22624 * config/i386/i386-builtin.def
22625 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
22626 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
22627 Require also OPTION_MASK_ISA_AVX512F in addition to
22628 OPTION_MASK_ISA_GFNI.
22629 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
22630 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
22631 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
22632 to OPTION_MASK_ISA_GFNI.
22633 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
22634 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
22635 OPTION_MASK_ISA_AVX512BW.
22636 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
22637 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
22638 addition to OPTION_MASK_ISA_GFNI.
22639 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
22640 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
22641 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
22642 to OPTION_MASK_ISA_GFNI.
22643 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
22644 a requirement for all ISAs rather than any of them with a few
22646 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
22648 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
22649 bitmasks to be enabled with 3 exceptions, instead of requiring any
22650 enabled ISA with lots of exceptions.
22651 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
22652 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
22653 Change avx512bw in isa attribute to avx512f.
22654 * config/i386/sgxintrin.h: Add license boilerplate.
22655 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
22656 to __AVX512F__ and __AVX512VL to __AVX512VL__.
22657 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
22658 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
22660 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
22661 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
22662 temporarily sse2 rather than sse if not enabled already.
22665 * config/i386/sse.md (VI248_VLBW): Rename to ...
22666 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
22667 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
22668 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
22669 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
22670 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
22671 mode iterator instead of VI248_VLBW.
22673 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
22675 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
22676 (record_modified): Skip clobbers; add debug output.
22677 (param_change_prob): Use sreal frequencies.
22679 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
22681 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
22682 punt for user-aligned variables.
22684 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
22686 * tree-chrec.c (chrec_contains_symbols): Return true for
22689 2018-01-05 Sudakshina Das <sudi.das@arm.com>
22692 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
22693 of (x|y) == x for BICS pattern.
22695 2018-01-05 Jakub Jelinek <jakub@redhat.com>
22697 PR tree-optimization/83605
22698 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
22699 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
22702 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
22704 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
22705 * config/epiphany/rtems.h: New file.
22707 2018-01-04 Jakub Jelinek <jakub@redhat.com>
22708 Uros Bizjak <ubizjak@gmail.com>
22711 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
22712 QIreg_operand instead of register_operand predicate.
22713 * config/i386/i386.c (ix86_rop_should_change_byte_p,
22714 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
22715 comments instead of -fmitigate[-_]rop.
22717 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22720 * cgraphunit.c (symbol_table::compile): Switch to text_section
22721 before calling assembly_start debug hook.
22722 * run-rtl-passes.c (run_rtl_passes): Likewise.
22725 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
22727 * tree-vrp.c (extract_range_from_binary_expr_1): Check
22728 range_int_cst_p rather than !symbolic_range_p before calling
22729 extract_range_from_multiplicative_op_1.
22731 2018-01-04 Jeff Law <law@redhat.com>
22733 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
22734 redundant test in assertion.
22736 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
22738 * doc/rtl.texi: Document machine_mode wrapper classes.
22740 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
22742 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
22743 using tree_to_uhwi.
22745 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
22747 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
22748 the VEC_PERM_EXPR fold to fail.
22750 2018-01-04 Jakub Jelinek <jakub@redhat.com>
22753 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
22754 to switched_sections.
22756 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
22759 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
22760 test for d.testing.
22762 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
22765 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
22766 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
22768 2018-01-04 Jakub Jelinek <jakub@redhat.com>
22771 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
22772 is BLKmode and bitpos not zero or mode change is needed.
22774 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
22777 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
22780 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
22783 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
22784 instead of MULT rtx. Update all corresponding splitters.
22785 (*saddl_se): Ditto.
22786 (*ssub<modesuffix>): Ditto.
22787 (*ssubl_se): Ditto.
22788 (*cmp_sadd_di): Update split patterns.
22789 (*cmp_sadd_si): Ditto.
22790 (*cmp_sadd_sidi): Ditto.
22791 (*cmp_ssub_di): Ditto.
22792 (*cmp_ssub_si): Ditto.
22793 (*cmp_ssub_sidi): Ditto.
22794 * config/alpha/predicates.md (const23_operand): New predicate.
22795 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
22796 Look for ASHIFT, not MULT inner operand.
22797 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
22799 2018-01-04 Martin Liska <mliska@suse.cz>
22801 PR gcov-profile/83669
22802 * gcov.c (output_intermediate_file): Add version to intermediate
22804 * doc/gcov.texi: Document new field 'version' in intermediate
22805 file format. Fix location of '-k' option of gcov command.
22807 2018-01-04 Martin Liska <mliska@suse.cz>
22810 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
22812 2018-01-04 Jakub Jelinek <jakub@redhat.com>
22814 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
22816 2018-01-03 Martin Sebor <msebor@redhat.com>
22818 PR tree-optimization/83655
22819 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
22820 checking calls with invalid arguments.
22822 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22824 * tree-vect-stmts.c (vect_get_store_rhs): New function.
22825 (vectorizable_mask_load_store): Delete.
22826 (vectorizable_call): Return false for masked loads and stores.
22827 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
22828 instead of gimple_assign_rhs1.
22829 (vectorizable_load): Handle IFN_MASK_LOAD.
22830 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
22832 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22834 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
22836 (vectorizable_mask_load_store): ...here.
22837 (vectorizable_load): ...and here.
22839 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22841 * tree-vect-stmts.c (vect_build_all_ones_mask)
22842 (vect_build_zero_merge_argument): New functions, split out from...
22843 (vectorizable_load): ...here.
22845 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22847 * tree-vect-stmts.c (vect_check_store_rhs): New function,
22849 (vectorizable_mask_load_store): ...here.
22850 (vectorizable_store): ...and here.
22852 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22854 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
22856 (vectorizable_mask_load_store): ...here.
22858 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22860 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
22861 (vect_model_store_cost): Take a vec_load_store_type instead of a
22863 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
22864 (vect_model_store_cost): Take a vec_load_store_type instead of a
22866 (vectorizable_mask_load_store): Update accordingly.
22867 (vectorizable_store): Likewise.
22868 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
22870 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22872 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
22873 IFN_MASK_LOAD calls here rather than...
22874 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
22876 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22877 Alan Hayward <alan.hayward@arm.com>
22878 David Sherwood <david.sherwood@arm.com>
22880 * expmed.c (extract_bit_field_1): For vector extracts,
22881 fall back to extract_bit_field_as_subreg if vec_extract
22884 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22885 Alan Hayward <alan.hayward@arm.com>
22886 David Sherwood <david.sherwood@arm.com>
22888 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
22889 they are variable or constant sized.
22890 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
22891 slots for constant-sized data.
22893 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22894 Alan Hayward <alan.hayward@arm.com>
22895 David Sherwood <david.sherwood@arm.com>
22897 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
22898 handling COND_EXPRs with boolean comparisons, try to find a better
22899 basis for the mask type than the boolean itself.
22901 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22903 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
22904 is calculated and how it can be overridden.
22905 * genmodes.c (max_bitsize_mode_any_mode): New variable.
22906 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
22908 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
22911 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22912 Alan Hayward <alan.hayward@arm.com>
22913 David Sherwood <david.sherwood@arm.com>
22915 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
22916 Remove the mode argument.
22917 (aarch64_simd_valid_immediate): Remove the mode and inverse
22919 * config/aarch64/iterators.md (bitsize): New iterator.
22920 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
22921 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
22922 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
22923 aarch64_simd_valid_immediate.
22924 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
22925 (aarch64_reg_or_bic_imm): Likewise.
22926 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
22927 with an insn_type enum and msl with a modifier_type enum.
22928 Replace element_width with a scalar_mode. Change the shift
22929 to unsigned int. Add constructors for scalar_float_mode and
22930 scalar_int_mode elements.
22931 (aarch64_vect_float_const_representable_p): Delete.
22932 (aarch64_can_const_movi_rtx_p)
22933 (aarch64_simd_scalar_immediate_valid_for_move)
22934 (aarch64_simd_make_constant): Update call to
22935 aarch64_simd_valid_immediate.
22936 (aarch64_advsimd_valid_immediate_hs): New function.
22937 (aarch64_advsimd_valid_immediate): Likewise.
22938 (aarch64_simd_valid_immediate): Remove mode and inverse
22939 arguments. Rewrite to use the above. Use const_vec_duplicate_p
22940 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
22941 and aarch64_float_const_representable_p on the result.
22942 (aarch64_output_simd_mov_immediate): Remove mode argument.
22943 Update call to aarch64_simd_valid_immediate and use of
22944 simd_immediate_info.
22945 (aarch64_output_scalar_simd_mov_immediate): Update call
22948 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22949 Alan Hayward <alan.hayward@arm.com>
22950 David Sherwood <david.sherwood@arm.com>
22952 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
22953 (mode_nunits): Likewise CONST_MODE_NUNITS.
22954 * machmode.def (ADJUST_NUNITS): Document.
22955 * genmodes.c (mode_data::need_nunits_adj): New field.
22956 (blank_mode): Update accordingly.
22957 (adj_nunits): New variable.
22958 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
22960 (emit_mode_size_inline): Set need_bytesize_adj for all modes
22961 listed in adj_nunits.
22962 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
22963 listed in adj_nunits. Don't emit case statements for such modes.
22964 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
22965 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
22966 nothing if adj_nunits is nonnull.
22967 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
22968 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
22969 (emit_mode_fbit): Update use of print_maybe_const_decl.
22970 (emit_move_size): Likewise. Treat the array as non-const
22972 (emit_mode_adjustments): Handle adj_nunits.
22974 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22976 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
22977 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
22978 (VECTOR_MODES): Use it.
22979 (make_vector_modes): Take the prefix as an argument.
22981 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
22982 Alan Hayward <alan.hayward@arm.com>
22983 David Sherwood <david.sherwood@arm.com>
22985 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
22986 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
22987 for MODE_VECTOR_BOOL.
22988 * machmode.def (VECTOR_BOOL_MODE): Document.
22989 * genmodes.c (VECTOR_BOOL_MODE): New macro.
22990 (make_vector_bool_mode): New function.
22991 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
22993 * lto-streamer-in.c (lto_input_mode_table): Likewise.
22994 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
22996 * stor-layout.c (int_mode_for_mode): Likewise.
22997 * tree.c (build_vector_type_for_mode): Likewise.
22998 * varasm.c (output_constant_pool_2): Likewise.
22999 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
23000 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
23001 for MODE_VECTOR_BOOL.
23002 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
23003 of mode class checks.
23004 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
23005 instead of a list of mode class checks.
23006 (expand_vector_scalar_condition): Likewise.
23007 (type_for_widest_vector_mode): Handle BImode as an inner mode.
23009 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23010 Alan Hayward <alan.hayward@arm.com>
23011 David Sherwood <david.sherwood@arm.com>
23013 * machmode.h (mode_size): Change from unsigned short to
23015 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
23016 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
23017 or if measurement_type is not polynomial.
23018 (fixed_size_mode::includes_p): Check for constant-sized modes.
23019 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
23020 return a poly_uint16 rather than an unsigned short.
23021 (emit_mode_size): Change the type of mode_size from unsigned short
23022 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
23023 (emit_mode_adjustments): Cope with polynomial vector sizes.
23024 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
23026 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
23028 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
23029 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
23030 * caller-save.c (setup_save_areas): Likewise.
23031 (replace_reg_with_saved_mem): Likewise.
23032 * calls.c (emit_library_call_value_1): Likewise.
23033 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
23034 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
23035 (gen_lowpart_for_combine): Likewise.
23036 * convert.c (convert_to_integer_1): Likewise.
23037 * cse.c (equiv_constant, cse_insn): Likewise.
23038 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
23039 (cselib_subst_to_values): Likewise.
23040 * dce.c (word_dce_process_block): Likewise.
23041 * df-problems.c (df_word_lr_mark_ref): Likewise.
23042 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
23043 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
23044 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
23045 (rtl_for_decl_location): Likewise.
23046 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
23047 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
23048 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
23049 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
23050 (expand_expr_real_1): Likewise.
23051 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
23052 (pad_below): Likewise.
23053 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
23054 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
23055 * ira.c (get_subreg_tracking_sizes): Likewise.
23056 * ira-build.c (ira_create_allocno_objects): Likewise.
23057 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
23058 (ira_sort_regnos_for_alter_reg): Likewise.
23059 * ira-costs.c (record_operand_costs): Likewise.
23060 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
23061 (resolve_simple_move): Likewise.
23062 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
23063 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
23064 (lra_constraints): Likewise.
23065 (CONST_POOL_OK_P): Reject variable-sized modes.
23066 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
23067 (add_pseudo_to_slot, lra_spill): Likewise.
23068 * omp-low.c (omp_clause_aligned_alignment): Likewise.
23069 * optabs-query.c (get_best_extraction_insn): Likewise.
23070 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
23071 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
23072 (expand_mult_highpart, valid_multiword_target_p): Likewise.
23073 * recog.c (offsettable_address_addr_space_p): Likewise.
23074 * regcprop.c (maybe_mode_change): Likewise.
23075 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
23076 * regrename.c (build_def_use): Likewise.
23077 * regstat.c (dump_reg_info): Likewise.
23078 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
23079 (find_reloads, find_reloads_subreg_address): Likewise.
23080 * reload1.c (eliminate_regs_1): Likewise.
23081 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
23082 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
23083 (simplify_binary_operation_1, simplify_subreg): Likewise.
23084 * targhooks.c (default_function_arg_padding): Likewise.
23085 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
23086 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
23087 (verify_gimple_assign_ternary): Likewise.
23088 * tree-inline.c (estimate_move_cost): Likewise.
23089 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
23090 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
23091 (get_address_cost_ainc): Likewise.
23092 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
23093 (vect_supportable_dr_alignment): Likewise.
23094 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
23095 (vectorizable_reduction): Likewise.
23096 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
23097 (vectorizable_operation, vectorizable_load): Likewise.
23098 * tree.c (build_same_sized_truth_vector_type): Likewise.
23099 * valtrack.c (cleanup_auto_inc_dec): Likewise.
23100 * var-tracking.c (emit_note_insn_var_location): Likewise.
23101 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
23102 (ADDR_VEC_ALIGN): Likewise.
23104 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23105 Alan Hayward <alan.hayward@arm.com>
23106 David Sherwood <david.sherwood@arm.com>
23108 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
23110 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
23111 or if measurement_type is polynomial.
23112 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
23113 * combine.c (make_extraction): Likewise.
23114 * dse.c (find_shift_sequence): Likewise.
23115 * dwarf2out.c (mem_loc_descriptor): Likewise.
23116 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
23117 (extract_bit_field, extract_low_bits): Likewise.
23118 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
23119 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
23120 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
23121 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
23122 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
23123 * reload.c (find_reloads): Likewise.
23124 * reload1.c (alter_reg): Likewise.
23125 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
23126 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
23127 * tree-if-conv.c (predicate_mem_writes): Likewise.
23128 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
23129 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
23130 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
23131 * valtrack.c (dead_debug_insert_temp): Likewise.
23132 * varasm.c (mergeable_constant_section): Likewise.
23133 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
23135 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23136 Alan Hayward <alan.hayward@arm.com>
23137 David Sherwood <david.sherwood@arm.com>
23139 * expr.c (expand_assignment): Cope with polynomial mode sizes
23140 when assigning to a CONCAT.
23142 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23143 Alan Hayward <alan.hayward@arm.com>
23144 David Sherwood <david.sherwood@arm.com>
23146 * machmode.h (mode_precision): Change from unsigned short to
23148 (mode_to_precision): Return a poly_uint16 rather than an unsigned
23150 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
23151 or if measurement_type is not polynomial.
23152 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
23153 in which the mode is already known to be a scalar_int_mode.
23154 * genmodes.c (emit_mode_precision): Change the type of mode_precision
23155 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
23157 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
23158 for GET_MODE_PRECISION.
23159 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
23160 for GET_MODE_PRECISION.
23161 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
23163 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
23164 (expand_field_assignment, make_extraction): Likewise.
23165 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
23166 (get_last_value): Likewise.
23167 * convert.c (convert_to_integer_1): Likewise.
23168 * cse.c (cse_insn): Likewise.
23169 * expr.c (expand_expr_real_1): Likewise.
23170 * lra-constraints.c (simplify_operand_subreg): Likewise.
23171 * optabs-query.c (can_atomic_load_p): Likewise.
23172 * optabs.c (expand_atomic_load): Likewise.
23173 (expand_atomic_store): Likewise.
23174 * ree.c (combine_reaching_defs): Likewise.
23175 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
23176 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
23177 * tree.h (type_has_mode_precision_p): Likewise.
23178 * ubsan.c (instrument_si_overflow): Likewise.
23180 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23181 Alan Hayward <alan.hayward@arm.com>
23182 David Sherwood <david.sherwood@arm.com>
23184 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
23185 polynomial numbers of units.
23186 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
23187 (valid_vector_subparts_p): New function.
23188 (build_vector_type): Remove temporary shim and take the number
23189 of units as a poly_uint64 rather than an int.
23190 (build_opaque_vector_type): Take the number of units as a
23191 poly_uint64 rather than an int.
23192 * tree.c (build_vector_from_ctor): Handle polynomial
23193 TYPE_VECTOR_SUBPARTS.
23194 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
23195 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
23196 (build_vector_from_val): If the number of units is variable,
23197 use build_vec_duplicate_cst for constant operands and
23198 VEC_DUPLICATE_EXPR otherwise.
23199 (make_vector_type): Remove temporary is_constant ().
23200 (build_vector_type, build_opaque_vector_type): Take the number of
23201 units as a poly_uint64 rather than an int.
23202 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
23204 * cfgexpand.c (expand_debug_expr): Likewise.
23205 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
23206 (store_constructor, expand_expr_real_1): Likewise.
23207 (const_scalar_mask_from_tree): Likewise.
23208 * fold-const-call.c (fold_const_reduction): Likewise.
23209 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
23210 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
23211 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
23212 (fold_relational_const): Likewise.
23213 (native_interpret_vector): Likewise. Change the size from an
23214 int to an unsigned int.
23215 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
23216 TYPE_VECTOR_SUBPARTS.
23217 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
23218 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
23219 duplicating a non-constant operand into a variable-length vector.
23220 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
23221 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
23222 * ipa-icf.c (sem_variable::equals): Likewise.
23223 * match.pd: Likewise.
23224 * omp-simd-clone.c (simd_clone_subparts): Likewise.
23225 * print-tree.c (print_node): Likewise.
23226 * stor-layout.c (layout_type): Likewise.
23227 * targhooks.c (default_builtin_vectorization_cost): Likewise.
23228 * tree-cfg.c (verify_gimple_comparison): Likewise.
23229 (verify_gimple_assign_binary): Likewise.
23230 (verify_gimple_assign_ternary): Likewise.
23231 (verify_gimple_assign_single): Likewise.
23232 * tree-pretty-print.c (dump_generic_node): Likewise.
23233 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
23234 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
23235 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
23236 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
23237 (vect_shift_permute_load_chain): Likewise.
23238 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
23239 (expand_vector_condition, optimize_vector_constructor): Likewise.
23240 (lower_vec_perm, get_compute_type): Likewise.
23241 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
23242 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
23243 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
23244 (vect_recog_mask_conversion_pattern): Likewise.
23245 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
23246 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
23247 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
23248 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
23249 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
23250 (vectorizable_shift, vectorizable_operation, vectorizable_store)
23251 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
23252 (supportable_widening_operation): Likewise.
23253 (supportable_narrowing_operation): Likewise.
23254 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
23256 * varasm.c (output_constant): Likewise.
23258 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23259 Alan Hayward <alan.hayward@arm.com>
23260 David Sherwood <david.sherwood@arm.com>
23262 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
23263 so that both the length == 3 and length != 3 cases set up their
23264 own permute vectors. Add comments explaining why we know the
23265 number of elements is constant.
23266 (vect_permute_load_chain): Likewise.
23268 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23269 Alan Hayward <alan.hayward@arm.com>
23270 David Sherwood <david.sherwood@arm.com>
23272 * machmode.h (mode_nunits): Change from unsigned char to
23274 (ONLY_FIXED_SIZE_MODES): New macro.
23275 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
23276 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
23277 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
23279 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
23280 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
23281 or if measurement_type is not polynomial.
23282 * genmodes.c (ZERO_COEFFS): New macro.
23283 (emit_mode_nunits_inline): Make mode_nunits_inline return a
23285 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
23286 Use ZERO_COEFFS when emitting initializers.
23287 * data-streamer.h (bp_pack_poly_value): New function.
23288 (bp_unpack_poly_value): Likewise.
23289 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
23290 for GET_MODE_NUNITS.
23291 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
23292 for GET_MODE_NUNITS.
23293 * tree.c (make_vector_type): Remove temporary shim and make
23294 the real function take the number of units as a poly_uint64
23295 rather than an int.
23296 (build_vector_type_for_mode): Handle polynomial nunits.
23297 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
23298 * emit-rtl.c (const_vec_series_p_1): Likewise.
23299 (gen_rtx_CONST_VECTOR): Likewise.
23300 * fold-const.c (test_vec_duplicate_folding): Likewise.
23301 * genrecog.c (validate_pattern): Likewise.
23302 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
23303 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
23304 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
23305 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
23306 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
23307 * rtlanal.c (subreg_get_info): Likewise.
23308 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
23309 (vect_grouped_load_supported): Likewise.
23310 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
23311 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
23312 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
23313 (simplify_const_unary_operation, simplify_binary_operation_1)
23314 (simplify_const_binary_operation, simplify_ternary_operation)
23315 (test_vector_ops_duplicate, test_vector_ops): Likewise.
23316 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
23317 instead of CONST_VECTOR_NUNITS.
23318 * varasm.c (output_constant_pool_2): Likewise.
23319 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
23320 explicit-encoded elements in the XVEC for variable-length vectors.
23322 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23324 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
23326 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23327 Alan Hayward <alan.hayward@arm.com>
23328 David Sherwood <david.sherwood@arm.com>
23330 * coretypes.h (fixed_size_mode): Declare.
23331 (fixed_size_mode_pod): New typedef.
23332 * builtins.h (target_builtins::x_apply_args_mode)
23333 (target_builtins::x_apply_result_mode): Change type to
23334 fixed_size_mode_pod.
23335 * builtins.c (apply_args_size, apply_result_size, result_vector)
23336 (expand_builtin_apply_args_1, expand_builtin_apply)
23337 (expand_builtin_return): Update accordingly.
23339 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23341 * cse.c (hash_rtx_cb): Hash only the encoded elements.
23342 * cselib.c (cselib_hash_rtx): Likewise.
23343 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
23344 CONST_VECTOR encoding.
23346 2018-01-03 Jakub Jelinek <jakub@redhat.com>
23347 Jeff Law <law@redhat.com>
23350 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
23351 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
23352 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
23353 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
23356 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
23357 explicitly probe *sp in a noreturn function if there were any callee
23358 register saves or frame pointer is needed.
23360 2018-01-03 Jakub Jelinek <jakub@redhat.com>
23363 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
23364 BLKmode for ternary, binary or unary expressions.
23367 * var-tracking.c (delete_vta_debug_insn): New inline function.
23368 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
23369 insns from get_insns () to NULL instead of each bb separately.
23370 Use delete_vta_debug_insn. No longer static.
23371 (vt_debug_insns_local, variable_tracking_main_1): Adjust
23372 delete_vta_debug_insns callers.
23373 * rtl.h (delete_vta_debug_insns): Declare.
23374 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
23375 instead of variable_tracking_main.
23377 2018-01-03 Martin Sebor <msebor@redhat.com>
23379 PR tree-optimization/83603
23380 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
23381 arguments past the endof the argument list in functions declared
23382 without a prototype.
23383 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
23384 Avoid checking when arguments are null.
23386 2018-01-03 Martin Sebor <msebor@redhat.com>
23389 * doc/extend.texi (attribute const): Fix a typo.
23390 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
23391 issuing -Wsuggest-attribute for void functions.
23393 2018-01-03 Martin Sebor <msebor@redhat.com>
23395 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
23396 offset_int::from instead of wide_int::to_shwi.
23397 (maybe_diag_overlap): Remove assertion.
23398 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
23399 * gimple-ssa-sprintf.c (format_directive): Same.
23400 (parse_directive): Same.
23401 (sprintf_dom_walker::compute_format_length): Same.
23402 (try_substitute_return_value): Same.
23404 2018-01-03 Jeff Law <law@redhat.com>
23406 PR middle-end/83654
23407 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
23408 non-constant residual for zero at runtime and avoid probing in
23409 that case. Reorganize code for trailing problem to mirror handling
23412 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23414 PR tree-optimization/83501
23415 * tree-ssa-strlen.c (get_string_cst): New.
23416 (handle_char_store): Call get_string_cst.
23418 2018-01-03 Martin Liska <mliska@suse.cz>
23420 PR tree-optimization/83593
23421 * tree-ssa-strlen.c: Include tree-cfg.h.
23422 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
23423 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
23424 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
23426 (strlen_dom_walker::before_dom_children): Call
23427 gimple_purge_dead_eh_edges. Dump tranformation with details
23429 (strlen_dom_walker::before_dom_children): Update call by adding
23430 new argument cleanup_eh.
23431 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
23433 2018-01-03 Martin Liska <mliska@suse.cz>
23436 * cif-code.def (VARIADIC_THUNK): New enum value.
23437 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
23440 2018-01-03 Jan Beulich <jbeulich@suse.com>
23442 * sse.md (mov<mode>_internal): Tighten condition for when to use
23443 vmovdqu<ssescalarsize> for TI and OI modes.
23445 2018-01-03 Jakub Jelinek <jakub@redhat.com>
23447 Update copyright years.
23449 2018-01-03 Martin Liska <mliska@suse.cz>
23452 * ipa-visibility.c (function_and_variable_visibility): Skip
23453 functions with noipa attribure.
23455 2018-01-03 Jakub Jelinek <jakub@redhat.com>
23457 * gcc.c (process_command): Update copyright notice dates.
23458 * gcov-dump.c (print_version): Ditto.
23459 * gcov.c (print_version): Ditto.
23460 * gcov-tool.c (print_version): Ditto.
23461 * gengtype.c (create_file): Ditto.
23462 * doc/cpp.texi: Bump @copying's copyright year.
23463 * doc/cppinternals.texi: Ditto.
23464 * doc/gcc.texi: Ditto.
23465 * doc/gccint.texi: Ditto.
23466 * doc/gcov.texi: Ditto.
23467 * doc/install.texi: Ditto.
23468 * doc/invoke.texi: Ditto.
23470 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23472 * vector-builder.h (vector_builder::m_full_nelts): Change from
23473 unsigned int to poly_uint64.
23474 (vector_builder::full_nelts): Update prototype accordingly.
23475 (vector_builder::new_vector): Likewise.
23476 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
23477 (vector_builder::operator ==): Likewise.
23478 (vector_builder::finalize): Likewise.
23479 * int-vector-builder.h (int_vector_builder::int_vector_builder):
23480 Take the number of elements as a poly_uint64 rather than an
23482 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
23483 from unsigned int to poly_uint64.
23484 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
23485 (vec_perm_indices::new_vector): Likewise.
23486 (vec_perm_indices::length): Likewise.
23487 (vec_perm_indices::nelts_per_input): Likewise.
23488 (vec_perm_indices::input_nelts): Likewise.
23489 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
23490 number of elements per input as a poly_uint64 rather than an
23491 unsigned int. Use the original encoding for variable-length
23492 vectors, rather than clamping each individual element.
23493 For the second and subsequent elements in each pattern,
23494 clamp the step and base before clamping their sum.
23495 (vec_perm_indices::series_p): Handle polynomial element counts.
23496 (vec_perm_indices::all_in_range_p): Likewise.
23497 (vec_perm_indices_to_tree): Likewise.
23498 (vec_perm_indices_to_rtx): Likewise.
23499 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
23500 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
23501 (tree_vector_builder::new_binary_operation): Handle polynomial
23502 element counts. Return false if we need to know the number
23503 of elements at compile time.
23504 * fold-const.c (fold_vec_perm): Punt if the number of elements
23505 isn't known at compile time.
23507 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23509 * vec-perm-indices.h (vec_perm_builder): Change element type
23510 from HOST_WIDE_INT to poly_int64.
23511 (vec_perm_indices::element_type): Update accordingly.
23512 (vec_perm_indices::clamp): Handle polynomial element_types.
23513 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
23514 (vec_perm_indices::all_in_range_p): Likewise.
23515 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
23517 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
23518 polynomial vec_perm_indices element types.
23519 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
23520 * fold-const.c (fold_vec_perm): Likewise.
23521 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
23522 * tree-vect-generic.c (lower_vec_perm): Likewise.
23523 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
23524 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
23525 element type to HOST_WIDE_INT.
23527 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23528 Alan Hayward <alan.hayward@arm.com>
23529 David Sherwood <david.sherwood@arm.com>
23531 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
23532 rather than an int. Use plus_constant.
23533 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
23534 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
23536 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23537 Alan Hayward <alan.hayward@arm.com>
23538 David Sherwood <david.sherwood@arm.com>
23540 * calls.c (emit_call_1, expand_call): Change struct_value_size from
23541 a HOST_WIDE_INT to a poly_int64.
23543 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23544 Alan Hayward <alan.hayward@arm.com>
23545 David Sherwood <david.sherwood@arm.com>
23547 * calls.c (load_register_parameters): Cope with polynomial
23548 mode sizes. Require a constant size for BLKmode parameters
23549 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
23550 forces a parameter to be padded at the lsb end in order to
23551 fill a complete number of words, require the parameter size
23552 to be ordered wrt UNITS_PER_WORD.
23554 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23555 Alan Hayward <alan.hayward@arm.com>
23556 David Sherwood <david.sherwood@arm.com>
23558 * reload1.c (spill_stack_slot_width): Change element type
23559 from unsigned int to poly_uint64_pod.
23560 (alter_reg): Treat mode sizes as polynomial.
23562 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23563 Alan Hayward <alan.hayward@arm.com>
23564 David Sherwood <david.sherwood@arm.com>
23566 * reload.c (complex_word_subreg_p): New function.
23567 (reload_inner_reg_of_subreg, push_reload): Use it.
23569 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23570 Alan Hayward <alan.hayward@arm.com>
23571 David Sherwood <david.sherwood@arm.com>
23573 * lra-constraints.c (process_alt_operands): Reject matched
23574 operands whose sizes aren't ordered.
23575 (match_reload): Refer to this check here.
23577 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23578 Alan Hayward <alan.hayward@arm.com>
23579 David Sherwood <david.sherwood@arm.com>
23581 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
23582 that the mode size is in the set {1, 2, 4, 8, 16}.
23584 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23585 Alan Hayward <alan.hayward@arm.com>
23586 David Sherwood <david.sherwood@arm.com>
23588 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
23589 Use plus_constant instead of gen_rtx_PLUS.
23591 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23592 Alan Hayward <alan.hayward@arm.com>
23593 David Sherwood <david.sherwood@arm.com>
23595 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
23596 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
23597 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
23598 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
23599 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
23600 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
23601 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
23602 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
23603 * config/i386/i386.c (ix86_push_rounding): ...this new function.
23604 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
23606 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
23607 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
23608 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
23609 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
23610 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
23611 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
23612 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
23613 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
23614 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
23615 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
23617 * expr.c (emit_move_resolve_push): Treat the input and result
23618 of PUSH_ROUNDING as a poly_int64.
23619 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
23620 (emit_push_insn): Likewise.
23621 * lra-eliminations.c (mark_not_eliminable): Likewise.
23622 * recog.c (push_operand): Likewise.
23623 * reload1.c (elimination_effects): Likewise.
23624 * rtlanal.c (nonzero_bits1): Likewise.
23625 * calls.c (store_one_arg): Likewise. Require the padding to be
23626 known at compile time.
23628 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23629 Alan Hayward <alan.hayward@arm.com>
23630 David Sherwood <david.sherwood@arm.com>
23632 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
23633 Use plus_constant instead of gen_rtx_PLUS.
23635 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23636 Alan Hayward <alan.hayward@arm.com>
23637 David Sherwood <david.sherwood@arm.com>
23639 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
23640 rather than an int.
23642 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23643 Alan Hayward <alan.hayward@arm.com>
23644 David Sherwood <david.sherwood@arm.com>
23646 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
23647 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
23648 via stack temporaries. Treat the mode size as polynomial too.
23650 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23651 Alan Hayward <alan.hayward@arm.com>
23652 David Sherwood <david.sherwood@arm.com>
23654 * expr.c (expand_expr_real_2): When handling conversions involving
23655 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
23656 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
23657 as a poly_uint64 too.
23659 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23660 Alan Hayward <alan.hayward@arm.com>
23661 David Sherwood <david.sherwood@arm.com>
23663 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
23665 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23666 Alan Hayward <alan.hayward@arm.com>
23667 David Sherwood <david.sherwood@arm.com>
23669 * combine.c (can_change_dest_mode): Handle polynomial
23670 REGMODE_NATURAL_SIZE.
23671 * expmed.c (store_bit_field_1): Likewise.
23672 * expr.c (store_constructor): Likewise.
23673 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
23674 and polynomial REGMODE_NATURAL_SIZE.
23675 (gen_lowpart_common): Likewise.
23676 * reginfo.c (record_subregs_of_mode): Likewise.
23677 * rtlanal.c (read_modify_subreg_p): Likewise.
23679 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23680 Alan Hayward <alan.hayward@arm.com>
23681 David Sherwood <david.sherwood@arm.com>
23683 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
23684 numbers of elements.
23686 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23687 Alan Hayward <alan.hayward@arm.com>
23688 David Sherwood <david.sherwood@arm.com>
23690 * match.pd: Cope with polynomial numbers of vector elements.
23692 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23693 Alan Hayward <alan.hayward@arm.com>
23694 David Sherwood <david.sherwood@arm.com>
23696 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
23697 in a POINTER_PLUS_EXPR.
23699 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23700 Alan Hayward <alan.hayward@arm.com>
23701 David Sherwood <david.sherwood@arm.com>
23703 * omp-simd-clone.c (simd_clone_subparts): New function.
23704 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
23705 (ipa_simd_modify_function_body): Likewise.
23707 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23708 Alan Hayward <alan.hayward@arm.com>
23709 David Sherwood <david.sherwood@arm.com>
23711 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
23712 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
23713 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
23714 (expand_vector_condition, vector_element): Likewise.
23715 (subparts_gt): New function.
23716 (get_compute_type): Use subparts_gt.
23717 (count_type_subparts): Delete.
23718 (expand_vector_operations_1): Use subparts_gt instead of
23719 count_type_subparts.
23721 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23722 Alan Hayward <alan.hayward@arm.com>
23723 David Sherwood <david.sherwood@arm.com>
23725 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
23726 (vect_compile_time_alias): ...this new function. Do the calculation
23727 on poly_ints rather than trees.
23728 (vect_prune_runtime_alias_test_list): Update call accordingly.
23730 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23731 Alan Hayward <alan.hayward@arm.com>
23732 David Sherwood <david.sherwood@arm.com>
23734 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
23736 (vect_schedule_slp_instance): Likewise.
23738 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23739 Alan Hayward <alan.hayward@arm.com>
23740 David Sherwood <david.sherwood@arm.com>
23742 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
23743 constant and extern definitions for variable-length vectors.
23744 (vect_get_constant_vectors): Note that the number of units
23745 is known to be constant.
23747 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23748 Alan Hayward <alan.hayward@arm.com>
23749 David Sherwood <david.sherwood@arm.com>
23751 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
23752 of units as polynomial. Choose between WIDE and NARROW based
23755 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23756 Alan Hayward <alan.hayward@arm.com>
23757 David Sherwood <david.sherwood@arm.com>
23759 * tree-vect-stmts.c (simd_clone_subparts): New function.
23760 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
23762 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23763 Alan Hayward <alan.hayward@arm.com>
23764 David Sherwood <david.sherwood@arm.com>
23766 * tree-vect-stmts.c (vectorizable_call): Treat the number of
23767 vectors as polynomial. Use build_index_vector for
23768 IFN_GOMP_SIMD_LANE.
23770 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23771 Alan Hayward <alan.hayward@arm.com>
23772 David Sherwood <david.sherwood@arm.com>
23774 * tree-vect-stmts.c (get_load_store_type): Treat the number of
23775 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
23776 for variable-length vectors.
23777 (vectorizable_mask_load_store): Treat the number of units as
23778 polynomial, asserting that it is constant if the condition has
23779 already been enforced.
23780 (vectorizable_store, vectorizable_load): Likewise.
23782 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23783 Alan Hayward <alan.hayward@arm.com>
23784 David Sherwood <david.sherwood@arm.com>
23786 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
23787 of units as polynomial. Punt if we can't tell at compile time
23788 which vector contains the final result.
23790 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23791 Alan Hayward <alan.hayward@arm.com>
23792 David Sherwood <david.sherwood@arm.com>
23794 * tree-vect-loop.c (vectorizable_induction): Treat the number
23795 of units as polynomial. Punt on SLP inductions. Use an integer
23796 VEC_SERIES_EXPR for variable-length integer reductions. Use a
23797 cast of such a series for variable-length floating-point
23800 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23801 Alan Hayward <alan.hayward@arm.com>
23802 David Sherwood <david.sherwood@arm.com>
23804 * tree.h (build_index_vector): Declare.
23805 * tree.c (build_index_vector): New function.
23806 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
23807 of units as polynomial, forcibly converting it to a constant if
23808 vectorizable_reduction has already enforced the condition.
23809 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
23810 to create a {1,2,3,...} vector.
23811 (vectorizable_reduction): Treat the number of units as polynomial.
23812 Choose vectype_in based on the largest scalar element size rather
23813 than the smallest number of units. Enforce the restrictions
23816 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23817 Alan Hayward <alan.hayward@arm.com>
23818 David Sherwood <david.sherwood@arm.com>
23820 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
23821 number of units as polynomial.
23823 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23824 Alan Hayward <alan.hayward@arm.com>
23825 David Sherwood <david.sherwood@arm.com>
23827 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
23828 * target.def (autovectorize_vector_sizes): Return the vector sizes
23829 by pointer, using vector_sizes rather than a bitmask.
23830 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
23831 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
23832 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
23834 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
23835 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
23836 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
23837 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
23838 * omp-general.c (omp_max_vf): Likewise.
23839 * omp-low.c (omp_clause_aligned_alignment): Likewise.
23840 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
23841 * tree-vect-loop.c (vect_analyze_loop): Likewise.
23842 * tree-vect-slp.c (vect_slp_bb): Likewise.
23843 * doc/tm.texi: Regenerate.
23844 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
23846 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
23847 the vector size as a poly_uint64 rather than an unsigned int.
23848 (current_vector_size): Change from an unsigned int to a poly_uint64.
23849 (get_vectype_for_scalar_type): Update accordingly.
23850 * tree.h (build_truth_vector_type): Take the size and number of
23851 units as a poly_uint64 rather than an unsigned int.
23852 (build_vector_type): Add a temporary overload that takes
23853 the number of units as a poly_uint64 rather than an unsigned int.
23854 * tree.c (make_vector_type): Likewise.
23855 (build_truth_vector_type): Take the number of units as a poly_uint64
23856 rather than an unsigned int.
23858 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23859 Alan Hayward <alan.hayward@arm.com>
23860 David Sherwood <david.sherwood@arm.com>
23862 * target.def (get_mask_mode): Take the number of units and length
23863 as poly_uint64s rather than unsigned ints.
23864 * targhooks.h (default_get_mask_mode): Update accordingly.
23865 * targhooks.c (default_get_mask_mode): Likewise.
23866 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
23867 * doc/tm.texi: Regenerate.
23869 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23870 Alan Hayward <alan.hayward@arm.com>
23871 David Sherwood <david.sherwood@arm.com>
23873 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
23874 * omp-general.c (omp_max_vf): Likewise.
23875 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
23876 (expand_omp_simd): Handle polynomial safelen.
23877 * omp-low.c (omplow_simd_context): Add a default constructor.
23878 (omplow_simd_context::max_vf): Change from int to poly_uint64.
23879 (lower_rec_simd_input_clauses): Update accordingly.
23880 (lower_rec_input_clauses): Likewise.
23882 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23883 Alan Hayward <alan.hayward@arm.com>
23884 David Sherwood <david.sherwood@arm.com>
23886 * tree-vectorizer.h (vect_nunits_for_cost): New function.
23887 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
23888 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
23889 (vect_analyze_slp_cost): Likewise.
23890 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
23891 (vect_model_load_cost): Likewise.
23893 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23894 Alan Hayward <alan.hayward@arm.com>
23895 David Sherwood <david.sherwood@arm.com>
23897 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
23898 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
23899 from an unsigned int * to a poly_uint64_pod *.
23900 (calculate_unrolling_factor): New function.
23901 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
23903 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23904 Alan Hayward <alan.hayward@arm.com>
23905 David Sherwood <david.sherwood@arm.com>
23907 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
23908 from an unsigned int to a poly_uint64.
23909 (_loop_vec_info::slp_unrolling_factor): Likewise.
23910 (_loop_vec_info::vectorization_factor): Change from an int
23912 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
23913 (vect_get_num_vectors): New function.
23914 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
23915 (vect_get_num_copies): Use vect_get_num_vectors.
23916 (vect_analyze_data_ref_dependences): Change max_vf from an int *
23917 to an unsigned int *.
23918 (vect_analyze_data_refs): Change min_vf from an int * to a
23920 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
23921 than an unsigned HOST_WIDE_INT.
23922 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
23923 (vect_analyze_data_ref_dependence): Change max_vf from an int *
23924 to an unsigned int *.
23925 (vect_analyze_data_ref_dependences): Likewise.
23926 (vect_compute_data_ref_alignment): Handle polynomial vf.
23927 (vect_enhance_data_refs_alignment): Likewise.
23928 (vect_prune_runtime_alias_test_list): Likewise.
23929 (vect_shift_permute_load_chain): Likewise.
23930 (vect_supportable_dr_alignment): Likewise.
23931 (dependence_distance_ge_vf): Take the vectorization factor as a
23932 poly_uint64 rather than an unsigned HOST_WIDE_INT.
23933 (vect_analyze_data_refs): Change min_vf from an int * to a
23935 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
23936 vfm1 as a poly_uint64 rather than an int. Make the same change
23937 for the returned bound_scalar.
23938 (vect_gen_vector_loop_niters): Handle polynomial vf.
23939 (vect_do_peeling): Likewise. Update call to
23940 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
23941 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
23943 * tree-vect-loop.c (vect_determine_vectorization_factor)
23944 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
23945 (vect_get_known_peeling_cost): Likewise.
23946 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
23947 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
23948 (vect_transform_loop): Likewise. Use the lowest possible VF when
23949 updating the upper bounds of the loop.
23950 (vect_min_worthwhile_factor): Make static. Return an unsigned int
23951 rather than an int.
23952 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
23953 polynomial unroll factors.
23954 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
23955 (vect_make_slp_decision): Likewise.
23956 (vect_supported_load_permutation_p): Likewise, and polynomial
23958 (vect_analyze_slp_cost): Handle polynomial vf.
23959 (vect_slp_analyze_node_operations): Likewise.
23960 (vect_slp_analyze_bb_1): Likewise.
23961 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
23962 than an unsigned HOST_WIDE_INT.
23963 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
23964 (vectorizable_load): Handle polynomial vf.
23965 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
23967 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
23969 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23970 Alan Hayward <alan.hayward@arm.com>
23971 David Sherwood <david.sherwood@arm.com>
23973 * match.pd: Handle bit operations involving three constants
23974 and try to fold one pair.
23976 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23978 * tree-vect-loop-manip.c: Include gimple-fold.h.
23979 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
23980 niters_maybe_zero parameters. Handle other cases besides a step of 1.
23981 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
23982 Add a path that uses a step of VF instead of 1, but disable it
23984 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
23985 and niters_no_overflow parameters. Update calls to
23986 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
23987 Create a new SSA name if the latter choses to use a ste other
23988 than zero, and return it via niters_vector_mult_vf_var.
23989 * tree-vect-loop.c (vect_transform_loop): Update calls to
23990 vect_do_peeling, vect_gen_vector_loop_niters and
23991 slpeel_make_loop_iterate_ntimes.
23992 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
23993 (vect_gen_vector_loop_niters): Update declarations after above changes.
23995 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
23997 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
23998 128-bit round to integer instructions.
23999 (ceil<mode>2): Likewise.
24000 (btrunc<mode>2): Likewise.
24001 (round<mode>2): Likewise.
24003 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24005 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
24006 unaligned VSX load/store on P8/P9.
24007 (expand_block_clear): Allow the use of unaligned VSX
24008 load/store on P8/P9.
24010 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24012 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
24014 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
24015 swap associated with both a load and a store.
24017 2018-01-02 Andrew Waterman <andrew@sifive.com>
24019 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
24020 * config/riscv/riscv.md (clear_cache): Use it.
24022 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
24024 * web.c: Remove out-of-date comment.
24026 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
24028 * expr.c (fixup_args_size_notes): Check that any existing
24029 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
24030 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
24031 (emit_single_push_insn): ...here.
24033 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
24035 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
24036 (const_vector_encoded_nelts): New function.
24037 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
24038 (const_vector_int_elt, const_vector_elt): Declare.
24039 * emit-rtl.c (const_vector_int_elt_1): New function.
24040 (const_vector_elt): Likewise.
24041 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
24042 of CONST_VECTOR_ELT.
24044 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
24046 * expr.c: Include rtx-vector-builder.h.
24047 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
24048 directly on the tree encoding.
24049 (const_vector_from_tree): Likewise.
24050 * optabs.c: Include rtx-vector-builder.h.
24051 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
24052 sequence of "u" values.
24053 * vec-perm-indices.c: Include rtx-vector-builder.h.
24054 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
24055 directly on the vec_perm_indices encoding.
24057 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
24059 * doc/rtl.texi (const_vector): Describe new encoding scheme.
24060 * Makefile.in (OBJS): Add rtx-vector-builder.o.
24061 * rtx-vector-builder.h: New file.
24062 * rtx-vector-builder.c: Likewise.
24063 * rtl.h (rtx_def::u2): Add a const_vector field.
24064 (CONST_VECTOR_NPATTERNS): New macro.
24065 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
24066 (CONST_VECTOR_DUPLICATE_P): Likewise.
24067 (CONST_VECTOR_STEPPED_P): Likewise.
24068 (CONST_VECTOR_ENCODED_ELT): Likewise.
24069 (const_vec_duplicate_p): Check for a duplicated vector encoding.
24070 (unwrap_const_vec_duplicate): Likewise.
24071 (const_vec_series_p): Check for a non-duplicated vector encoding.
24072 Say that the function only returns true for integer vectors.
24073 * emit-rtl.c: Include rtx-vector-builder.h.
24074 (gen_const_vec_duplicate_1): Delete.
24075 (gen_const_vector): Call gen_const_vec_duplicate instead of
24076 gen_const_vec_duplicate_1.
24077 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
24078 (gen_const_vec_duplicate): Use rtx_vector_builder.
24079 (gen_const_vec_series): Likewise.
24080 (gen_rtx_CONST_VECTOR): Likewise.
24081 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
24082 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
24083 Build a new vector rather than modifying a CONST_VECTOR in-place.
24084 (handle_special_swappables): Update call accordingly.
24085 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
24086 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
24087 Build a new vector rather than modifying a CONST_VECTOR in-place.
24088 (handle_special_swappables): Update call accordingly.
24090 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
24092 * simplify-rtx.c (simplify_const_binary_operation): Use
24093 CONST_VECTOR_ELT instead of XVECEXP.
24095 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
24097 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
24098 the selector elements to be different from the data elements
24099 if the selector is a VECTOR_CST.
24100 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
24101 ssizetype for the selector.
24103 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
24105 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
24106 before testing each element individually.
24107 * tree-vect-generic.c (lower_vec_perm): Likewise.
24109 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
24111 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
24112 * selftest-run-tests.c (selftest::run_tests): Call it.
24113 * vector-builder.h (vector_builder::operator ==): New function.
24114 (vector_builder::operator !=): Likewise.
24115 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
24116 (vec_perm_indices::all_from_input_p): New function.
24117 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
24118 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
24119 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
24120 instead of reading the VECTOR_CST directly. Detect whether both
24121 vector inputs are the same before constructing the vec_perm_indices,
24122 and update the number of inputs argument accordingly. Use the
24123 utility functions added above. Only construct sel2 if we need to.
24125 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
24127 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
24128 the broadcast of the low byte.
24129 (expand_mult_highpart): Use an explicit encoding for the permutes.
24130 * optabs-query.c (can_mult_highpart_p): Likewise.
24131 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
24132 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
24133 (vectorizable_bswap): Likewise.
24134 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
24135 explicit encoding for the power-of-2 permutes.
24136 (vect_permute_store_chain): Likewise.
24137 (vect_grouped_load_supported): Likewise.
24138 (vect_permute_load_chain): Likewise.
24140 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
24142 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
24143 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
24144 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
24145 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
24146 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
24147 (vect_gen_perm_mask_any): Likewise.
24149 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
24151 * int-vector-builder.h: New file.
24152 * vec-perm-indices.h: Include int-vector-builder.h.
24153 (vec_perm_indices): Redefine as an int_vector_builder.
24154 (auto_vec_perm_indices): Delete.
24155 (vec_perm_builder): Redefine as a stand-alone class.
24156 (vec_perm_indices::vec_perm_indices): New function.
24157 (vec_perm_indices::clamp): Likewise.
24158 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
24159 (vec_perm_indices::new_vector): New function.
24160 (vec_perm_indices::new_expanded_vector): Update for new
24161 vec_perm_indices class.
24162 (vec_perm_indices::rotate_inputs): New function.
24163 (vec_perm_indices::all_in_range_p): Operate directly on the
24164 encoded form, without computing elided elements.
24165 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
24166 encoding. Update for new vec_perm_indices class.
24167 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
24168 the given vec_perm_builder.
24169 (expand_vec_perm_var): Update vec_perm_builder constructor.
24170 (expand_mult_highpart): Use vec_perm_builder instead of
24171 auto_vec_perm_indices.
24172 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
24173 vec_perm_indices instead of auto_vec_perm_indices. Use a single
24174 or double series encoding as appropriate.
24175 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
24176 vec_perm_indices instead of auto_vec_perm_indices.
24177 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
24178 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
24179 (vect_permute_store_chain): Likewise.
24180 (vect_grouped_load_supported): Likewise.
24181 (vect_permute_load_chain): Likewise.
24182 (vect_shift_permute_load_chain): Likewise.
24183 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
24184 (vect_transform_slp_perm_load): Likewise.
24185 (vect_schedule_slp_instance): Likewise.
24186 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
24187 (vectorizable_mask_load_store): Likewise.
24188 (vectorizable_bswap): Likewise.
24189 (vectorizable_store): Likewise.
24190 (vectorizable_load): Likewise.
24191 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
24192 vec_perm_indices instead of auto_vec_perm_indices. Use
24193 tree_to_vec_perm_builder to read the vector from a tree.
24194 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
24195 vec_perm_builder instead of a vec_perm_indices.
24196 (have_whole_vector_shift): Use vec_perm_builder and
24197 vec_perm_indices instead of auto_vec_perm_indices. Leave the
24198 truncation to calc_vec_perm_mask_for_shift.
24199 (vect_create_epilog_for_reduction): Likewise.
24200 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
24201 from auto_vec_perm_indices to vec_perm_indices.
24202 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
24203 instead of changing individual elements.
24204 (aarch64_vectorize_vec_perm_const): Use new_vector to install
24205 the vector in d.perm.
24206 * config/arm/arm.c (expand_vec_perm_d::perm): Change
24207 from auto_vec_perm_indices to vec_perm_indices.
24208 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
24209 instead of changing individual elements.
24210 (arm_vectorize_vec_perm_const): Use new_vector to install
24211 the vector in d.perm.
24212 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
24213 Update vec_perm_builder constructor.
24214 (rs6000_expand_interleave): Likewise.
24215 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
24216 (rs6000_expand_interleave): Likewise.
24218 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
24220 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
24221 to qimode could truncate the indices.
24222 * optabs.c (expand_vec_perm_var): Likewise.
24224 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
24226 * Makefile.in (OBJS): Add vec-perm-indices.o.
24227 * vec-perm-indices.h: New file.
24228 * vec-perm-indices.c: Likewise.
24229 * target.h (vec_perm_indices): Replace with a forward class
24231 (auto_vec_perm_indices): Move to vec-perm-indices.h.
24232 * optabs.h: Include vec-perm-indices.h.
24233 (expand_vec_perm): Delete.
24234 (selector_fits_mode_p, expand_vec_perm_var): Declare.
24235 (expand_vec_perm_const): Declare.
24236 * target.def (vec_perm_const_ok): Replace with...
24237 (vec_perm_const): ...this new hook.
24238 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
24239 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
24240 * doc/tm.texi: Regenerate.
24241 * optabs.def (vec_perm_const): Delete.
24242 * doc/md.texi (vec_perm_const): Likewise.
24243 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
24244 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
24245 expand_vec_perm for constant permutation vectors. Assert that
24246 the mode of variable permutation vectors is the integer equivalent
24247 of the mode that is being permuted.
24248 * optabs-query.h (selector_fits_mode_p): Declare.
24249 * optabs-query.c: Include vec-perm-indices.h.
24250 (selector_fits_mode_p): New function.
24251 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
24252 is defined, instead of checking whether the vec_perm_const_optab
24253 exists. Use targetm.vectorize.vec_perm_const instead of
24254 targetm.vectorize.vec_perm_const_ok. Check whether the indices
24255 fit in the vector mode before using a variable permute.
24256 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
24257 vec_perm_indices instead of an rtx.
24258 (expand_vec_perm): Replace with...
24259 (expand_vec_perm_const): ...this new function. Take the selector
24260 as a vec_perm_indices rather than an rtx. Also take the mode of
24261 the selector. Update call to shift_amt_for_vec_perm_mask.
24262 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
24263 Use vec_perm_indices::new_expanded_vector to expand the original
24264 selector into bytes. Check whether the indices fit in the vector
24265 mode before using a variable permute.
24266 (expand_vec_perm_var): Make global.
24267 (expand_mult_highpart): Use expand_vec_perm_const.
24268 * fold-const.c: Includes vec-perm-indices.h.
24269 * tree-ssa-forwprop.c: Likewise.
24270 * tree-vect-data-refs.c: Likewise.
24271 * tree-vect-generic.c: Likewise.
24272 * tree-vect-loop.c: Likewise.
24273 * tree-vect-slp.c: Likewise.
24274 * tree-vect-stmts.c: Likewise.
24275 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
24277 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
24278 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
24279 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
24280 (aarch64_vectorize_vec_perm_const): ...this new function.
24281 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
24282 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
24283 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
24284 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
24285 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
24286 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
24287 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
24289 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
24290 check for NEON modes.
24291 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
24292 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
24293 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
24294 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
24296 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
24297 the old VEC_PERM_CONST conditions.
24298 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
24299 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
24300 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
24301 (ia64_vectorize_vec_perm_const_ok): Merge into...
24302 (ia64_vectorize_vec_perm_const): ...this new function.
24303 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
24304 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
24305 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
24306 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
24307 * config/mips/mips.c (mips_expand_vec_perm_const)
24308 (mips_vectorize_vec_perm_const_ok): Merge into...
24309 (mips_vectorize_vec_perm_const): ...this new function.
24310 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
24311 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
24312 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
24313 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
24314 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
24315 (rs6000_expand_vec_perm_const): Delete.
24316 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
24318 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
24319 (altivec_expand_vec_perm_const_le): Take each operand individually.
24320 Operate on constant selectors rather than rtxes.
24321 (altivec_expand_vec_perm_const): Likewise. Update call to
24322 altivec_expand_vec_perm_const_le.
24323 (rs6000_expand_vec_perm_const): Delete.
24324 (rs6000_vectorize_vec_perm_const_ok): Delete.
24325 (rs6000_vectorize_vec_perm_const): New function.
24326 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
24327 an element count and rtx array.
24328 (rs6000_expand_extract_even): Update call accordingly.
24329 (rs6000_expand_interleave): Likewise.
24330 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
24331 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
24332 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
24333 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
24334 (rs6000_expand_vec_perm_const): Delete.
24335 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
24336 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
24337 (altivec_expand_vec_perm_const_le): Take each operand individually.
24338 Operate on constant selectors rather than rtxes.
24339 (altivec_expand_vec_perm_const): Likewise. Update call to
24340 altivec_expand_vec_perm_const_le.
24341 (rs6000_expand_vec_perm_const): Delete.
24342 (rs6000_vectorize_vec_perm_const_ok): Delete.
24343 (rs6000_vectorize_vec_perm_const): New function. Remove stray
24344 reference to the SPE evmerge intructions.
24345 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
24346 an element count and rtx array.
24347 (rs6000_expand_extract_even): Update call accordingly.
24348 (rs6000_expand_interleave): Likewise.
24349 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
24350 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
24352 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
24354 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
24356 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
24357 vector mode and that that mode matches the mode of the data
24359 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
24360 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
24361 directly using expand_vec_perm_1 when forcing selectors into
24363 (expand_vec_perm_var): New function, split out from expand_vec_perm.
24365 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
24367 * optabs-query.h (can_vec_perm_p): Delete.
24368 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
24369 * optabs-query.c (can_vec_perm_p): Split into...
24370 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
24371 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
24372 particular selector is valid.
24373 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
24374 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
24375 (vect_grouped_load_supported): Likewise.
24376 (vect_shift_permute_load_chain): Likewise.
24377 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
24378 (vect_transform_slp_perm_load): Likewise.
24379 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
24380 (vectorizable_bswap): Likewise.
24381 (vect_gen_perm_mask_checked): Likewise.
24382 * fold-const.c (fold_ternary_loc): Likewise. Don't take
24383 implementations of variable permutation vectors into account
24384 when deciding which selector to use.
24385 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
24386 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
24387 with a false third argument.
24388 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
24389 to test whether the constant selector is valid and can_vec_perm_var_p
24390 to test whether a variable selector is valid.
24392 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
24394 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
24395 * optabs-query.c (can_vec_perm_p): Likewise.
24396 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
24397 instead of vec_perm_indices.
24398 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
24399 (vect_gen_perm_mask_checked): Likewise,
24400 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
24401 (vect_gen_perm_mask_checked): Likewise,
24403 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
24405 * optabs-query.h (qimode_for_vec_perm): Declare.
24406 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
24407 (qimode_for_vec_perm): ...this new function.
24408 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
24410 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24412 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
24413 does not have a conditional at the top.
24415 2018-01-02 Richard Biener <rguenther@suse.de>
24417 * ipa-inline.c (big_speedup_p): Fix expression.
24419 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
24422 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
24425 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
24429 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
24430 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
24431 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
24432 cond_taken_branch_cost 3->4.
24434 2018-01-01 Jakub Jelinek <jakub@redhat.com>
24436 PR tree-optimization/83581
24437 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
24438 TODO_cleanup_cfg if any changes have been made.
24440 PR middle-end/83608
24441 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
24442 convert_modes if target mode has the right side, but different mode
24445 PR middle-end/83609
24446 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
24447 last argument when extracting from CONCAT. If either from_real or
24448 from_imag is NULL, use expansion through memory. If result is not
24449 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
24450 the parts directly to inner mode, if even that fails, use expansion
24453 PR middle-end/83623
24454 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
24455 check for bswap in mode rather than HImode and use that in expand_unop
24458 Copyright (C) 2018 Free Software Foundation, Inc.
24460 Copying and distribution of this file, with or without modification,
24461 are permitted in any medium without royalty provided the copyright
24462 notice and this notice are preserved.